DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-16 and 21-26 are the currently pending claims hereby under examination. Claims 17-20 have been canceled.
Drawings
The drawings are objected to because FIG. 4 and 6 are blurry such that they are not readable, see MPEP § 507. Corrected drawing sheets in compliance with 37 CFR 1.121(d) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. The figure or figure number of an amended drawing should not be labeled as "amended." If a drawing figure is to be canceled, the appropriate figure must be removed from the replacement sheet, and where necessary, the remaining figures must be renumbered and appropriate changes made to the brief description of the several views of the drawings for consistency. Additional replacement sheets may be necessary to show the renumbering of the remaining figures. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either "Replacement Sheet" or "New Sheet" pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Objections
Claims 6, 8, 13, 21, and 25 are objected to because of the following informalities:
In claims 6 and 21, lines 1 and 2 respectively: “gait characteristics data” is inconsistent with “gait characteristic data” recited in claims 1, 5, 8, and 12, and should be revised for consistent terminology to clarify whether the same data set is intended;
In claim 8, line 4: “access or execute a trained a machine learning model” has an apparent typographical and grammatical error (duplicative article), and should be revised to "access or execute a trained machine learning model";
In claim 13, line 2: “trained machine learning mode” appears to be a typographical error for “trained machine learning model”, and should be revised; and
Claim 25 is objected to as being of improper dependent form because claim 25 is written as a method claim but depends from claim 24, which is a system claim ("The method of claim 24, wherein the processor is further configured", line 1).
Appropriate correction is required.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 6 and 12 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as failing to set forth the subject matter which the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the applicant regards as the invention.
Claim 6 recites “biomechanics data including one or more of braking impulse, braking peak, propulsive peak, propulsive impulse, other forces derived from ground reaction forces (GRF) data, and joint torques and powers, including hip, knee and/or ankle angles, torques and powers” in lines 2-5. The nested “including” within an “including one or more of” list renders it unclear what constitutes the required biomechanics data, and whether angles are a subset of “joint torques and powers” or additional distinct biomechanics data, such that the metes and bounds of the claimed data are indefinite. The Examiner is interpreting “biomechanics data” under a broadest reasonable interpretation (BRI) to mean kinetic and or kinematic measurements associated with gait, including forces, impulses, joint angles, torques, and powers.
Claim 12 recites “gait characteristic data includes one or more of vertical acceleration, anterior acceleration, a number of steps, a period of time between steps, braking impulse, braking peak, propulsive peak, propulsive impulse, other forces derived from ground reaction forces (GRF) data, and joint torques and powers, including hip, knee and/or ankle angles, torques and powers” in lines 5-7. The nested “including” within an “includes one or more of” list renders it unclear what constitutes the required gait characteristic data, and whether angles are a subset of “joint torques and powers” or additional distinct data, such that the scope is indefinite. The Examiner is interpreting “gait characteristic data” under a broadest reasonable interpretation (BRI) to mean data associated with gait, including acceleration measurements, derived temporal gait parameters, and biomechanics-related measurements.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-16, and 21-26 are rejected under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception (i.e., a law of nature, a natural phenomenon, or an abstract idea) without significantly more. Claims 1-16, and 21-26 are directed to receiving acceleration data related to a patient, extracting or using gait characteristic data, using a trained machine learning model to identify gait features, and outputting a diagnosis of peripheral artery disease (PAD), which is an abstract idea. Claims 1-16, and 21-26 do not include additional elements that integrate the exception into a practical application or that are sufficient to amount to significantly more than the judicial exception for the reasons provided below which are in line with the 2014 Interim Guidance on Patent Subject Matter Eligibility (Federal Register, Vol. 79, No. 241, p 74618, December 16, 2014), the July 2015 Update on Subject Matter Eligibility (Federal Register, Vol. 80, No. 146, p. 45429, July 30, 2015), the May 2016 Subject Matter Eligibility Update (Federal Register, Vol. 81, No. 88, p. 27381, May 6, 2016), and the 2019 Revised Patent Subject Matter Eligibility Guidance (Federal Register, Vol. 84, No. 4, page 50, January 7, 2019).
The analysis of claim 1 is as follows:
Step 1: Claim 1 is drawn to a process.
Step 2A – Prong One: Claim 1 recites an abstract idea. In particular, claim 1 recites the following limitations:
[A1] training a machine learning model with gait characteristic data, extracted from acceleration data for patients known to have PAD and patients that do not have PAD;
[B1] extracting gait characteristic data from the acceleration data related to the specific patient;
[C1] feeding the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient; and
[D1] diagnosing the specific patient as having PAD or not having PAD based on the one or more identified gait features.
These elements [A1]-[D1] of claim 1 are drawn to an abstract idea since (1) they involve mathematical concepts in the form of mathematical relationships, mathematical formulas or equations, and/or mathematical calculations (for example, training and applying a machine learning model to classify a patient based on extracted gait features); and/or (2) they involve a mental process that can be practically performed in the human mind including observation, evaluation, judgment, and opinion and using pen and paper (for example, evaluating gait characteristics and concluding PAD versus not PAD based on the evaluated characteristics).
Step 2A – Prong Two: Claim 1 recites the following additional elements that are beyond the judicial exception:
[A2] receiving acceleration data related to a specific patient.
This additional element does not integrate the exception into a practical application. In particular, the element [A2] is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g).
Desjardins. Although Desjardins recognizes that claims reciting mathematical concepts may integrate an exception into a practical application when the specification identifies a specific technological improvement, the present specification does not describe any such improvement. Instead, the specification explains that the claimed invention uses known wearable sensors, such as accelerometers or inertial measurement units, to collect motion data, applies known gait features and biomechanical metrics derived from that data, and employs conventional machine learning models selected from well-known algorithm classes (e.g., neural networks, random forest, support vector machines, Logit, recurrent neural networks, and long short-term memory models) to perform classification and analysis. The specification further describes feature extraction, signal processing, model training, and inference in functional and results-oriented terms, without identifying any improvement to machine learning technology itself, any new or improved sensor operation, any novel gait signal processing technique, or any improvement to computer functionality. Rather, the specification characterizes the machine learning models and computing components as tools for performing medical evaluation, diagnosis, severity assessment, treatment selection, and alert generation. Accordingly, when the claims are evaluated in light of the specification as a whole, neither independent claim 1 nor dependent claims 2–7, 14–16, and 21–23 reflect a specific technological improvement that would integrate the abstract idea into a practical application, but instead recite the use of conventional computing and machine learning techniques to analyze data and produce clinical conclusions.
Step 2B: Claim 1 does not recite additional elements that amount to significantly more than the judicial exception itself. In particular, the recitations of receiving acceleration data and extracting gait characteristic data do not qualify as significantly more because these limitations are merely insignificant extrasolution activity to the judicial exception, e.g., mere data gathering in conjunction with the abstract idea that uses conventional, routine, and well known elements. Although the rejected claims are method claims and do not expressly recite a system configuration, the specification makes clear that the claimed methods are implemented using conventional hardware components operating in their ordinary capacity. In particular, the specification describes collecting motion data using known wearable sensors, such as accelerometers or inertial measurement units, which are treated as interchangeable, off-the-shelf sensing devices for measuring patient movement. As described by Vining et al. (Vining, US 2017/0205789 A1, [0018]: "commercial off-the-shelf accelerometer"), accelerometers are commercial off-the-shelf devices used for sensing motion, confirming that the accelerometer relied upon by the claims constitutes a well-understood, routine, and conventional sensing component. Where alerts or outputs are provided, the specification describes displaying information using a generic display device, without any claimed improvement to display technology or human-machine interaction, which constitutes the use of conventional, off-the-shelf display technology as recognized in Electric Power Group v. Alstom, 830 F.3d 1350 (Fed. Cir. 2016). Thus, when the claims are evaluated in light of the specification as a whole, the hardware used to perform the claimed steps amounts to well-understood, routine, and conventional components employed for their ordinary purposes, and does not supply an inventive concept or a technological improvement sufficient to transform the abstract idea into patent-eligible subject matter.
Further, the recitation of using a trained machine learning model does not qualify as significantly more because it is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int’l, 110 USPQ2d 1976 (2014)) and/or a claim to an abstract idea requiring no more than being stored on a computer readable medium which is a well-understood, routine and conventional activity previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int’l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations as an ordered combination (that is, as a whole) adds nothing that is not already present when looking at the elements taking individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
Claims 2-7, 14-16, and 21-23 depend from claim 1 or recite a similar abstract idea as claim 1, and therefore recite the same abstract idea as claim 1. Furthermore, these claims only contain recitations that further limit the abstract idea (that is, the claims only recite limitations that further limit the algorithm), with the following exceptions:
Claim 3: “obtained from one or more sensors worn by the specific patient”;
Claim 21: “determining a severity of PAD in the specific patient based on the gait characteristics data” or “determining a change in the severity of PAD in the specific patient based on the gait characteristics data”;
Claim 22: “selecting or modifying a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient”; and
Claim 23: “providing an alert on a display device based on the specific patient being diagnosed as having PAD, or based on the severity of PAD in the specific patient, or based on the change in severity of PAD in the specific patient”.
Each of these claims limitations does not integrate the exception into a practical application. In particular, the element of claim 3 is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g). The elements of claims 21-22 are directed to further analysis and decision-making based on the abstract classification, which remains within the abstract idea and does not recite a particular technological solution. The element of claim 23 is merely adding insignificant post-solution activity to the judicial exception, i.e., outputting or displaying the result at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g).
Also, each of these limitations does not recite additional elements that amount to significantly more than the judicial exception itself because they are merely insignificant extrasolution activity to the judicial exception, e.g., mere data gathering in conjunction with the abstract idea that uses conventional, routine, and well known elements or simply displaying the results of the algorithm that uses conventional, routine, and well known elements. Also, each of these limitations is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions (that is, one of data acquisition and display) that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int'l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations of each claim as an ordered combination in conjunction with the claims from which they depend (that is, as a whole) adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
The analysis of claim 8 is as follows:
Step 1: Claim 8 is drawn to a machine.
Step 2A – Prong One: Claim 8 recites an abstract idea. In particular, claim 8 recites the following limitations:
[A1] access or execute a trained a machine learning model, the trained machine learning model having been trained with biometric data and gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD;
[B1] extract gait characteristic data from the acceleration data related to the specific patient;
[C1] feed the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient; and
[D1] diagnose the specific patient as having PAD or not having PAD based on the one or more identified gait features.
These elements [A1]–[D1] of claim 8 are drawn to an abstract idea since (1) they involve mathematical concepts in the form of mathematical relationships, mathematical formulas or equations, and/or mathematical calculations (for example, training and applying a machine learning model to classify a patient based on extracted gait features); and/or (2) they involve mental processes that can be practically performed in the human mind including observation, evaluation, judgment, and opinion and using pen and paper (for example, evaluating gait characteristics and concluding PAD versus not PAD based on the evaluated characteristics).
Step 2A – Prong Two: Claim 8 recites the following additional elements that are beyond the judicial exception:
[A2] a processor and a memory storing instructions; and
[B2] one or more sensors configured to be worn by a specific patient.
These additional elements do not integrate the exception into a practical application. In particular, the element [B2] is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality – see MPEP 2106.04(d) and MPEP 2106.05(g). The element [A2] is merely an instruction to implement the abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea – see MPEP 2106.04(d) and MPEP 2106.05(f). Further, the claim does not recite a particular technological solution, a particular sensor improvement, or a particular improvement to computer functionality, but instead generally links the use of the judicial exception to a field of use, namely PAD diagnosis – see MPEP 2106.05(h).
Desjardins. Although Desjardins recognizes that claims reciting mathematical concepts may integrate an exception into a practical application when the specification identifies a specific technological improvement, the present specification does not describe any such improvement. Instead, the specification explains that the claimed invention uses known wearable sensors, such as accelerometers or inertial measurement units, to collect motion data, applies known gait features and biomechanical metrics derived from that data, and employs conventional machine learning models selected from well-known algorithm classes (e.g., neural networks, random forest, support vector machines, Logit, recurrent neural networks, and long short-term memory models) to perform classification and analysis. The specification further describes feature extraction, signal processing, model training, and inference in functional and results-oriented terms, without identifying any improvement to machine learning technology itself, any new or improved sensor operation, any novel gait signal processing technique, or any improvement to computer functionality. Rather, the specification characterizes the machine learning models and computing components as tools for performing medical evaluation, diagnosis, severity assessment, treatment selection, and alert generation. Accordingly, when the claims are evaluated in light of the specification as a whole, neither independent claim 8 nor dependent claims 9–13 and 24–26 reflect a specific technological improvement that would integrate the abstract idea into a practical application, but instead recite the use of conventional computing and machine learning techniques to analyze data and produce clinical conclusions.
Step 2B: Claim 8 does not recite additional elements that amount to significantly more than the judicial exception itself. In particular, the recitations of a processor, a memory storing instructions, and wearable sensors do not qualify as significantly more because these limitations are well-understood, routine, and conventional components used for their ordinary purposes in conjunction with the abstract idea. As described by Vining et al. (Vining, US 2017/0205789 A1, [0018]: "commercial off-the-shelf accelerometer"), accelerometers are commercial off-the-shelf devices used for sensing motion, confirming that the sensor relied upon by the claims constitutes a well-understood, routine, and conventional sensing component. Further, the use of conventional, off-the-shelf display technology for outputting results (when claimed) does not add significantly more as recognized in Electric Power Group v. Alstom, 830 F.3d 1350 (Fed. Cir. 2016). Thus, when the claim is evaluated as a whole, claim 8 does not include an inventive concept sufficient to transform the abstract idea into patent-eligible subject matter.
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations as an ordered combination (that is, as a whole) adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
Claims 9-13 and 24-26 depend from claim 8 and therefore recite the same abstract idea as claim 8. Furthermore, these claims only contain recitations that further limit the abstract idea (that is, the claims only recite limitations that further limit the algorithm), with the following exceptions:
Claim 9: “code and data associated with the trained machine learning model is stored in the memory”;
Claim 24: “the processor is further configured to determine a severity of PAD in the specific patient based on the gait characteristics data or determine a change in the severity of PAD in the specific patient based on the gait characteristics data”; and
Claim 26: “the processor is further configured to provide an alert on a display device based on the specific patient being diagnosed as having PAD, or based on the severity of PAD in the specific patient, or based on the change in severity of PAD in the specific patient”.
Each of these claim limitations does not integrate the exception into a practical application. In particular, the element of claim 9 merely recites storing code and data in memory, which is a generic computer function and does not amount to a technological improvement or a practical application. The elements of claim 24 are directed to further analysis and classification based on the abstract evaluation, which remains within the abstract idea and does not recite a particular technological solution. The element of claim 26 is merely adding insignificant post-solution activity to the judicial exception, i.e., outputting or displaying the result at a higher level of generality – see MPEP 2106.04(d) and MPEP 2106.05(g).
Also, each of these limitations does not recite additional elements that amount to significantly more than the judicial exception itself because they are merely insignificant extrasolution activity to the judicial exception, e.g., generic data storage, mere data analysis, and displaying the results of the algorithm using conventional, routine, and well known elements. Also, each of these limitations is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions (that is, data acquisition, storage, and display) that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int'l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations of each claim as an ordered combination in conjunction with the claims from which they depend (that is, as a whole) adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
The analysis of claim 14 is as follows:
Step 1: Claim 14 is drawn to a process.
Step 2A – Prong One: Claim 14 recites an abstract idea. In particular, claim 14 recites the following limitations:
[A1] training a machine learning model with acceleration or accelerometer data for patients known to have PAD and patients that do not have PAD;
[B1] feeding the acceleration data to the trained machine learning model to identify one or more gait features for the specific patient; and
[C1] diagnosing the specific patient as having PAD or not having PAD based on the one or more identified gait features.
These elements [A1]-[C1] of claim 14 are drawn to an abstract idea since (1) they involve mathematical concepts in the form of mathematical relationships, mathematical formulas or equations, and/or mathematical calculations (for example, training and applying a machine learning model to classify a patient based on acceleration or accelerometer data and identified gait features); and/or (2) they involve a mental process that can be practically performed in the human mind including observation, evaluation, judgment, and opinion and using pen and paper (for example, evaluating acceleration-derived gait features and concluding PAD versus not PAD based on the evaluated features).
Step 2A – Prong Two: Claim 14 recites the following additional element that is beyond the judicial exception:
[A2] receiving acceleration data related to a specific patient.
This additional element does not integrate the exception into a practical application. In particular, the element [A2] is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g).
Desjardins. Although Desjardins recognizes that claims reciting mathematical concepts may integrate an exception into a practical application when the specification identifies a specific technological improvement, the present specification does not describe any such improvement. Instead, the specification explains that the claimed invention uses known wearable sensors, such as accelerometers or inertial measurement units, to collect motion data, applies known gait features and biomechanical metrics derived from that data, and employs conventional machine learning models selected from well-known algorithm classes (e.g., neural networks, random forest, support vector machines, Logit, recurrent neural networks, and long short-term memory models) to perform classification and analysis. The specification further describes feature extraction, signal processing, model training, and inference in functional and results-oriented terms, without identifying any improvement to machine learning technology itself, any new or improved sensor operation, any novel gait signal processing technique, or any improvement to computer functionality. Rather, the specification characterizes the machine learning models and computing components as tools for performing medical evaluation, diagnosis, severity assessment, treatment selection, and alert generation. Accordingly, when the claims are evaluated in light of the specification as a whole, neither independent claim 14 nor dependent claims 15, 16, 21–23 reflect a specific technological improvement that would integrate the abstract idea into a practical application, but instead recite the use of conventional computing and machine learning techniques to analyze data and produce clinical conclusions.
Step 2B: Claim 14 does not recite additional elements that amount to significantly more than the judicial exception itself. In particular, the recitation of receiving acceleration data does not qualify as significantly more because this limitation is merely insignificant extrasolution activity to the judicial exception, e.g., mere data gathering in conjunction with the abstract idea that uses conventional, routine, and well known elements. Although the rejected claims are method claims and do not expressly recite a system configuration, the specification makes clear that the claimed methods are implemented using conventional hardware components operating in their ordinary capacity. In particular, the specification describes collecting motion data using known wearable sensors, such as accelerometers or inertial measurement units, which are treated as interchangeable, off-the-shelf sensing devices for measuring patient movement. As described by Vining et al. (Vining, US 2017/0205789 A1, [0018]: "commercial off-the-shelf accelerometer"), accelerometers are commercial off-the-shelf devices used for sensing motion, confirming that the accelerometer relied upon by the claims constitutes a well-understood, routine, and conventional sensing component. Where alerts or outputs are provided, the specification describes displaying information using a generic display device, without any claimed improvement to display technology or human-machine interaction, which constitutes the use of conventional, off-the-shelf display technology as recognized in Electric Power Group v. Alstom, 830 F.3d 1350 (Fed. Cir. 2016). Thus, when the claims are evaluated in light of the specification as a whole, the hardware used to perform the claimed steps amounts to well-understood, routine, and conventional components employed for their ordinary purposes, and does not supply an inventive concept or a technological improvement sufficient to transform the abstract idea into patent-eligible subject matter.
Further, the recitation of using a trained machine learning model does not qualify as significantly more because it is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int’l, 110 USPQ2d 1976 (2014)) and/or a claim to an abstract idea requiring no more than being stored on a computer readable medium which is a well-understood, routine and conventional activity previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int’l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations as an ordered combination (that is, as a whole) adds nothing that is not already present when looking at the elements taking individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
Claim 15 depends from claim 14 and further recites that the machine learning model comprises a recurrent neural network or a long short-term memory (LSTM) model, which merely specifies a particular type of well-known machine learning algorithm and does not add any additional element that integrates the abstract idea into a practical application or amounts to significantly more.
Each of these claims limitations does not integrate the exception into a practical application. In particular, the element of claim 3 is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g). The elements of claims 21-22 are directed to further analysis and decision-making based on the abstract classification, which remains within the abstract idea and does not recite a particular technological solution. The element of claim 23 is merely adding insignificant post-solution activity to the judicial exception, i.e., outputting or displaying the result at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g).
Also, each of these limitations does not recite additional elements that amount to significantly more than the judicial exception itself because they are merely insignificant extrasolution activity to the judicial exception, e.g., mere data gathering in conjunction with the abstract idea that uses conventional, routine, and well known elements or simply displaying the results of the algorithm that uses conventional, routine, and well known elements. Also, each of these limitations is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions (that is, one of data acquisition and display) that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int'l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations of each claim as an ordered combination in conjunction with the claims from which they depend (that is, as a whole) adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
The analysis of claim 16 is as follows:
Step 1: Claim 16 is drawn to a manufacture (i.e., a non-transitory computer-readable medium).
Step 2A – Prong One: Claim 16 recites an abstract idea. In particular, claim 16 recites the following limitations:
[A1] training a machine learning model with gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD;
[B1] extracting gait characteristic data from the acceleration data related to a specific patient;
[C1] feeding the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient; and
[D1] diagnosing the specific patient as having PAD or not having PAD based on the one or more identified gait features.
These elements [A1]–[D1] of claim 16 are drawn to an abstract idea since (1) they involve mathematical concepts in the form of mathematical relationships, mathematical formulas or equations, and/or mathematical calculations (for example, training and applying a machine learning model to classify a patient based on extracted gait features); and/or (2) they involve mental processes that can be practically performed in the human mind including observation, evaluation, judgment, and opinion and using pen and paper (for example, evaluating gait characteristics and concluding PAD versus not PAD based on the evaluated characteristics).
Step 2A – Prong Two: Claim 16 recites the following additional element that is beyond the judicial exception:
[A2] receiving acceleration data related to a specific patient.
This additional element does not integrate the exception into a practical application. In particular, the element [A2] is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality – see MPEP 2106.04(d) and MPEP 2106.05(g).
Desjardins. Although Desjardins recognizes that claims reciting mathematical concepts may integrate an exception into a practical application when the specification identifies a specific technological improvement, the present specification does not describe any such improvement. Instead, the specification explains that the claimed invention uses known wearable sensors, such as accelerometers or inertial measurement units, to collect motion data, applies known gait features and biomechanical metrics derived from that data, and employs conventional machine learning models selected from well-known algorithm classes to perform classification and analysis. The specification further describes feature extraction, signal processing, model training, and inference in functional and results-oriented terms, without identifying any improvement to machine learning technology itself, any new or improved sensor operation, any novel gait signal processing technique, or any improvement to computer functionality. Rather, the specification characterizes the machine learning models and computing components as tools for performing medical evaluation, diagnosis, severity assessment, treatment selection, and alert generation. Accordingly, when the claims are evaluated in light of the specification as a whole, claim 16 does not reflect a specific technological improvement that would integrate the abstract idea into a practical application, but instead recites the use of conventional computing and machine learning techniques to analyze data and produce clinical conclusions.
Step 2B: Claim 16 does not recite additional elements that amount to significantly more than the judicial exception itself. In particular, the recitation of a non-transitory computer-readable medium storing instructions merely links the abstract idea to a generic computer-readable medium, which is a well-understood, routine, and conventional activity previously known in the industry. As discussed above with respect to the specification, the claimed steps are implemented using conventional hardware components operating in their ordinary capacity, including commercial off-the-shelf accelerometers for motion sensing and generic processors and memory for data processing and storage. Thus, when evaluated as a whole, claim 16 does not include an inventive concept sufficient to transform the abstract idea into patent-eligible subject matter.
Each of these claims limitations does not integrate the exception into a practical application. In particular, the element of claim 3 is merely adding insignificant extra-solution activity to the judicial exception, i.e., mere data gathering at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g). The elements of claims 21-22 are directed to further analysis and decision-making based on the abstract classification, which remains within the abstract idea and does not recite a particular technological solution. The element of claim 23 is merely adding insignificant post-solution activity to the judicial exception, i.e., outputting or displaying the result at a higher level of generality - see MPEP 2106.04(d) and MPEP 2106.05(g).
Also, each of these limitations does not recite additional elements that amount to significantly more than the judicial exception itself because they are merely insignificant extrasolution activity to the judicial exception, e.g., mere data gathering in conjunction with the abstract idea that uses conventional, routine, and well known elements or simply displaying the results of the algorithm that uses conventional, routine, and well known elements. Also, each of these limitations is simply appending well-understood, routine and conventional activities previously known in the industry, specified at a high level of generality, to the judicial exception, e.g., a claim to an abstract idea requiring no more than a generic computer to perform generic computer functions (that is, one of data acquisition and display) that are well-understood, routine and conventional activities previously known in the industry (see Electric Power Group, 830 F.3d 1350 (Fed. Cir. 2016); Alice Corp. v. CLS Bank Int'l, 110 USPQ2d 1976 (2014); SAP Am. v. InvestPic, 890 F.3d 1016 (Fed. Circ. 2018)).
In view of the above, the additional elements individually do not integrate the exception into a practical application and do not amount to significantly more than the above-judicial exception (the abstract idea). Looking at the limitations of each claim as an ordered combination in conjunction with the claims from which they depend (that is, as a whole) adds nothing that is not already present when looking at the elements taken individually. There is no indication that the combination of elements improves the functioning of a computer, for example, or improves any other technology. There is no indication that the combination of elements permits automation of specific tasks that previously could not be automated. There is no indication that the combination of elements includes a particular solution to a computer-based problem or a particular way to achieve a desired computer-based outcome. Rather, the collective functions of the claimed invention merely provide conventional computer implementation, i.e., the computer is simply a tool to perform the process.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1-3, 6, 14, 16, 21, and 24 are rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman.
Regarding claim 1, Huang teaches a method of receiving acceleration data related to a specific patient (Huang, ¶[0015]: "The wearable device can provide a portable, user friendly solution for detecting data related to a subject's gait, balance or posture in the subject's natural environment", ¶[0006]: “The method includes receiving a signal comprising sensor data from sensors according to a short range communication protocol. The sensors include a three-axis accelerometer”, Huang receives accelerometer-based sensor data as the acceleration data related to a specific subject/patient); extracting gait characteristic data from the acceleration data related to the specific patient (Huang, ¶[0015]: “The portable computing device can analyze the data to determine identify any unstable pattern and apply a statistical or machine learning-based classification to assign one or more clinical parameters to the unstable pattern”, Huang teaches analyzing the data from the wearable device to determine an unstable pattern associated with gait; claim 18: “analyzing, by the system, the sensor data to identify a pattern related to gait, balance or posture within the sensor data", Huang teaches analyzing sensor data to identify a gait-related pattern; ¶[0029]: “The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data. The unstable patterns can be associated with one or more of gait”, Huang extracts gait-related characteristic information by analyzing the accelerometer data to identify a gait-related pattern); and feeding the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient (Huang, ¶[0031]: “one or more pattern recognition classifiers, each of which utilize the extracted features or a subset of the extracted features to determine an appropriate clinical parameter… From the provided feature vector, an outcome class is selected”, Huang feeds extracted gait-related features as a feature vector to a trained classifier to produce an output class and associated confidence).
Also regarding claim 1, Huang does not teach a computer-implemented method of diagnosing peripheral artery disease (PAD) in a patient. Rather, Huang teaches a computer-implemented method performed by a processor that receives sensor data from wearable sensors, analyzes the sensor data to identify a pattern related to gait, and applies a statistical or machine learning-based classification to the gait-related pattern to assign a clinical parameter (Huang, ¶[0004]-[0005]: “apply a statistical or machine learning-based classification to the pattern related to gait, balance or posture to assign a clinical parameter”). However, Huang does not expressly teach that the method is for diagnosing peripheral artery disease (PAD).
Chidean, which investigates gait analysis in peripheral arterial disease patients, teaches that “Peripheral arterial disease (PAD) is an artherosclerotic occlusive disorder of distal arteries” and that “full spectra analysis allowed to better characterize gait in PAD patients than classical spectral analysis" where "Acceleration gait signals were recorded using... four wireless sensor nodes located at ankle and hip..." (Chidean, Abstract). "It allowed to better discriminate PAD patients and control subjects” and that the analysis “could be used for clinical early diagnosis” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters have “diagnostic ability” to "identify the presence of PAD", stating that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” and that receiver operating characteristic curve analyses were used to discriminate “individuals with PAD-IC from those without PAD-IC” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang in view of Chidean and Rahman to diagnose peripheral artery disease (PAD) in a patient, by configuring Huang’s statistical or machine learning-based classification of gait-related patterns to discriminate PAD patients from non-PAD patients using acceleration gait signal characteristics. The modification would have been possible because Huang’s classifier is expressly trained on classes of interest using extracted gait-related patterns from sensor data, Chidean teaches PAD-specific gait characterization from acceleration signals, and Rahman teaches that gait variability parameters have diagnostic value for identifying the presence of PAD using classification and discrimination analysis. The benefit of this combination would have been to provide an objective, wearable-sensor-based computer-implemented PAD diagnosis using gait-derived features with established diagnostic discrimination capability, improving reliability and clinical usefulness of PAD screening outside specialized laboratory testing.
Also regarding claim 1, the modified Huang does not fully teach training a machine learning model with gait characteristic data, extracted from acceleration data for patients known to have PAD and patients that do not have PAD. Rather, the modified Huang teaches training a classifier on “training patterns representing various classes of interest” and that the classifier “utilize[s] the extracted features or a subset of the extracted features” as a feature vector to select an “outcome class” (Huang, ¶[0029]:"inertial data from the three-dimensional accelerometer and three-dimensional gyroscope can be filtered and calibrated... The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data"; ¶[0031]: “Each classifier is trained on a plurality of training patterns representing various classes of interest. The training process of the a given classifier will vary with its implementation, but the training generally involves a statistical aggregation of training data from a plurality of training images into one or more parameters associated with the output class”, Huang uses the term “training images” as generic pattern-recognition terminology, but here the training patterns are abstract feature representations derived from inertial sensor data processed to identify gait-related patterns, not visual image content). However, the modified Huang does not expressly teach that the training data is gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD.
Chidean, which investigates gait analysis in PAD patients, teaches characterizing gait for “both control subjects and PAD patients” using “acceleration gait signals” recorded from wearable sensors and that the proposed analysis could be used to “discriminate PAD patients and control subjects” (Chidean, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean to train Huang’s classifier using gait characteristic data extracted from acceleration gait signals for patients known to have PAD and patients that do not have PAD, such that the “classes of interest” in Huang correspond to PAD and non-PAD subjects. The modification would have been possible because Huang expressly teaches training a machine learning classifier using training data aggregated into parameters associated with an output class, and Chidean teaches collecting acceleration gait signals from PAD patients and control subjects and using derived gait characteristics to discriminate between those groups. The benefit of this combination would have been to enable automated, wearable-sensor-based PAD discrimination by using Huang’s trained classifier framework on Chidean’s PAD versus control acceleration gait characteristics, improving classification robustness relative to threshold-only methods.
Also regarding claim 1, the modified Huang does not fully teach diagnosing the specific patient as having PAD or not having PAD based on the one or more identified gait features. Rather, the modified Huang teaches applying machine learning-based classification to extracted gait-related features to select an “outcome class” representing a clinical parameter, thereby producing a classified output based on the extracted features of PAD patients and control subjects as previously established (Huang, ¶[0031]). However, it does not expressly teach diagnosing the specific patient as having PAD or not having PAD based on the identified gait features.
Chidean teaches that acceleration gait signal characterization “allowed to better discriminate PAD patients and control subjects” and that “The full spectral analysis could be used for clinical early diagnosis and also to monitor the disease evolution in PAD patients” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters provide diagnostic discrimination for PAD, stating that “Receiver operating characteristics curve analyses of the pain free walking condition were performed to determine the optimal cut-off values for separating individuals with PAD-IC from those without PAD-IC” and that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean and Rahman to diagnose the specific patient as having PAD or not having PAD based on the gait features identified from the acceleration gait signals, by defining Huang’s classifier output classes to correspond to PAD and non-PAD consistent with established diagnostic discrimination of gait variability parameters. One of ordinary skill in the art would have found this modification feasible because Huang already selects an outcome class from a feature vector of extracted gait features (Huang, ¶[0031]), and Chidean and Rahman teach that gait-derived parameters discriminate PAD patients from controls and can be used for clinical early diagnosis, thereby providing the PAD/non-PAD class definitions and diagnostic interpretation for Huang’s selected outcome class. The benefit of the combination would have been enabling earlier and more objective identification of peripheral artery disease outside specialized vascular laboratories by leveraging wearable accelerometer data and machine learning-based gait analysis to support scalable screening, monitoring, and clinical decision-making for PAD.
Regarding claim 2, the modified Huang teaches that the machine learning model comprises one of a neural network algorithm, a random forest algorithm, a support vector machine (SVM) algorithm and a Logit algorithm (Huang, ¶[0031], “Any of a variety of optimization techniques can be utilized for the classification algorithm, including support vector machines, self-organized maps, fuzzy logic systems, data fusion processes, ensemble methods, statistical or machine learning-based systems or artificial neural networks”, Huang expressly teaches using support vector machines and artificial neural networks as examples of the classification algorithm in the machine learning model; ¶[0032], “a support vector machine (SVM) classifier can process the training data”, Huang expressly teaches a support vector machine classifier; ¶[0032], “a convolutional neural network (CNN) classifier can process the training data”, Huang expressly teaches a neural network classifier as an example of the machine learning model).
Regarding claim 3, the modified Huang teaches that the acceleration data related to the specific patient is obtained from one or more sensors worn by the specific patient (Huang, ¶[0006]: “The sensors include a three-axis accelerometer, a three-axis gyroscope and an array of pressure sensors embedded within a wearable device”, Huang expressly teaches that acceleration data is obtained from accelerometer sensors embedded within a wearable device worn by the patient; ¶[0019]: “a subject can wear one or more inserts 22 (e.g., an insert 22 in each shoe) to facilitate testing of gait, balance or posture”, Huang expressly teaches that the sensors providing acceleration data are worn by the subject); ¶[0015]: “the wearable device can be one or more shoe insoles embedded with sensors that can be configured to capture data indicative of movement”, Huang further teaches wearable sensors configured to capture movement-related data from the subject).
Regarding claim 6, the modified Huang teaches the method of claim 1, but does not expressly teach that the gait characteristics data includes biomechanics data including one or more of braking impulse, braking peak, propulsive peak, propulsive impulse, other forces derived from ground reaction forces (GRF) data, and joint torques and powers, including hip, knee and/or ankle angles, torques and powers. Rather, the modified Huang teaches collecting multi-axis accelerometer data and force-related underfoot data during gait using wearable sensors, including a three-axis accelerometer and insole-based pressure sensors, which provide time-varying acceleration and force signals associated with gait dynamics (Huang, ¶[0006]; ¶[0045]), but does not expressly teach biomechanics data including braking impulse, braking peak, propulsive peak, propulsive impulse, other forces derived from ground reaction forces (GRF) data, or joint torques and powers, including hip, knee and/or ankle angles, torques and powers.
Rahman teaches biomechanical joint angle measurements during gait in PAD-related analysis, stating “Ranges of motion of the ankle, knee, and hip joint angles were calculated for the gait cycles in every trial” (Rahman, p. 5, Sec. 2.3.1). Rahman thus teaches extracting kinematic biomechanics data, including hip, knee, and ankle joint angles, from gait data for patients with PAD.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Rahman to have the gait characteristics data include biomechanics data including hip, knee and/or ankle angles. One of ordinary skill in the art would have found it obvious and feasible to incorporate Rahman’s joint angle biomechanics measurements into Huang’s wearable gait analysis workflow because Huang already includes sensors for gait measurement and analysis and further discloses a “Joint Angular and EMG Sensing Unit” for recording leg IMU data concurrently with gait measurements (Huang, ¶[0046]). The benefit of the combination would have been enabling the modified Huang’s gait characterization to incorporate joint angle biomechanics features for improved discrimination of gait-related clinical differences in patients, including PAD-related gait differences.
Regarding claim 14, Huang teaches a system comprising a processor and a memory storing instructions, which when executed by the processor causes the processor to (Huang, ¶[0005]: "The portable computing device includes: a communications unit to receive the signal streamed according to the short range communication protocol; a non-transitory memory storing machine readable instructions; and a processor to execute the machine readable instructions", describing a processor with memory for storing and executing instructions): receive acceleration data related to a specific patient (Huang, ¶[0015]: “The wearable device can provide a portable, user friendly solution for detecting data related to a subject's gait, balance or posture in the subject's natural environment”, Huang teaches detecting/collecting gait-related data for a subject in the subject’s natural environment; ¶[0015]: “The wearable device can send the data related to the subject's gate wirelessly to the mobile computing device for analysis", Huang teaches sending the detected gait-related data to a mobile computing device for analysis; ¶[0004]: “The sensors include a three-axis accelerometer, a three-axis gyroscope and an array of pressure sensors embedded within the wearable device”, Huang teaches that the wearable device sensors include a three-axis accelerometer, which provides the acceleration data related to the subject/patient); extract gait characteristic data from the acceleration data related to the specific patient (Huang, ¶[0015]: “The portable computing device can analyze the data to determine identify any unstable pattern and apply a statistical or machine learning-based classification to assign one or more clinical parameters to the unstable pattern”, Huang teaches analyzing the data from the wearable device to determine an unstable pattern associated with gait; claim 18: “analyzing, by the system, the sensor data to identify a pattern related to gait, balance or posture within the sensor data", Huang teaches analyzing sensor data to identify a gait-related pattern; ¶[0029]: “The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data. The unstable patterns can be associated with one or more of gait”, Huang extracts gait-related characteristic information by analyzing the accelerometer data to identify a gait-related pattern); feed the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient (Huang, ¶[0031]: “one or more pattern recognition classifiers, each of which utilize the extracted features or a subset of the extracted features to determine an appropriate clinical parameter… From the provided feature vector, an outcome class is selected”, Huang feeds extracted gait-related features as a feature vector to a trained classifier to produce an output class and associated confidence).
Also regarding claim 14, Huang does not expressly teach: a system for diagnosing peripheral artery disease (PAD) in a patient. Rather, Huang teaches a wireless portable gait system performed by a processor that receives sensor data from wearable sensors, analyzes the sensor data to identify a pattern related to gait, and applies a statistical or machine learning-based classification to the pattern related to gait to assign a clinical parameter (Huang, claim 18: “receiving , by a system comprising a processor , a signal comprising sensor data from sensors according to a short range communication protocol , wherein the sensors comprise a three - axis accelerometer …”; “analyzing , by the system , the sensor data to identify a pattern related to gait , balance or posture within the sensor data ;”; and “applying , by the system , a statistical or machine learning based classification to the pattern related to gait , balance or posture to assign a clinical parameter to the pattern …”). However, Huang does not expressly teach that the system is for diagnosing peripheral artery disease (PAD).
Chidean, which investigates gait analysis in peripheral arterial disease patients, teaches that acceleration gait signal characterization “allowed to better discriminate PAD patients and control subjects” and that “[t]he full spectral analysis could be used for clinical early diagnosis and also to monitor the disease evolution in PAD patients” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters provide diagnostic discrimination for PAD, stating that “Receiver operating characteristics curve analyses … were performed to determine the optimal cut-off values for separating individuals with PAD-IC from those without PAD-IC” and that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang in view of Chidean and Rahman to diagnose peripheral artery disease (PAD) in a patient by configuring Huang’s statistical or machine learning-based classification of gait-related patterns to discriminate PAD patients from non-PAD patients using acceleration gait signal characteristics. One of ordinary skill in the art would have found this modification feasible because Huang already (i) receives wearable accelerometer sensor data, (ii) analyzes the sensor data to identify gait-related patterns, and (iii) applies statistical or machine learning-based classification to select an outcome class from a provided feature vector (Huang, claim 18; ¶[0031]), and Chidean and Rahman provide PAD-versus-control gait-characterization and diagnostic discrimination context to define the “classes of interest” as PAD and non-PAD. The benefit of this combination would have been to provide an objective, wearable-sensor-based, computer-implemented PAD diagnosis using gait-derived features with established diagnostic discrimination capability, improving reliability and clinical usefulness of PAD screening outside specialized laboratory testing.
Also regarding claim 14, the modified Huang does not fully teach: the trained machine learning model having been trained with gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD. Rather, the modified Huang teaches training a classifier on “training patterns representing various classes of interest” and that the classifier “utilize[s] the extracted features or a subset of the extracted features” as a feature vector to select an “outcome class” (Huang, ¶[0029]:"inertial data from the three-dimensional accelerometer and three-dimensional gyroscope can be filtered and calibrated... The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data"; ¶[0031]: “Each classifier is trained on a plurality of training patterns representing various classes of interest. The training process of the a given classifier will vary with its implementation, but the training generally involves a statistical aggregation of training data from a plurality of training images into one or more parameters associated with the output class”, Huang uses the term “training images” as generic pattern-recognition terminology, but here the training patterns are abstract feature representations derived from inertial sensor data processed to identify gait-related patterns, not visual image content). However, the modified Huang does not expressly teach that the training data is gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD.
Chidean, which investigates gait analysis in PAD patients, teaches characterizing gait for “both control subjects and PAD patients” using “acceleration gait signals” recorded from wearable sensors and that the proposed analysis could be used to “discriminate PAD patients and control subjects” (Chidean, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean to train Huang’s classifier using gait characteristic data extracted from acceleration gait signals for patients known to have PAD and patients that do not have PAD, such that the “classes of interest” in Huang correspond to PAD and non-PAD subjects. One of ordinary skill in the art would have found this modification feasible because Huang expressly teaches training a classifier on training patterns for classes of interest and selecting an outcome class from a provided feature vector (Huang, ¶[0031]), and Chidean teaches collecting acceleration gait signals from PAD patients and control subjects and using gait characterization to discriminate those groups. The benefit of the combination would have been to enable automated, wearable-sensor-based PAD discrimination by using Huang’s trained classifier framework on PAD-versus-control acceleration gait characteristics, improving PAD screening and classification robustness.
Also regarding claim 14, the modified Huang does not fully teach: diagnose the specific patient as having PAD or not having PAD based on the one or more identified gait features. Rather, the modified Huang teaches applying machine learning-based classification to extracted gait-related features to select an “outcome class” representing a clinical parameter, thereby producing a classified output based on the extracted features of PAD patients and control subjects as previously established (Huang, ¶[0031]). However, it does not expressly teach diagnosing the specific patient as having PAD or not having PAD based on the identified gait features.
Chidean teaches that acceleration gait signal characterization “allowed to better discriminate PAD patients and control subjects” and that “The full spectral analysis could be used for clinical early diagnosis and also to monitor the disease evolution in PAD patients” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters provide diagnostic discrimination for PAD, stating that “Receiver operating characteristics curve analyses of the pain free walking condition were performed to determine the optimal cut-off values for separating individuals with PAD-IC from those without PAD-IC” and that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean and Rahman to diagnose the specific patient as having PAD or not having PAD based on the gait features identified from the acceleration gait signals by defining Huang’s classifier output classes to correspond to PAD and non-PAD consistent with established diagnostic discrimination of gait parameters. One of ordinary skill in the art would have found this modification feasible because Huang already selects an outcome class from a feature vector of extracted gait features (Huang, ¶[0031]), and Chidean and Rahman teach that gait-derived parameters discriminate PAD patients from controls and can be used for clinical early diagnosis, thereby providing the PAD/non-PAD class definitions and diagnostic interpretation for Huang’s selected outcome class. The benefit of the combination would have been enabling earlier and more objective identification of peripheral artery disease outside specialized vascular laboratories by leveraging wearable accelerometer data and machine learning-based gait analysis to support scalable screening, monitoring, and clinical decision-making for PAD.
Regarding claim 16, Huang teaches a non-transitory computer-readable medium storing instructions, which when executed by one or more processors (Huang, ¶[0005]: "The portable computing device includes: a communications unit to receive the signal streamed according to the short range communication protocol; a non-transitory memory storing machine readable instructions; and a processor to execute the machine readable instructions", describing a processor with memory/non-transitory computer-readable medium for storing and executing instructions), cause the one or more processors to implement a method of receiving acceleration data related to a specific patient (Huang, ¶[0015]: "The wearable device can provide a portable, user friendly solution for detecting data related to a subject's gait, balance or posture in the subject's natural environment"; ¶[0006]: “The method includes receiving a signal comprising sensor data from sensors according to a short range communication protocol. The sensors include a three-axis accelerometer”, Huang receives accelerometer-based sensor data as the acceleration data related to a specific subject/patient); extracting gait characteristic data from the acceleration data related to the specific patient (Huang, ¶[0015]: “The portable computing device can analyze the data to determine identify any unstable pattern and apply a statistical or machine learning-based classification to assign one or more clinical parameters to the unstable pattern”, Huang teaches analyzing the data from the wearable device to determine an unstable pattern associated with gait; claim 18: “analyzing, by the system, the sensor data to identify a pattern related to gait, balance or posture within the sensor data", Huang teaches analyzing sensor data to identify a gait-related pattern; ¶[0029]: “The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data. The unstable patterns can be associated with one or more of gait”, Huang extracts gait-related characteristic information by analyzing the accelerometer data to identify a gait-related pattern); and feeding the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient (Huang, ¶[0031]: “one or more pattern recognition classifiers, each of which utilize the extracted features or a subset of the extracted features to determine an appropriate clinical parameter… From the provided feature vector, an outcome class is selected”, Huang feeds extracted gait-related features as a feature vector to a trained classifier to produce an output class and associated confidence).
Also regarding claim 1, Huang does not teach that the executed instructions cause the one or more processors to implement a method of diagnosing peripheral artery disease (PAD) in a patient. Rather, Huang teaches a computer-implemented method performed by a processor that receives sensor data from wearable sensors, analyzes the sensor data to identify a pattern related to gait, and applies a statistical or machine learning-based classification to the gait-related pattern to assign a clinical parameter (Huang, ¶[0004]-[0005]: “apply a statistical or machine learning-based classification to the pattern related to gait, balance or posture to assign a clinical parameter”). However, Huang does not expressly teach that the method is for diagnosing peripheral artery disease (PAD).
Chidean, which investigates gait analysis in peripheral arterial disease patients, teaches that “Peripheral arterial disease (PAD) is an artherosclerotic occlusive disorder of distal arteries” and that “full spectra analysis allowed to better characterize gait in PAD patients than classical spectral analysis" where "Acceleration gait signals were recorded using... four wireless sensor nodes located at ankle and hip..." (Chidean, Abstract). "It allowed to better discriminate PAD patients and control subjects” and that the analysis “could be used for clinical early diagnosis” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters have “diagnostic ability” to "identify the presence of PAD", stating that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” and that receiver operating characteristic curve analyses were used to discriminate “individuals with PAD-IC from those without PAD-IC” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang in view of Chidean and Rahman to diagnose peripheral artery disease (PAD) in a patient, by configuring Huang’s statistical or machine learning-based classification of gait-related patterns to discriminate PAD patients from non-PAD patients using acceleration gait signal characteristics. The modification would have been possible because Huang’s classifier is expressly trained on classes of interest using extracted gait-related patterns from sensor data, Chidean teaches PAD-specific gait characterization from acceleration signals, and Rahman teaches that gait variability parameters have diagnostic value for identifying the presence of PAD using classification and discrimination analysis. The benefit of this combination would have been to provide an objective, wearable-sensor-based computer-implemented PAD diagnosis using gait-derived features with established diagnostic discrimination capability, improving reliability and clinical usefulness of PAD screening outside specialized laboratory testing.
Also regarding claim 16, Huang does not teach training a machine learning model with gait characteristic data extracted from acceleration data for patients known to have peripheral artery disease (PAD) and patients that do not have PAD. Rather, the modified Huang teaches training a classifier on “training patterns representing various classes of interest” and that the classifier “utilize[s] the extracted features or a subset of the extracted features” as a feature vector to select an “outcome class” (Huang, ¶[0029]:"inertial data from the three-dimensional accelerometer and three-dimensional gyroscope can be filtered and calibrated... The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data"; ¶[0031]: “Each classifier is trained on a plurality of training patterns representing various classes of interest. The training process of the a given classifier will vary with its implementation, but the training generally involves a statistical aggregation of training data from a plurality of training images into one or more parameters associated with the output class”, Huang uses the term “training images” as generic pattern-recognition terminology, but here the training patterns are abstract feature representations derived from inertial sensor data processed to identify gait-related patterns, not visual image content). However, the modified Huang does not expressly teach that the training data is gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD.
Chidean, which investigates gait analysis in PAD patients, teaches characterizing gait for “both control subjects and PAD patients” using “acceleration gait signals” recorded from wearable sensors and that the proposed analysis could be used to “discriminate PAD patients and control subjects” (Chidean, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean to train Huang’s classifier using gait characteristic data extracted from acceleration gait signals for patients known to have PAD and patients that do not have PAD, such that the “classes of interest” in Huang correspond to PAD and non-PAD subjects. The modification would have been possible because Huang expressly teaches training a machine learning classifier using training data aggregated into parameters associated with an output class, and Chidean teaches collecting acceleration gait signals from PAD patients and control subjects and using derived gait characteristics to discriminate between those groups. The benefit of this combination would have been to enable automated, wearable-sensor-based PAD discrimination by using Huang’s trained classifier framework on Chidean’s PAD versus control acceleration gait characteristics, improving classification robustness relative to threshold-only methods.
Also regarding claim 16, Huang does not teach diagnosing the specific patient as having PAD or not having PAD based on the one or more identified gait features. Rather, the modified Huang teaches applying machine learning-based classification to extracted gait-related features to select an “outcome class” representing a clinical parameter, thereby producing a classified output based on the extracted features of PAD patients and control subjects as previously established (Huang, ¶[0031]). However, it does not expressly teach diagnosing the specific patient as having PAD or not having PAD based on the identified gait features.
Chidean teaches that acceleration gait signal characterization “allowed to better discriminate PAD patients and control subjects” and that “The full spectral analysis could be used for clinical early diagnosis and also to monitor the disease evolution in PAD patients” (Chidean, Abstract; p. 5-7, Sec. 6.).
Rahman teaches that gait variability parameters provide diagnostic discrimination for PAD, stating that “Receiver operating characteristics curve analyses of the pain free walking condition were performed to determine the optimal cut-off values for separating individuals with PAD-IC from those without PAD-IC” and that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean and Rahman to diagnose the specific patient as having PAD or not having PAD based on the gait features identified from the acceleration gait signals, by defining Huang’s classifier output classes to correspond to PAD and non-PAD consistent with established diagnostic discrimination of gait variability parameters. One of ordinary skill in the art would have found this modification feasible because Huang already selects an outcome class from a feature vector of extracted gait features (Huang, ¶[0031]), and Chidean and Rahman teach that gait-derived parameters discriminate PAD patients from controls and can be used for clinical early diagnosis, thereby providing the PAD/non-PAD class definitions and diagnostic interpretation for Huang’s selected outcome class. The benefit of the combination would have been enabling earlier and more objective identification of peripheral artery disease outside specialized vascular laboratories by leveraging wearable accelerometer data and machine learning-based gait analysis to support scalable screening, monitoring, and clinical decision-making for PAD.
Regarding claim 21, the modified Huang does not fully teach determining a severity of PAD in the specific patient based on the gait characteristics data or determining a change in the severity of PAD in the specific patient based on the gait characteristics data. Rather, Huang teaches analyzing gait related sensor data, applying a statistical or machine learning based classification, and assigning a “clinical parameter” that can be used for “diagnosis stratification or monitoring” of a medical condition (Huang, ¶[0036]: “The clinical parameter assigned by the classifier 46 can relate to a certain medical condition. In some examples the clinical parameter can be used in the diagnosis stratification or monitoring of a medical condition that affects the musculoskeletal system”). However, Huang does not expressly teach determining a severity of PAD or determining a change in the severity of PAD based on gait characteristics data.
Chidean teaches using analysis of acceleration gait signals in PAD patients and expressly states that the analysis can be used “to monitor the disease evolution in PAD patients” and that it “showed promising results to assess severity of PAD” (Chidean, p. 5-7, Sec. 6).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean to determine a severity of PAD in the specific patient based on the gait characteristics data or to determine a change in the severity of PAD in the specific patient based on the gait characteristics data. This modification would have been feasible because Huang already produces a clinical parameter from classified gait related data for stratification and monitoring, and Chidean teaches that gait acceleration signal analysis can be used to assess PAD severity and monitor disease evolution in PAD patients, thereby providing an express PAD severity and progression basis for Huang’s stratification and monitoring outputs. The benefit of the combination would have been enabling PAD severity assessment and longitudinal monitoring using wearable gait acceleration data within a machine learning based classification framework, supporting earlier detection of worsening disease and more objective follow up in clinical or remote settings.
Regarding claim 24, the modified Huang does not fully teach that the processor is further configured to determine a severity of PAD in the specific patient based on the gait characteristics data or determine a change in the severity of PAD in the specific patient based on the gait characteristics data. Rather, Huang teaches analyzing gait related sensor data, applying a statistical or machine learning based classification, and assigning a “clinical parameter” that can be used for “diagnosis stratification or monitoring” of a medical condition (Huang, ¶[0036]: “The clinical parameter assigned by the classifier 46 can relate to a certain medical condition. In some examples the clinical parameter can be used in the diagnosis stratification or monitoring of a medical condition that affects the musculoskeletal system”). However, Huang does not expressly teach determining a severity of PAD or determining a change in the severity of PAD based on gait characteristics data.
Chidean teaches using analysis of acceleration gait signals in PAD patients and expressly states that the analysis can be used “to monitor the disease evolution in PAD patients” and that it “showed promising results to assess severity of PAD” (Chidean, p. 5-7, Sec. 6).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean to configure the processor to determine a severity of PAD in the specific patient based on the gait characteristics data or to determine a change in the severity of PAD in the specific patient based on the gait characteristics data. This modification would have been feasible because Huang already implements processor based classification of gait related data to produce a clinical parameter used for stratification and monitoring, and Chidean teaches that gait acceleration signal analysis can be used to assess PAD severity and monitor disease evolution in PAD patients, thereby providing an express PAD severity and progression basis for Huang’s stratification and monitoring outputs. The benefit of the combination would have been enabling PAD severity assessment and longitudinal monitoring using wearable gait acceleration data within a machine learning based classification framework, supporting earlier detection of worsening disease and more objective follow up in clinical or remote settings.
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Szymczak et al. (Szymczak M, Krupa P, Oszkinis G, Majchrzycki M. Gait pattern in patients with peripheral artery disease. BMC Geriatr. 2018 Feb 20), hereto referred as Szymczak.
The modified Huang teaches claim 1 as described above.
Regarding claim 4, the modified Huang teaches the method of claim 1, but does not expressly teach that the one or more gait features include one or more of step time asymmetry, step time variability, step time, stance time, stride time, and swing time. Rather, the modified Huang teaches measuring sensor based parameters from a Portable Gait Lab system, including timing and symmetry related parameters, but does not expressly teach using gait features that include step time, stride time, stance time, or swing time.
Rahman teaches that stride time and step time are recognized stride characteristics used in assessing gait variability in patients with peripheral artery disease (Rahman, p. 5, Sec. 2.3: “Joint kinematic variability has been shown to be a more sensitive measure of differences between groups than variability of stride characteristics (stride time, step time) [62]”).
Szymczak teaches that the proportional share of gait cycle phases is altered in peripheral artery disease, including that the stance phase is extended and the swing phase is reduced (Szymczak, p. 1, 'Background': “the proportional share of particular phases of the gait cycle is altered. The stance phase is extended, while the swing phase is reduced [6]”). These teachings show that step time, stride time, stance time, and swing time are recognized gait features relevant to peripheral artery disease and can be extracted and used as features for analysis or classification.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Rahman and Szymczak to have the one or more gait features include one or more of step time, stride time, stance time, and swing time. One of ordinary skill in the art would have found it obvious and feasible to extract step time and stride time features and gait phase timing features from Huang’s sensor data because Rahman and Szymczak show these are standard spatiotemporal gait features used in assessing differences between patients with peripheral artery disease and controls, and these features can be computed from gait signals and gait cycle timing. The benefit of the combination would have been improved discrimination of peripheral artery disease related gait abnormalities by using spatiotemporal gait features that are recognized to change with peripheral artery disease, thereby improving diagnostic performance.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Zhang et al. (Zhang, Yuting et al. “Accelerometer-Based Gait Recognition by Sparse Representation of Signature Points With Clusters.” IEEE transactions on cybernetics 45.9 (2015)), hereto referred as Zhang.
The modified Huang teaches claim 1 as described above.
Regarding claim 5, the modified Huang teaches the method of claim 1, but does not expressly teach that the gait characteristic data includes one or more of vertical acceleration data, anterior acceleration data, a number of steps, and a period of time between steps. Rather, the modified Huang teaches collecting multi-axis accelerometer data (including X, Y, and Z axes) and applying a “statistical or machine learning - based classification” to identify gait-related patterns (Huang, ¶[0041]; ¶[0045]: “The 9 - axis inertial motion sensor 74 recorded accelerometer , gyroscope and magnetometer data , each in three dimensions , with the X , Y and Z axes of all three parameters sampled”), but does not expressly teach specifying the gait characteristic data as including vertical acceleration data, anterior acceleration data, a number of steps, or a period of time between steps.
Chidean teaches that acceleration gait analysis for PAD includes evaluating specific acceleration axes corresponding to vertical and anterior-posterior directions, stating that “this difference is statistically significant in different axes depending on the node location, namely, in the X-axis (V acceleration) on the hip, and the Y-axis (AP acceleration) on the ankle” (Chidean, p. 4-5, Sec. 5). Chidean also teaches that “Periodicity can be estimated in the frequency domain using the fundamental frequency, f0 (Hz), which, in pseudo-periodic signals, represents the inverse of the period” (Chidean, p. 3, Periodicity Characterization). This teaching evidences that gait acceleration signals are treated as periodic (or pseudo-periodic) and have an estimable period.
Zhang teaches identifying step cycles and quantifying how many step cycles occur within an accelerometer gait record segment, stating that “we provided the manual annotations of the step cycles” and that “we manually annotated7 the cycle borders at the cyclic valley points of the gait acceleration series recorded at the ankle8”, and further stating that “Normally, 7~14 full step cycles (one left and one right step constitute a full step) existed within the useful segment of a record” (Zhang, p. 1869-1870, Sec. V). Zhang thus shows determining step cycle borders in an acceleration series and deriving a “number of steps” from the step cycles. Zhang further teaches segmenting the record to a “useful” portion and provides a typical duration for that segment, stating “We manually annotated the starting and ending points of the useful segments” and “Normally, the useful length of a record are 7~15s” (Zhang, p. 1869-1870, Sec. V), defining a segment duration that can be used together with step cycle borders to determine time between steps).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean and Zhang to have the gait characteristic data include one or more of vertical acceleration data, anterior acceleration data, a number of steps, and a period of time between steps. One of ordinary skill in the art would have found it obvious and feasible to apply Chidean’s axis-specific treatment of acceleration gait signals to the modified Huang’s three-axis accelerometer data because Huang already acquires tri-axial acceleration signals from a wearable accelerometer, and Chidean merely assigns physiological meaning to individual axes of that same acceleration data by associating one axis with vertical acceleration and another with anterior-posterior acceleration. This modification requires only selecting and labeling existing acceleration components from Huang’s sensor output and does not require any change to Huang’s hardware or data acquisition pipeline. Further, one of ordinary skill in the art would have found it obvious and feasible to apply Zhang’s step cycle identification and counting approach to Huang’s gait acceleration data because Zhang identifies step cycle borders directly from periodic features in the acceleration time series recorded at the ankle, and Huang likewise analyzes accelerometer time series data to identify gait-related patterns. Step cycle borders defined at cyclic valley points in an acceleration signal can be determined using standard peak and valley detection techniques routinely applied to wearable gait acceleration data, enabling determination of a number of steps and timing between steps using the same acceleration data already collected by Huang. The benefit of the combination would have been enabling more complete and standardized gait characterization for PAD-related analysis by incorporating directional acceleration components and step timing metrics derived from wearable accelerometer gait signals, thereby providing a more structured and physiologically meaningful feature set for gait pattern classification across patients.
Claims 7-10, 13, 22-23, and 25-26 are rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Flores et al. (Flores, Alyssa M et al. “Leveraging Machine Learning and Artificial Intelligence to Improve Peripheral Artery Disease Detection, Treatment, and Outcomes.” Circulation research 128.12 (2021)), hereto referred as Flores.
The modified Huang teaches claim 1 as described above.
The modified Huang teaches claim 8 as described above.
Regarding claim 7, the modified Huang teaches the method of claim 1, but does not expressly teach that the step of feeding further includes feeding biometric data of the specific patient to the trained machine learning model. Rather, the modified Huang teaches feeding extracted gait-related features derived from wearable sensor data into a machine learning based classifier to assign a clinical outcome class, but does not expressly teach additionally feeding biometric data of the specific patient into the trained machine learning model as part of the classification process.
Flores teaches incorporating patient specific biometric and physiological data into machine learning models for peripheral artery disease, stating that “A wealth of data is not only accumulating in the health care setting but also from large-scale genetic studies and consumer devices such as wearables and smartphones that may contribute physiological and behavioral data” (Flores, p. 1834, DATA TYPES FOR ML AND AI) and that machine learning models for PAD may integrate “the full breadth of demographic, biological, and clinical data” (Flores, p. 1833; p. 1835-1837, VASCULAR DISEASE DIAGNOSIS). Flores further teaches that patient demographic and biometric attributes such as sex and race are relevant inputs for PAD-related machine learning models, noting that bias must be monitored “in relation to sex, race, and socioeconomic status” and that datasets should reflect “a diverse patient profile” (Flores, p. 1846, Algorithm Bias). Taken together, Flores teaches feeding patient-specific biometric and demographic data as inputs to machine learning models used for PAD-related analysis.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Flores to further feed biometric data of the specific patient to the trained machine learning model. One of ordinary skill in the art would have found it obvious and feasible to incorporate biometric data into Huang’s machine learning workflow because Huang already processes subject specific sensor data and applies machine learning classification, and Flores teaches that machine learning models for PAD commonly integrate physiological and demographic biometric data alongside sensor-derived measurements. Incorporating biometric data would have required only providing additional input features to the existing classifier without altering the sensor hardware or classification architecture. The benefit of the combination would have been enabling more individualized and clinically informative PAD assessment by allowing the trained machine learning model to consider both gait characteristics and patient biometric data when performing classification.
Regarding claim 8, Huang teaches that the system comprises: a processor and a memory storing instructions (Huang, ¶[0005]: "The portable computing device includes: a communications unit to receive the signal streamed according to the short range communication protocol; a non-transitory memory storing machine readable instructions; and a processor to execute the machine readable instructions.", describing a processor with memory); and one or more sensors configured to be worn by a specific patient (Huang, ¶[0004]: "The sensors include a three-axis accelerometer, a three-axis gyroscope and an array of pressure sensors embedded within a wearable device", Huang teaches one or more sensors embedded within a wearable device configured to be worn by a subject); wherein the processor is further configured to: receive acceleration data related to the specific patient generated by the one or more sensors (Huang, ¶[0005]: "The portable computing device includes: a communications unit to receive the signal streamed according to the short range communication protocol; ¶[0006]: "The method includes receiving a signal comprising sensor data from sensors according to a short range communication protocol. The sensors include a three-axis accelerometer", Huang teaches receiving accelerometer-based sensor data as the claimed acceleration data); extract gait characteristic data from the acceleration data related to the specific patient (Huang, ¶[0029]: "The analyzer 45 can perform additional analysis tasks on the preprocessed data, including identifying one or more unstable patterns within the data. The unstable patterns can be associated with one or more of gait", ¶[0005]: “analyze the data to identify a pattern related to gait, balance or posture within the data” Huang teaches extracting gait characteristic information by analyzing the accelerometer data to identify gait-related patterns); feed the extracted gait characteristic data to the trained machine learning model to identify one or more gait features for the specific patient (Huang, ¶[0031]: “one or more pattern recognition classifiers, each of which utilize the extracted features or a subset of the extracted features to determine an appropriate clinical parameter… From the provided feature vector, an outcome class is selected”, ¶[0005]: "apply a statistical or machine learning-based classification to the pattern related to gait, balance or posture to assign a clinical parameter to the pattern", Huang teaches feeding extracted gait related features as a feature vector to a trained classifier to produce an output class).
Also regarding claim 8, Huang does not fully teach a system for diagnosing peripheral artery disease (PAD) in a patient, the system comprising: a processor and a memory storing instructions, which when executed by the processor causes the processor to access or execute a trained a machine learning model, the trained machine learning model having been trained with biometric data and gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD and diagnose the specific patient as having PAD or not having PAD based on the one or more identified gait features. Rather, Huang teaches training a pattern recognition classifier such that "Each classifier is trained on a plurality of training patterns representing various classes of interest" and that training "generally involves a statistical aggregation of training data... into one or more parameters associated with the output class", and further teaches selecting an "outcome class" from a "feature vector" (Huang, ¶[0031]: "Each classifier is trained on a plurality of training patterns representing various classes of interest. The training process of the a given classifier will vary with its implementation, but the training generally involves a statistical aggregation of training data from a plurality of training images into one or more parameters associated with the output class"; Huang, ¶[0031]: "From the provided feature vector, an outcome class is selected"). However, Huang does not expressly teach that the classifier is trained using acceleration derived gait characteristic data for patients known to have PAD and patients that do not have PAD, does not expressly teach using biometric data in the training, and does not expressly teach diagnosing PAD versus not PAD based on the identified gait features.
Chidean teaches PAD specific acceleration gait datasets that distinguish PAD patients from controls, stating "Peripheral arterial disease (PAD) is an artherosclerotic occlusive disorder of distal arteries" and that "full spectra analysis allowed to better characterize gait for both control subjects and PAD patients. Acceleration gait signals were recorded using... four wireless sensor nodes located at ankle and hip height on both sides" (Chidean, Abstract)., and further stating "It allowed to better discriminate PAD patients and control subjects” and that the analysis “could be used for clinical early diagnosis and also to monitor the disease evolution in PAD patients" (Chidean, Abstract; p. 5-7, Sec. 6.).
Flores teaches PAD specific machine learning training using patient biometric and baseline characteristics, stating that "our group developed a classification model that was trained in a hypothesis-free fashion to identify cases of patients with PAD who were previously undiagnosed" and that the "final model included over 120 baseline characteristics", and further teaches ML cohort based classification involving "no PAD (controls)" and using "data on demographics" (Flores, p. 1835-1837, VASCULAR DISEASE DIAGNOSIS).
Rahman provides concrete examples of biometric data collected for PAD and non-PAD cohorts, listing control participant and PAD patient "mass", "height", and "gender" values, and stating that "subjects height, body mass, and anthropometric measures were taken" (Rahman, p. 4, Sec. 2.1: "Thirty healthy older control participants (age: 60.1 ± 8.03 years, age range: 45–74 years, mass: 86.6 ± 16.1 kg, height: 176.7 ± 8.3 cm, gender: 25 males, 5 females) and thirty age-matched, symptomatic patients with PAD-IC... (age: 63.8 ± 9.10 years, age range: 48–80 years, mass: 81.1 ± 14.7 kg, height: 171.8 ± 5.1 cm, gender: 28 males, 2 females... ) participated in the study"; Rahman, p. 4: "subjects height, body mass, and anthropometric measures were taken"). Rahman also teaches that gait variability parameters have “diagnostic ability” to "identify the presence of PAD", stating that “A combination of gait variability parameters correctly identifies PAD-IC disease 70% of the time or more” and that receiver operating characteristic curve analyses were used to discriminate “individuals with PAD-IC from those without PAD-IC” (Rahman, Abstract).
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Huang in view of Chidean, Rahman, and Flores to diagnose the specific patient as having PAD or not having PAD based on one or more gait features identified from acceleration data by training Huang’s statistical or machine learning-based classification using gait characteristic data extracted from acceleration data for patients known to have PAD and patients that do not have PAD and further using biometric data as additional input features. This modification would have been feasible because Huang already teaches training a classifier using training patterns for "classes of interest" and selecting an "outcome class" from a "feature vector", and Chidean provides acceleration gait signals for both PAD patients and control subjects with discrimination between those groups, and Flores teaches PAD specific trained classification models that use demographic and baseline characteristics, and Rahman shows that biometric data such as height, body mass, and gender are routinely collected for PAD and non-PAD cohorts. The benefit of the combination would have been enabling Huang’s wearable sensor based gait classification framework to output PAD status using PAD validated gait acceleration characteristics while incorporating patient biometric variables to improve individualized classification performance and clinical usefulness.
Regarding claim 9, the modified Huang teaches that code and data associated with the trained machine learning model is stored in the memory (Huang, ¶[0004]: "According to an aspect, a non-transitory computer readable medium is described. The non-transitory computer readable medium can comprise computer executable instructions that when executed by a processing resource facilitate the performance of operations", ¶[0005]: "The portable computing device includes: a communications unit to receive the signal streamed according to the short range communication protocol; a non-transitory memory storing machine readable instructions; and a processor to execute the machine readable instructions", Huang teaches code in the form of machine readable instructions stored in non-transitory memory for executing the machine learning-based classification; ¶[0030]: "However, in most examples, the classifier 46 can access historical data related to the unstable pattern identified by the analyzer", Huang teaches data associated with the classifier, including historical data used for the machine learning-based classification).
Regarding claim 10, the modified Huang teaches that the machine learning model comprises one of a neural network algorithm, a random forest algorithm, a support vector machine (SVM) algorithm and a Logit algorithm (Huang, ¶[0031]: “Any of a variety of optimization techniques can be utilized for the classification algorithm, including support vector machines, self-organized maps, fuzzy logic systems, data fusion processes, ensemble methods, statistical or machine learning-based systems or artificial neural networks”, Huang expressly teaches support vector machines and artificial neural networks as example machine learning algorithms for the classification algorithm; ¶[0032]: “a support vector machine (SVM) classifier can process the training data” , Huang expressly teaches a support vector machine classifier; ¶[0032]: “a convolutional neural network (CNN) classifier can process the training data”, Huang expressly teaches a neural network classifier as an example of the machine learning model).
Regarding claim 13, the modified Huang teaches the system of claim 8, but does not expressly teach that the instructions to feed the extracted gait characteristic data to the trained machine learning model further include instructions to feed biometric data of the specific patient to the trained machine learning model. Rather, the modified Huang teaches feeding extracted gait-related features derived from wearable sensor data into a machine learning based classifier to assign a clinical outcome class, but does not expressly teach additionally feeding biometric data of the specific patient into the trained machine learning model as part of the classification process.
Flores teaches incorporating patient specific biometric and physiological data into machine learning models for peripheral artery disease, stating that “A wealth of data is not only accumulating in the health care setting but also from large-scale genetic studies and consumer devices such as wearables and smartphones that may contribute physiological and behavioral data” (Flores, p. 1834, DATA TYPES FOR ML AND AI) and that machine learning models for PAD may integrate “the full breadth of demographic, biological, and clinical data” (Flores, p. 1833; p. 1835-1837, VASCULAR DISEASE DIAGNOSIS). Flores further teaches that patient demographic and biometric attributes such as sex and race are relevant inputs for PAD-related machine learning models, noting that bias must be monitored “in relation to sex, race, and socioeconomic status” and that datasets should reflect “a diverse patient profile” (Flores, p. 1846, Algorithm Bias). Taken together, Flores teaches feeding patient-specific biometric and demographic data as inputs to machine learning models used for PAD-related analysis.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Flores to further feed biometric data of the specific patient to the trained machine learning model. One of ordinary skill in the art would have found it obvious and feasible to incorporate biometric data into Huang’s machine learning workflow because Huang already processes subject specific sensor data and applies machine learning classification, and Flores teaches that machine learning models for PAD commonly integrate physiological and demographic biometric data alongside sensor-derived measurements. Incorporating biometric data would have required only providing additional input features to the existing classifier without altering the sensor hardware or classification architecture. The benefit of the combination would have been enabling more individualized and clinically informative PAD assessment by allowing the trained machine learning model to consider both gait characteristics and patient biometric data when performing classification.
Regarding claim 22, the modified Huang does not fully teach selecting or modifying a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient. Rather, the modified Huang teaches assigning a clinical parameter based on machine learning classification of extracted gait related features and using the clinical parameter to “determine, track or ensure the subject’s compliance with certain exercise or rehabilitation programs”, thereby supporting use of a gait derived clinical parameter in connection with an exercise or rehabilitation program (Huang, ¶[0036]: “The clinical parameter, as another example, can be used to determine, track or ensure the subject’s compliance with certain exercise or rehabilitation programs”). However, Huang does not expressly teach selecting or modifying a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient.
Flores teaches selecting an appropriate treatment strategy for PAD patients and using machine learning to support PAD treatment planning, stating that “the increased costs of newer treatments and their attendant risks must be considered when selecting an appropriate treatment strategy and prescribing newer agents” and that “ML models can assist with matching preventative efforts to patients who may most benefit from new or existing drugs, and achieve appropriate polypharmacy, which is especially prevalent in the highly comorbid PAD population”, and further stating that “With numerous treatment guidelines for PAD and its frequent comorbid conditions, training such algorithms on patients with PAD can help develop plans that synchronize therapy and limit interactions leading to side effects” (Flores, p. 1840, 'Identifying Appropriate Medical Treatment for Vascular Disease').
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Flores to select or modify a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient. This modification would have been feasible because the modified Huang already produces a clinical parameter from gait related feature classification and applies that clinical parameter in the context of ensuring compliance with an exercise or rehabilitation program, and to determine a severity of PAD in the specific patient or a change in severity of PAD in the specific patient based on gait characteristics data, and Flores teaches selecting an appropriate treatment strategy and using machine learning outputs to match treatment efforts to patients, thereby providing an express rationale to use the determined PAD severity or progression output as an input to treatment course selection or modification. The benefit of the combination would have been enabling more personalized and responsive PAD management by adjusting exercise, rehabilitation, or medical therapy strategies based on severity or changes in severity derived from objective gait analysis.
Regarding claim 23, the modified Huang teaches providing, via a display device of the medical device, an alert to the specific patient based on (i) the specific patient being diagnosed as having PAD, or (ii) the severity of PAD in the specific patient, or (iii) the change in severity of PAD in the specific patient (Huang, ¶[0027]: "The user interface 43 can display visual or audio information to the user. The user interface 43 can also receive inputs from the user. As an example, the user interface 43 can be a touch screen that can both display visual information and receive user inputs"; Huang, ¶[0037]: "The alert generator 47 can be used in some examples to provide an alert based on the clinical parameter...However, when the alert generator 47 is employed, the alert can be, for example, a tactile signal (e.g., vibration of the mobile device), an audio signal or a visual signal"; Huang teaches a user interface (display device) that displays visual information to the user, and an alert generator that provides an alert (including a visual signal) based on the clinical parameter, which corresponds to the diagnosed PAD status, PAD severity, or change in PAD severity as determined by the modified Huang from claims 21-22).
Regarding claim 25, the modified Huang does not fully teach that the processor is further configured to select or modify a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient. Rather, the modified Huang teaches assigning a clinical parameter based on machine learning classification of extracted gait related features and using the clinical parameter to “determine, track or ensure the subject’s compliance with certain exercise or rehabilitation programs”, thereby supporting use of a gait derived clinical parameter in connection with an exercise or rehabilitation program (Huang, ¶[0036]: “The clinical parameter, as another example, can be used to determine, track or ensure the subject’s compliance with certain exercise or rehabilitation programs”). However, Huang does not expressly teach selecting or modifying a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient.
Flores teaches selecting an appropriate treatment strategy for PAD patients and using machine learning to support PAD treatment planning, stating that “the increased costs of newer treatments and their attendant risks must be considered when selecting an appropriate treatment strategy and prescribing newer agents” and that “ML models can assist with matching preventative efforts to patients who may most benefit from new or existing drugs, and achieve appropriate polypharmacy, which is especially prevalent in the highly comorbid PAD population”, and further stating that “With numerous treatment guidelines for PAD and its frequent comorbid conditions, training such algorithms on patients with PAD can help develop plans that synchronize therapy and limit interactions leading to side effects” (Flores, p. 1840, 'Identifying Appropriate Medical Treatment for Vascular Disease').
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Flores to configure the processor to select or modify a treatment course for the specific patient based on the severity of PAD in the specific patient or the change in severity of PAD in the specific patient. This modification would have been feasible because the modified Huang already produces a clinical parameter from gait related feature classification and applies that clinical parameter in the context of ensuring compliance with an exercise or rehabilitation program, and to determine a severity of PAD in the specific patient or a change in severity of PAD in the specific patient based on gait characteristics data, and Flores teaches selecting an appropriate treatment strategy and using machine learning outputs to match treatment efforts to patients, thereby providing an express rationale to use the determined PAD severity or progression output as an input to treatment course selection or modification. The benefit of the combination would have been enabling more personalized and responsive PAD management by adjusting exercise, rehabilitation, or medical therapy strategies based on severity or changes in severity derived from objective gait analysis.
Regarding claim 26, the modified Huang teaches that the processor is further configured to provide an alert on a display device based on the specific patient being diagnosed as having PAD, or based on the severity of PAD in the specific patient, or based on the change in severity of PAD in the specific patient (Huang, ¶[0027]: "The user interface 43 can display visual or audio information to the user. The user interface 43 can also receive inputs from the user. As an example, the user interface 43 can be a touch screen that can both display visual information and receive user inputs"; Huang, ¶[0037]: "The alert generator 47 can be used in some examples to provide an alert based on the clinical parameter...However, when the alert generator 47 is employed, the alert can be, for example, a tactile signal (e.g., vibration of the mobile device), an audio signal or a visual signal"; Huang teaches a user interface (display device) that displays visual information to the user, and an alert generator that provides an alert (including a visual signal) based on the clinical parameter, which corresponds to the diagnosed PAD status, PAD severity, or change in PAD severity as determined by the modified Huang from claims 23-24).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Flores et al. (Flores, Alyssa M et al. “Leveraging Machine Learning and Artificial Intelligence to Improve Peripheral Artery Disease Detection, Treatment, and Outcomes.” Circulation research 128.12 (2021)), hereto referred as Flores, and further in view of Szymczak et al. (Szymczak M, Krupa P, Oszkinis G, Majchrzycki M. Gait pattern in patients with peripheral artery disease. BMC Geriatr. 2018 Feb 20), hereto referred as Szymczak.
The modified Huang teaches claim 8 as described above.
Regarding claim 11, the modified Huang teaches the system of claim 8, but does not expressly teach that the one or more gait features include one or more of step time asymmetry, step time variability, step time, stance time, stride time, and swing time. Rather, the modified Huang teaches measuring sensor based parameters from a Portable Gait Lab system, including timing and symmetry related parameters, but does not expressly teach using gait features that include step time, stride time, stance time, or swing time.
Rahman teaches that stride time and step time are recognized stride characteristics used in assessing gait variability in patients with peripheral artery disease (Rahman, p. 5, Sec. 2.3: “Joint kinematic variability has been shown to be a more sensitive measure of differences between groups than variability of stride characteristics (stride time, step time) [62]”).
Szymczak teaches that the proportional share of gait cycle phases is altered in peripheral artery disease, including that the stance phase is extended and the swing phase is reduced (Szymczak, p. 1, 'Background': “the proportional share of particular phases of the gait cycle is altered. The stance phase is extended, while the swing phase is reduced [6]”). These teachings show that step time, stride time, stance time, and swing time are recognized gait features relevant to peripheral artery disease and can be extracted and used as features for analysis or classification.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Rahman and Szymczak to have the one or more gait features include one or more of step time, stride time, stance time, and swing time. One of ordinary skill in the art would have found it obvious and feasible to extract step time and stride time features and gait phase timing features from Huang’s sensor data because Rahman and Szymczak show these are standard spatiotemporal gait features used in assessing differences between patients with peripheral artery disease and controls, and these features can be computed from gait signals and gait cycle timing. The benefit of the combination would have been improved discrimination of peripheral artery disease related gait abnormalities by using spatiotemporal gait features that are recognized to change with peripheral artery disease, thereby improving diagnostic performance.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Flores et al. (Flores, Alyssa M et al. “Leveraging Machine Learning and Artificial Intelligence to Improve Peripheral Artery Disease Detection, Treatment, and Outcomes.” Circulation research 128.12 (2021)), hereto referred as Flores, and further in view of Zhang et al. (Zhang, Yuting et al. “Accelerometer-Based Gait Recognition by Sparse Representation of Signature Points With Clusters.” IEEE transactions on cybernetics 45.9 (2015)), hereto referred as Zhang.
The modified Huang teaches claim 8 as described above.
Regarding claim 12, the modified Huang teaches the system of claim 8, but does not expressly teach that the gait characteristic data includes one or more of vertical acceleration, anterior acceleration, a number of steps, a period of time between steps, braking impulse, braking peak, propulsive peak, propulsive impulse, other forces derived from ground reaction forces (GRF) data, and joint torques and powers, including hip, knee and/or ankle angles, torques and powers. Rather, the modified Huang teaches collecting multi-axis accelerometer data (including X, Y, and Z axes) and applying a “statistical or machine learning - based classification” to identify gait-related patterns (Huang, ¶[0041]; ¶[0045]: “The 9 - axis inertial motion sensor 74 recorded accelerometer , gyroscope and magnetometer data , each in three dimensions , with the X , Y and Z axes of all three parameters sampled”), but does not expressly teach labeling the accelerometer axes as “vertical” or “anterior”, does not expressly teach extracting a “number of steps” or a “period of time between steps”, and does not expressly teach GRF-derived braking/propulsive impulses or peaks, or joint torques and powers
Chidean teaches that acceleration gait analysis for PAD includes evaluating specific acceleration axes corresponding to vertical and anterior-posterior directions, stating that “this difference is statistically significant in different axes depending on the node location, namely, in the X-axis (V acceleration) on the hip, and the Y-axis (AP acceleration) on the ankle” (Chidean, p. 4-5, Sec. 5). Chidean also teaches that “Periodicity can be estimated in the frequency domain using the fundamental frequency, f0 (Hz), which, in pseudo-periodic signals, represents the inverse of the period” (Chidean, p. 3, Periodicity Characterization). This teaching evidences that gait acceleration signals are treated as periodic (or pseudo-periodic) and have an estimable period.
Zhang teaches identifying step-cycle borders from ankle acceleration time series, stating “we manually annotated7 the cycle borders at the cyclic valley points of the gait acceleration series recorded at the ankle8” (Zhang, p. 1869-1870, Sec. V), and further states that “Normally, 7~14 full step cycles... existed within the useful segment of a record” and that “Normally, the useful length of a record are 7~15s” (Zhang, p. 1869-1870, Sec. V). These teachings evidence that step cycles may be identified within an acceleration record segment and that record duration may be used in analyzing those step cycles. Zhang further teaches segmenting the record to a “useful” portion and provides a typical duration for that segment, stating “We manually annotated the starting and ending points of the useful segments” and “Normally, the useful length of a record are 7~15s” (Zhang, p. 1869-1870, Sec. V), defining a segment duration that can be used together with step cycle borders to determine time between steps).
Rahman teaches biomechanical joint angle measurements during gait in PAD-related analysis, stating “Ranges of motion of the ankle, knee, and hip joint angles were calculated for the gait cycles in every trial” (Rahman, p. 5, Sec. 2.3.1). Rahman thus teaches extracting kinematic biomechanics data, including hip, knee, and ankle joint angles, from gait data for patients with PAD.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Chidean, Zhang, and Rahman to extract additional gait features from Huang’s already-collected sensor data, including (i) directional acceleration components and periodicity features from the accelerometer axes, and (ii) kinematic joint angle features for the lower extremity. The combination would have been feasible because Huang already records three-dimensional accelerometer data using a wearable inertial motion sensor and analyzes that data to identify gait-related patterns using machine learning classification, Chidean teaches interpreting specific accelerometer axes as vertical and anterior–posterior acceleration and characterizes gait signals as periodic with an estimable period f0 corresponding to step frequency, Zhang teaches identifying step-cycle borders directly from ankle acceleration time series and counting full step cycles within a record segment of known duration, and Rahman teaches calculating ankle, knee, and hip joint angle kinematics from gait cycles, such that each of the additional claimed gait characteristics can be derived from the same wearable sensor data streams already collected and processed by Huang without substantially altering Huang’s sensor hardware or data acquisition pipeline. The benefit of the combination would have been enabling a richer feature set (directional acceleration, periodicity/step-cycle features, and joint-angle kinematics) for improved gait characterization and subsequent classification.
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Huang et al. (US 20180279915 A1), hereto referred as Huang, and further in view of Chidean et al. (Chidean, Mihaela I et al. “Full Band Spectra Analysis of Gait Acceleration Signals for Peripheral Arterial Disease Patients.” Frontiers in physiology 9 (2018)), hereto referred as Chidean, and further in view of Rahman et al. (Rahman, Hafizur et al. “Gait Variability Is Affected More by Peripheral Artery Disease than by Vascular Occlusion.” PloS one 16.3 (2021)), hereto referred as Rahman, and further in view of Watanabe et al. (Watanabe, Yuji, and Masaki Kimura. “Gait Identification and Authentication Using LSTM Based on 3-Axis Accelerations of Smartphone.” Procedia Computer Science 176 (2020)), hereto referred as Watanabe.
The modified Huang teaches claim 14 as described above.
Regarding claim 15, Huang teaches the computer-implemented method of claim 14, but does not expressly teach that the machine learning model comprises a recurrent neural network or a long short-term memory (LSTM) model. Rather, Huang teaches applying statistical or machine learning-based classification to extracted gait-related features, including artificial neural networks as a general class of machine learning algorithms (Huang, ¶[0031]: “Any of a variety of optimization techniques can be utilized for the classification algorithm, including … statistical or machine learning-based systems or artificial neural networks”), but does not expressly teach using a recurrent neural network or a long short-term memory (LSTM) model for processing sequential acceleration data.
Watanabe teaches explicitly applying recurrent neural networks and long short-term memory models to time-series acceleration gait data, stating “We use DRNN and LSTM, which are deep learning methods” and further stating “Recurrent neural networks RNN receive time series data in the input layer” and that “One of the methods to solve this problem is Long Short-Term Memory (LSTM)” (Watanabe, p. 3875, Sec. 3.2). Watanabe further teaches that “the 3-axis accelerations of a window of 200 samples is directly applied to the input layer” and that “The hidden layer uses LSTM blocks” (Watanabe, p. 3875, Sec. 3.2), demonstrating use of LSTM architectures on sequential accelerometer gait data.
It would have been prima facie obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have further modified the modified Huang in view of Watanabe to implement the machine learning model as a recurrent neural network or a long short-term memory (LSTM) model for processing acceleration data associated with gait. The modification would have been possible and feasible because Huang already acquires and processes time-series accelerometer data and applies machine learning-based classification to gait-related patterns, and Watanabe demonstrates that recurrent neural network and LSTM architectures can be directly applied to raw 3-axis acceleration gait time series without requiring changes to the underlying sensor hardware or data acquisition pipeline. The benefit of the combination would have been improved modeling of temporal dependencies in gait acceleration signals by capturing sequential relationships over time, thereby improving classification performance for gait-based analysis.
Conclusion
Any inquiry concerning this communication or earlier communications from the
examiner should be directed to AARON MERRIAM whose telephone number is (703) 756-
5938. The examiner can normally be reached M-F 8:00 am - 5:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing
using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is
encouraged to use the USPTO Automated Interview Request (AIR) at
http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s
supervisor, Jason Sims can be reached on (571)272-4867. The fax phone number for the
organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained
from Patent Center. Unpublished application information in Patent Center is available to
registered users. To file and manage patent submissions in Patent Center, visit:
https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for
more information about Patent Center and https://www.uspto.gov/patents/docx for
information about filing in DOCX format. For additional questions, contact the Electronic
Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO
Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/AARON MERRIAM/Examiner, Art Unit 3791
/JASON M SIMS/Supervisory Patent Examiner, Art Unit 3791