DETAILED ACTION
This is responsive to application 19/166, 421 filed on 09/17/2025 in which claims 1-15 are presented for examination.
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 .
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 9-12, and 14 are rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim 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.
Regarding claims 9, the claim recites “- selecting a second time-series of values of the physiological parameter from the training data, wherein the second time series is part of a same data stream as the first time- series of values of the physiological parameter and is ahead in time of the first time series; and- training the machine learning model to predict the second time-series of values of the physiological parameter from the first time-series of values of the physiological parameter as input.”
It is not clear, how the second time series of values is part of a same data stream as the first time-series, and it is being selected; but then, the model is predicting the same second time-series of value? For compact prosecution, the examiner have interpreted the claim language in similar manner as claim 1, where the predicted second time-series of values is forecasted data value based on input of the first time-series of values.
Claim 14 has same issue as described with regard to claim 9 above, and it is treated in similar manner.
Claims 10-12 are rejected based on the rejected base claim 9.
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 1-15 are rejected under 35 U.S.C. 101 because the claimed invention is directed to an abstract idea without significantly more.
Regarding claim 1:
Step 1: Is the claim to a process, machine, manufacture or composition of matter?” Yes, it’s a system (machine) claim .
Step 2a Prong 1 (judicial exception)
Step 2A (1): “Does the claim recite an abstract idea, law of nature, or natural phenomenon? Yes , the claim comes under mental processes and mathematical concepts.
Claim 1 recites:
“A prediction system for providing a prediction for a physiological parameter of a patient and for outputting the prediction to a clinical decision support tool , comprising: an input interface configured to access a real-time or near real-time data stream of measured values of a physiological parameter of the patient; - a data storage configured to access a real-time or near real-time data stream of measured values of a physiological parameter of the patient; - a data storage comprising data defining machine learning model, wherein the machine learning model is configured to predict, using a first time-series of values of the physiological parameter as input, a second time-series of values of the physiological parameter, wherein the second time-series represents a numerical prediction of the values of the physiological parameter ahead in time of the first time series; - an output interface to the clinical decision support tool; - a processor subsystem configured to:- use the machine learning model to generate, using a time-series of measured values from the real-time or near real-time stream as input, a time-series of predicted values as output;- determine for the time-series of predicted values whether predicted values fall within a physiological range associated with the physiological parameter, thereby obtaining a time-series of classifications wherein a respective classification is indicative of whether a predicted value falls inside or outside of the physiological range; and- provide the time-series of classifications to the clinical decision support tool.
All the limitations above are abstract idea related to the mental process (concepts performed in the human mind (including an observation, evaluation, judgment, opinion) with the exception of bold and underlined limitations. Claim language pertains to analyzing patient’s data and predict/forecast future vital-sign values. Forecasting using known data is a mathematical concept. e.g. using regression . All this can be done using paper and pen and mentally.
Step 2A(2): Prong Two: evaluate whether the claim recites additional elements that integrate the exception into a practical application of the exception. NO
The claim does recite additional elements; however they don’t integrate the exception into a practical application of the exception.
prediction system (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
clinical decision support tool (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
an input interface configured to access.... (Adding insignificant extra-solution activity to the judicial exception - see MPEP 2106.05(g) )
real-time data stream(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
data storage(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
time-series of classifications(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
machine learning model(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
output interface(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
clinical decision support tool(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
processor subsystem(Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f))
Step 2B: evaluate whether the claim recites additional elements that amount to an inventive concept (aka “significantly more”) than the recited judicial exception? NO
As discussed previously with respect to Step 2A Prong Two, the additional element in the claim amounts to no more than mere instructions to apply the exception using a generic computer component.
Regarding the claim limitation,“ : an input interface configured to access....” the courts have recognized the computer functions as well‐understood, routine, and conventional functions when they are claimed in a merely generic manner (“i. Receiving or transmitting data over a network, e.g., using the Internet to gather data, Symantec, 838 F.3d at 1321, 120 USPQ2d at 1362 (utilizing an intermediary computer to forward information”); See, MPEP 2106.05 (d)(II)
The same analysis applies here in 2B, i.e., mere instructions to apply an exception using a generic computer component cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Dependent claims 2-8 further narrows the abstract idea defined in claim 1, and add the additional element of “patient monitor”,
Under step 2A, prong two, the additional elements don’t integrate the exception into a practical application of the exception as merely adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f).
As discussed previously with respect to Step 2A Prong Two, the additional elements in the claim amounts to no more than mere instructions to apply the exception using a generic computer component.
The same analysis applies here in 2B, i.e., mere instructions to apply an exception using a generic computer component cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Regarding claim 9, it is rejected under the same rationale as claim 1. In addition it adds the additional elements of “training system”, “recording of real-time or near real-time data streams”.
Under step 2A, prong two, the additional elements don’t integrate the exception into a practical application of the exception as merely adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f).
As discussed previously with respect to Step 2A Prong Two, the additional elements in the claim amounts to no more than mere instructions to apply the exception using a generic computer component.
The same analysis applies here in 2B, i.e., mere instructions to apply an exception using a generic computer component cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Dependent claims 10-12 further narrows the abstract idea defined in claim 9, and add the additional element of “training iterations”, “outlier detection technique”, “recurrent neural network,”, “long short-term memory neural network”, “a logistic regression-based model”, “decision-tree based model”.
Under step 2A, prong two, the additional elements don’t integrate the exception into a practical application of the exception as merely adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f).
As discussed previously with respect to Step 2A Prong Two, the additional elements in the claim amounts to no more than mere instructions to apply the exception using a generic computer component.
The same analysis applies here in 2B, i.e., mere instructions to apply an exception using a generic computer component cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Regarding claim 13, it is rejected under the same rationale as claim 1. It is a method claim.
Dependent claim 14, further narrows the abstract idea defined in claim 13.
Regarding claim 15, it is rejected under the same rationale as claim 13 above, as it substantially recites the same claim limitations. It is also a method claim, thus falls under process category. In addition, the claim fails the step 1, analysis, as the claim recites “a transitory or non-transitory computer-readable medium”, under broadest reasonable interpretation the claim could be interpreted as a transitory computer readable medium, which is signal per se, and doesn’t fall under one of the statutory categories. In addition, the claim recites additional element of “a transitory or non-transitory computer-readable medium”, which is being used as a tool (Adding the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea - see MPEP 2106.05(f)))
The same analysis applies here in 2B, i.e., mere instructions to apply an exception using a generic computer component cannot integrate a judicial exception into a practical application at Step 2A or provide an inventive concept in Step 2B.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1-3, 6-9, and 12-15 are rejected under 35 U.S.C. 102a(1) as anticipated by Khare et al. (US 20220323855 A1)
Regarding claim 1, Khare teaches a prediction system for providing a prediction for a physiological parameter of a patient and for outputting the prediction to a clinical decision support tool (Fig. 6, Fig. 7A), comprising:
- an input interface configured to access a real-time or near real-time data stream of measured values of a physiological parameter of the patient (Para, “[0007] In another aspect, a system for generating and providing simulated animal data by executing the methods herein is provided. The system including a computing device is operable to execute steps of receiving one or more sets of real animal data at least partially obtained from one or more sensors that receive, store, or send information related to one or more targeted individuals;..”
Para, “[0071] ... A biosensor can gather biological data (including readings and signals) such as physiological data, biometric data, chemical data, biomechanical data, genetic data, genomic data, location data or other biological data from one or more targeted individuals...”);
- a data storage comprising data defining machine learning model, wherein the machine learning model is configured to predict, using a first time-series of values of the physiological parameter as input, a second time-series of values of the physiological parameter, wherein the second time-series represents a numerical prediction of the values of the physiological parameter ahead in time of the first time series (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data. ...... Utilizing a GAN, the generator generates one or more new data values, which may comprise one or more new data sets, while the discriminator evaluates the one or more new values based on one or more user-defined criteria to certify, validate, or authenticate the newly created values.”);
- an output interface to the clinical decision support tool (Para, “[0140]...The computing device includes a network connection (e.g., internet) and browser application (e.g., executing browser software). The connection application, which executes within the browser, is configured (e.g., programmed) to establish one or more wireless communication links with each of the one or more sensors, receive one or more streams of data from the one or more sensors, and display, via the browser application, one or more readings derived from at least a portion of the streamed data.”
Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”);
- a processor subsystem configured to (See, Fig. 1, para 0041, 0043):
- use the machine learning model to generate, using a time-series of measured values from the real-time or near real-time stream as input, a time-series of predicted values as output (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data...”
Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”);
- determine for the time-series of predicted values whether predicted values fall within a physiological range associated with the physiological parameter, thereby obtaining a time-series of classifications, wherein a respective classification is indicative of whether a predicted value falls inside or outside of the physiological range (Para, “0112] In another refinement, the simulated data may be assigned to one or more classifications. Classifications (e.g., including groups) can be created to simplify the search process for a data acquirer (e.g., as one or more searchable tags) and may be based on data collection processes, practices, quality, or associations, as well as targeted individual and simulated targeted individual characteristics. Classifications can be identifiers for data...... Another classification may be assigned to an artificial data set that is representative of ECG data from a specific sensor with specific settings and following a specific data collection methodology. In another example, a classification may be created for data sets representing targeted individuals that have previously experienced a stroke, or for simulated data sets representing simulated targeted individuals that are based upon, at least in part, real-world targeted individuals..... simulated data range classifications (providing a range for the data, e.g., bilirubin levels between 0.2-1.2 mg/dL..”
Para, “[0078] ... Thresholds involve generally accepted values or principles (e.g., it may be established that a male over 90 years old should contact their doctor if their heart rate reaches over 200 beats per minute, or that the age-based max heart rate for a 33-year old male is n beats per minute). FIG. 3A provides a graph of the heart rate beats per minute (BPM) captured from a professional athlete, while FIG. 3B provides the autocorrelation function for the data in FIG. 3A.”); and
- provide the time-series of classifications to the clinical decision support tool (Para 0117, “...The real-world data may indicate the fatigue range of an athlete based on distance run or length of any given match. This information, in turn, can then be utilized by the simulation system to adjust the “energy level” within the game. This data may be utilized, for example, to gain an advantage within the game...” Also see, Claim 46.).
Regarding claim 2, Khare teaches the prediction system according to claim 1.
Khare further teaches wherein the time-series of predicted values spans a time period selected between seconds and tens of minutes, for example selected between 10 seconds and 1 hour (Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets.”
Para 0109, “...For example, in the context of tennis, an acquirer may want 1 hour of Player A's heart rate data when the temperature is at or above 95 degrees Fahrenheit for the entirety of a two-hour match.”
Para, “[0083] .... animal data readings are timestamped and occur at a predetermined time period (e.g., approximately every second). Initially, the model is trained using N such observations (length of LSTM sequence), which can be a few (e.g., 20), a couple hundred, thousands, millions, and more.” Note: from above citations, it is clear that tiem period can be predefined, and it can be any interval.)
Regarding claim 3, Khare teaches the prediction system according to claim 1.
Khare further teaches wherein the time-series of predicted values has a time resolution selected between half a second and minutes, for example comprising predicted values for every second, every 10 seconds, every minute, or every five minutes (Fig. 7A teaches predicted value for every second; Also, “[0083] .... The animal data readings are timestamped and occur at a predetermined time period (e.g., approximately every second). Initially, the model is trained using N such observations (length of LSTM sequence), which can be a few (e.g., 20), a couple hundred, thousands, millions, and more.”
Para 0090, “...For example, collecting animal data, situational context information related to the animal data, and trends based on the real-time or near real-time animal data enables comparison of real-time or near real-time data to historical data in the situational equivalent while enabling evaluation of both microtrends (e.g., within seconds or minutes) and macrotrends (e.g., full game).....” )
Regarding claim 6, Khare teaches the prediction system according to claim 1.
Khare further teaches wherein the machine learning model is trained to predict the time-series of values of the physiological parameter using a further time series of values of a further physiological parameter as further input (Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.” Note: data is predicted each second.);
Regarding claim 7, Khare teaches the prediction system according to claim 1.
Khare further teaches wherein the physiological parameter is one of: temperature, heart rate, pulse rate, ST- segment deviation, invasive arterial blood pressure, central venous pressure, respiratory rate, tidal volume, end-tidal CO2 concentration, train of four ratio, bispectral index, oxygen saturation, and non-invasive arterial blood pressure (Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”)
Regarding claim 8, Khare teaches a patient monitor comprising the prediction system according to claim 1 (See, claim 1 mapping; para, “[0093] ..... Advantageously, the simulated data can be used in an animal data prediction system, with a particular focus on wagering applications as well as probability assessment systems related to healthcare, telehealth, insurance, fitness, health/wellness monitoring, and the like...”)
Regarding claim 9, Khare teaches a training system for training a machine learning model to predict future values of a physiological parameter of a patient based on past measured values of the physiological parameter (see, Fig. 6), comprising:
- an input interface configured to access training data in form of recordings of real-time or near real-time data streams of measured values of a physiological parameter of one or more patients (Para, “[0007] In another aspect, a system for generating and providing simulated animal data by executing the methods herein is provided. The system including a computing device is operable to execute steps of receiving one or more sets of real animal data at least partially obtained from one or more sensors that receive, store, or send information related to one or more targeted individuals;..”
Para, “[0071] ... A biosensor can gather biological data (including readings and signals) such as physiological data, biometric data, chemical data, biomechanical data, genetic data, genomic data, location data or other biological data from one or more targeted individuals...”);
- a processor subsystem configured to iteratively train the machine learning model on the training data to obtain a machine learning model by (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data. ...... Utilizing a GAN, the generator generates one or more new data values, which may comprise one or more new data sets, while the discriminator evaluates the one or more new values based on one or more user-defined criteria to certify, validate, or authenticate the newly created values.”), in a training iteration:
- selecting a first time-series of values of the physiological parameter from the training data (Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets..” Note: here the system allows to include any number of seconds of data.);
- selecting a second time-series of values of the physiological parameter from the training data, wherein the second time series is part of a same data stream as the first time- series of values of the physiological parameter and is ahead in time of the first time series ((Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets..” Note: here the system allows selecting any subintervals. Also, note here claim is not explicit about first interval with first network, and second interval with second network, and combining etc... for example see claims 4-5 which can be explicitly addressed by Redell reference.); and
-training the machine learning model to predict the second time-series of values of the physiological parameter from the first time-series of values of the physiological parameter as input (Para, “[0140]...The computing device includes a network connection (e.g., internet) and browser application (e.g., executing browser software). The connection application, which executes within the browser, is configured (e.g., programmed) to establish one or more wireless communication links with each of the one or more sensors, receive one or more streams of data from the one or more sensors, and display, via the browser application, one or more readings derived from at least a portion of the streamed data.”
Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”).
Regarding claim 12, Khare teaches the training system according to claim 9.
Khare further teaches wherein the machine learning model is one of: recurrent neural network, such as a long short-term memory neural network, a logistic regression-based model, and a decision-tree based model (Para, “[0021] FIG. 5 provides details of a recurrent neural network that can be used for simulated animal data generation.”
Para, “[0022] FIG. 6 provides a schematic of a Long Short-Term Memory (LSTM) network that can be used to generate simulated animal data.”)
Regarding claim 13, Khare teaches a computer-implemented method for providing a prediction for a physiological parameter of a patient and for providing the prediction to a clinical decision support tool (Fig. 6, Fig. 7A), comprising:
- accessing a real-time or near real-time data stream of measured values of a physiological parameter of the patient (Para, “[0007] In another aspect, a system for generating and providing simulated animal data by executing the methods herein is provided. The system including a computing device is operable to execute steps of receiving one or more sets of real animal data at least partially obtained from one or more sensors that receive, store, or send information related to one or more targeted individuals;..”
Para, “[0071] ... A biosensor can gather biological data (including readings and signals) such as physiological data, biometric data, chemical data, biomechanical data, genetic data, genomic data, location data or other biological data from one or more targeted individuals...”);
- accessing data defining machine learning model, wherein the machine learning model is configured to predict, using a first time-series of values of the physiological parameter as input, a second time-series of values of the physiological parameter, wherein the second time- series represents a numerical prediction of the values of the physiological parameter ahead in time of the first time series (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data. ...... Utilizing a GAN, the generator generates one or more new data values, which may comprise one or more new data sets, while the discriminator evaluates the one or more new values based on one or more user-defined criteria to certify, validate, or authenticate the newly created values.”);
- using the machine learning model to generate, using a time-series of measured values from the real-time or near real-time stream as input, a time-series of predicted values as output (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data...”
Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”);
- determining for the time-series of predicted values whether predicted values fall within a physiological range associated with the physiological parameter, thereby obtaining a time- series of classifications, wherein a respective classification is indicative of whether a predicted value falls inside or outside of the physiological range (Para, “0112] In another refinement, the simulated data may be assigned to one or more classifications. Classifications (e.g., including groups) can be created to simplify the search process for a data acquirer (e.g., as one or more searchable tags) and may be based on data collection processes, practices, quality, or associations, as well as targeted individual and simulated targeted individual characteristics. Classifications can be identifiers for data...... Another classification may be assigned to an artificial data set that is representative of ECG data from a specific sensor with specific settings and following a specific data collection methodology. In another example, a classification may be created for data sets representing targeted individuals that have previously experienced a stroke, or for simulated data sets representing simulated targeted individuals that are based upon, at least in part, real-world targeted individuals..... simulated data range classifications (providing a range for the data, e.g., bilirubin levels between 0.2-1.2 mg/dL..”
Para, “[0078] ... Thresholds involve generally accepted values or principles (e.g., it may be established that a male over 90 years old should contact their doctor if their heart rate reaches over 200 beats per minute, or that the age-based max heart rate for a 33-year old male is n beats per minute). FIG. 3A provides a graph of the heart rate beats per minute (BPM) captured from a professional athlete, while FIG. 3B provides the autocorrelation function for the data in FIG. 3A.”); and
- providing the time-series of classifications to the clinical decision support tool (Para 0117, “...The real-world data may indicate the fatigue range of an athlete based on distance run or length of any given match. This information, in turn, can then be utilized by the simulation system to adjust the “energy level” within the game. This data may be utilized, for example, to gain an advantage within the game...” Also see, Claim 46.).
Regarding claim 14, Khare teaches a computer-implemented method for training a machine learning model to predict future values of a physiological parameter of a patient based on past measured values of the physiological parameter (see, Fig. 6, Fig. 7A), comprising:
- accessing training data in form of recordings of real-time or near real-time data streams of measured values of a physiological parameter of one or more patients (Para, “[0007] In another aspect, a system for generating and providing simulated animal data by executing the methods herein is provided. The system including a computing device is operable to execute steps of receiving one or more sets of real animal data at least partially obtained from one or more sensors that receive, store, or send information related to one or more targeted individuals;..”
Para, “[0071] ... A biosensor can gather biological data (including readings and signals) such as physiological data, biometric data, chemical data, biomechanical data, genetic data, genomic data, location data or other biological data from one or more targeted individuals...”);
- iteratively training the machine learning model on the training data to obtain a machine learning model by (Para, “[0077] In a variation, one or more neural networks are utilized to generate simulated animal data. In general, a neural network generates simulated animal data after being trained with real animal data. Animal data (e.g., ECG signals, heart rate, biological fluid readings) is collected from one or more sensors from one or more target individuals typically as a time series of observations. Sequence prediction machine learning algorithms can be applied to predict possible animal data values based on collected data. ...... Utilizing a GAN, the generator generates one or more new data values, which may comprise one or more new data sets, while the discriminator evaluates the one or more new values based on one or more user-defined criteria to certify, validate, or authenticate the newly created values.”), in a training iteration:
- selectin a first time-series of values of the physiological parameter from the training data (Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets..” Note: here the system allows to include any number of seconds of data.);
- selecting a second time-series of values of the physiological parameter from the training data, wherein the second time series is part of a same data stream as the first time- series of values of the physiological parameter and is ahead in time of the first time series ((Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets..” Note: here the system allows selecting any subintervals. Also, note here claim is not explicit about first interval with first network, and second interval with second network, and combining etc... for example see claims 4-5 which can be explicitly addressed by Redell reference.); and
- training the machine learning model to predict the second time-series of values of the physiological parameter from the first time-series of values of the physiological parameter as input (Para, “[0140]...The computing device includes a network connection (e.g., internet) and browser application (e.g., executing browser software). The connection application, which executes within the browser, is configured (e.g., programmed) to establish one or more wireless communication links with each of the one or more sensors, receive one or more streams of data from the one or more sensors, and display, via the browser application, one or more readings derived from at least a portion of the streamed data.”
Para, “[0023] FIG. 7A provides a plot of artificial heart rate data generated from real animal data in-sample, which occurs when forecasting for an observation that includes at least a portion of the animal data sample.”).
Regarding claim 15, Khare teaches a transitory or non-transitory computer-readable medium comprising data representing a computer program, the computer program comprising instructions for causing a processor system to perform the method according to claim 13 (See, Claim 13 above; Para, “[0045] The processes, methods, or algorithms disclosed herein can be deliverable to/implemented by a computer, controller, or other computing device, which can include any existing programmable electronic control unit or dedicated electronic control unit. Similarly, the processes, methods, or algorithms can be stored as data and instructions executable by a computer, controller, or other computing device in many forms including, but not limited to, information permanently stored on non-writable storage media such as ROM devices and information alterably stored on writeable storage media such as floppy disks, magnetic tapes, CDs, RAM devices, other magnetic and optical media, and shared or dedicated cloud computing resources. The processes, methods, or algorithms can also be implemented in an executable software object. Alternatively, the processes, methods, or algorithms can be embodied in whole or in part using suitable hardware components, such as Application Specific Integrated Circuits (ASICs), Field-Programmable Gate Arrays (FPGAs), state machines, controllers or other hardware components or devices, or a combination of hardware, software, and firmware components.”).
Claim Rejections - 35 USC § 103
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 4-5 are rejected under 35 U.S.C. 103 as being unpatentable over Khare in view of Nickalus Redell (“Forecasting Combinations”, 04/19/2020)
Regarding claim 4, Khare teaches the prediction system according to claim 1.
Khare further teaches wherein the prediction system is configured to access at least two machine learning models (Para 0076 “The new one or more artificial data sets may be created by application of one or more artificial intelligence techniques that will analyze previously captured data sets that match some or all of the characteristics required by the acquirer. The one or more artificial intelligence techniques (e.g., one or more trained neural networks, machine learning models) can recognize patterns in real data sets, be trained by the collected data to understand animal (e.g., human) biology and related profiles, be further trained by collected data to understand the impact of one or more parameters or variables on animal biology and related profiles, and create artificial data that factors in the one or more parameters or variables chosen
by the acquirer in order to match or meet the minimum requirements of the acquirer...”), wherein a first machine learning model is configured to make shorter term predictions (Para 0090, “...For example, collecting animal data, situational context information related to the animal data, and trends based on the real-time or near real-time animal data enables comparison of real-time or near real-time data to historical data in the situational equivalent while enabling evaluation of both microtrends (e.g., within seconds or minutes) and macrotrends (e.g., full game)....”) and a second machine learning model is configured to make longer term predictions (Para 0090, “...For example, collecting animal data, situational context information related to the animal data, and trends based on the real-time or near real-time animal data enables comparison of real-time or near real-time data to historical data in the situational equivalent while enabling evaluation of both microtrends (e.g., within seconds or minutes) and macrotrends (e.g., full game)...”), wherein the processor subsystem is configured to combine outputs of the [first machine learning model and the second machine] learning model to obtain one time-series of classifications to be provided to the clinical decision support tool (Para 0109, “... In a refinement, the system can be operable to combine dissimilar data sets to create or re-create one or more new data sets. For example, a user may want 1 hour of Player A's heart rate data when the temperature is above 95 degrees for the entirety of a two-hour match for a specific tournament, where one or more features such as elevation or humidity may impact performance. While this data has never been collected in its entirety, different data sets that comprise at least a portion of the requested data and feature the one or more desired parameters/variables (e.g., one or more data sets from Player A featuring heart rate, one or more data sets from Player A featuring playing tennis in temperatures above 95 degrees, one or more data sets at the required tournament with requested features such as elevation) can be identified by the simulation system. With the simulation system operable to identify these requested parameters within the data sets and across data sets and be trained to understand the impact of the one or more parameters/variables on collected animal data and associated outcomes, the simulation system can run one or more simulations to create one or more new artificial data sets that fulfill the user's request (which may be, for example, a predictive indicator, computed asset, or artificial animal data) based on these dissimilar sets of data.”).
Khare teaches the claimed invention, however it doesn’t explicitly recites combination of first model and second mode.
Redell explicitly teaches first model for short term prediction, and second model for long term prediction, and combining the output of the first and second model (Pg. 3-4, Combination:
Greedy: Models with shorter direct forecast horizons produce near-term forecasts, and models with longer direct forecast horizons only produce forecasts at horizons above and beyond those from the short-term models.
If multiple model training functions produce models with the same direct forecast horizon, forecasts for the shared horizon(s) are combined with the function passed in combine_forecasts(..., agregate = function) (see example 3 below).”)
It would have been obvious for a person of ordinary skill in the art to incorporate forecast combination teaching of Redell into teachings of Khare at the time the application was filed in order to give precise control over the algorithms used to produce short- and long-term forecasts. (Pg. 4, “Gives precise control over the algorithms used to produce short- and long-term forecasts..”)
Regarding claim 5, Khare as modified by Redell teaches the prediction system according to claim 4.
Khare further teaches wherein the first machine learning model is configured to output a first time-series of predicted values which has a first time resolution selected between half a second and tens of seconds (Para, “[0131] For heart rate values, the system may increase the amount of data used by the pre-filter logic processing the raw data to include n number of seconds worth of AFE data. An increase in the amount of data collected and utilized by the system enables the system to create a more predictable pattern of HR generated values as the number of intervals that are used to identify the QRS complex is increased. This occurs because HR is an average of the HR values calculated over one second sub-intervals. The n number of seconds is a tunable parameter that may be pre-determined or dynamic. In a refinement, one or more artificial intelligence techniques may be utilized to predict the n number of seconds of AFE data required to generate one or more values that fall within a given range based on one or more previously collected data sets.”), and wherein the second machine learning model is configured to output a second time-series of predicted values which has a second time resolution selected between tens of seconds and minutes (Para 0109, “...For example, in the context of tennis, an acquirer may want 1 hour of Player A's heart rate data when the temperature is at or above 95 degrees Fahrenheit for the entirety of a two-hour match.”
Para, “[0083] .... animal data readings are timestamped and occur at a predetermined time period (e.g., approximately every second). Initially, the model is trained using N such observations (length of LSTM sequence), which can be a few (e.g., 20), a couple hundred, thousands, millions, and more.” Note: from above citations, it is clear that time period can be predefined, and it can be any interval.” Note: the reference teaches determining both microtrends (short term), and macrotrend (long term), and further the reference teaches using predetermined time period, thus covering any variation of time selection.)
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Khare in view of LODDENKEMPER et al. (US 20230141496 A1)
Regarding claim 10, Khare teaches the training system according to claim 9.
Khare doesn’t explicitly teaches wherein the processor subsystem is configured to select, in respective training iterations, non- overlapping instances of the first time series of values of the physiological parameter from the training data.
LODDENKEMPER teaches wherein the processor subsystem is configured to select, in respective training iterations, non- overlapping instances of the first time series of values of the physiological parameter from the training data (Para, “[0079] In some embodiments, the seizure forecasting model 120 includes a combination of hardware and/or software for predicting seizure risk at a given time based on the wearable sensor data 102, or a subset of the wearable sensor data 102 pertaining to a selected segment of time preceding the given time. In some embodiments, the seizure forecasting model 120 may predict a seizure risk level once every prediction period. In some embodiments, the prediction period may be, e.g., every second, every ten seconds, every fifteen seconds, every twenty seconds, every thirty seconds, every minute, or other suitable period. Thus, for each prediction period, the seizure forecasting model 120 may develop a seizure risk prediction for that prediction period based on the wearable sensor data 102 and other health-related data associated with the prediction period. In some embodiments, the prediction periods are continuous, non-overlapping segments of time including the wearable sensor data 102 during that time.”)
It would have been obvious for a person of ordinary skill in the art to incorporate segment selection teaching of LODDENKEMPER into teachings of Khare at the time the application was filed in order to avoid over sampling and increased computations. (Para 0104, “Where overlapping segments are employed, such an approach may result in increased computation by down-sampling the overlapping portions of segments more than once. However, accuracy and effectiveness may nevertheless be comparable”).
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Khare in view of Zang et al. (US 20200364596 A1)
Regarding claim 11, Khare teaches the training system according to claim 9.
Khare further teaches wherein the processor subsystem is configured to, in the training iteration, apply an outlier detection technique to the first time-series of values of the physiological parameter (Para, “0112] In another refinement, the simulated data may be assigned to one or more classifications. Classifications (e.g., including groups) can be created to simplify the search process for a data acquirer (e.g., as one or more searchable tags) and may be based on data collection processes, practices, quality, or associations, as well as targeted individual and simulated targeted individual characteristics. Classifications can be identifiers for data...... Another classification may be assigned to an artificial data set that is representative of ECG data from a specific sensor with specific settings and following a specific data collection methodology. In another example, a classification may be created for data sets representing targeted individuals that have previously experienced a stroke, or for simulated data sets representing simulated targeted individuals that are based upon, at least in part, real-world targeted individuals..... simulated data range classifications (providing a range for the data, e.g., bilirubin levels between 0.2-1.2 mg/dL..”
Para, “[0078] ... Thresholds involve generally accepted values or principles (e.g., it may be established that a male over 90 years old should contact their doctor if their heart rate reaches over 200 beats per minute, or that the age-based max heart rate for a 33-year old male is n beats per minute). FIG. 3A provides a graph of the heart rate beats per minute (BPM) captured from a professional athlete, while FIG. 3B provides the autocorrelation function for the data in FIG. 3A.”).
Khare doesn’t explicitly teaches and if the first time- series of values of the physiological parameter is classified as an outlier with respect to the training data, apply a higher weight to an output of the machine learning model in the training than when the first time-series of values of the physiological parameter is not classified as an outlier with respect to the training data.
Zang teaches and if the first time- series of values of the physiological parameter is classified as an outlier with respect to the training data, apply a higher weight to an output of the machine learning model in the training than when the first time-series of values of the physiological parameter is not classified as an outlier with respect to the training data (Para, “[0034] Referring to FIGS. 2A-2B, historical weightings/weights 226a may be applied to each of the predicted time series (e.g., predicted time series 222a-1 through 222a-N; 222a-A; 222a-B) to generate a set of historically weighted (HW) time series (e.g., HW time series 1 230a-1, HW times series 2 230a-2, HW times series 3 230a-3, . . . HW times series N 230a-N, HW time series A 230a-A, HW times series B 230a-B). The historical weightings 226a may be selected in accordance with, and may be indicative of, the relevance of each of the respective time series in generating one or more outputs (e.g., one or more outputs included in the historical output time series 218a). Where applicable, adjustments may be made to the historical weights 226a to account for the introduction of one or more prospective time series (e.g., prospective time series A 210a-A, prospective time series B 210a-B, etc.). Collectively, the contribution made by each of the HW time series may sum to 100%.” Note: in instance case classification of an outlier is not an outlier data, rather it is event detection such as data being outside the expected range (data indicative of relevance).
Para “[0035] Predictive weightings/weights 234a may be applied to the set of HW time series to generate a set of predictive weighted (PW) time series (e.g., PW time series 1 238a-1, PW time series 2 238a-2, PW time series 3 238a-3, . . . PW time series N 238a-N, PW time series A 238a-A, PW time series B 238a-B). The predictive weightings 234a may be selected in accordance with, and may be indicative of, the perceived relevance of each of the respective time series in generating one or more outputs on the basis of events/conditions that have not yet occurred but may occur in the future with some deterministic probability. Such futuristic events/conditions would not necessarily/normally be reflected in the historical time series; the predictive weightings 234a may be used to account for known/probabilistic future events/conditions.”)
It would have been obvious for a person of ordinary skill in the art to incorporate weight assignment teaching of Zang into teachings of Khare at the time the application was filed in order to account for the detected events/conditions. (Para 0035, “Such futuristic events/conditions would not necessarily/normally be reflected in the historical time series; the predictive weightings 234a may be used to account for known/probabilistic future events/conditions.”)
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20130080374 A1, Abstract: “Systems and methods are provided for generating a forecasting model based on a set of measured values. Consistent with certain embodiments, the forecasting model may include a seasonal function and a trend function. Further, consistent with other embodiments, the computer-implemented systems and methods may include computing the forecasting model by minimizing an error function representing the error between the forecasting model and the measured values.”
Any inquiry concerning this communication or earlier communications from the examiner should be directed to HUMA WASEEM whose telephone number is (571)272-1316. The examiner can normally be reached Monday-Friday(9:00 am - 5 pm) EST.
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 B. Dunham can be reached on (571) 272-8109. 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.
/HUMA WASEEM/Examiner, Art Unit 3686
/JASON B DUNHAM/Supervisory Patent Examiner, Art Unit 3686
Prosecution Insights