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 .
Response to Amendment
The amendment filed on 08/12/2026 has been entered. Claims 1 and 5-6 have been amended. Claims 1 and 5-6 remain pending.
The previously raised rejections under 35 U.S.C. 112(f), regarding the features of “a data receiving unit” and “a blood pressure state prediction unit” in Claims 1 and 5, are withdrawn.
The previously raised rejections under 35 U.S.C. 112(a), regarding the features of “a data receiving unit” and “a blood pressure state prediction unit” in Claims 1 and 5, and the feature of determining blood pressures in Claim 6, are withdrawn.
The previously raised rejections under 35 U.S.C. 112(b), regarding the feature of “a data receiving unit” in Claims 1 and 5, are withdrawn.
Response to Arguments
35 U.S.C. 112(a) and 112(f)
Regarding “derived variable extraction unit” on Pages 9-10 of Remarks, Applicant argues that the recited derived variable extraction unit, specifically “one derived variable is generated for each time point across the prediction section” (Page 10), recites a definite algorithm and constitutes sufficient structure (Page 9). Examiner respectfully disagrees. Both the claimed limitation and the disclosures at paragraph [0031] and paragraph [0046] indicate that the “derived variable” relates to or is based on the “time elapsed”, which is not sufficient for one person of ordinary skill in the art to either understand what such derived variable is or how to obtain such derived variable. Therefore, the issued 112(f) and 112(a) for the limitation are not withdrawn.
35 U.S.C. 103 rejections
On Page 13 of Remarks at the point of A), Applicant argues that the cited references do not teach the amended third learning model, and specifically, Aguirre does not teach a Cross-Attention layer that receives and fuses the outputs of two different learning models, but operates within a single model but not across the outputs of two separate learning models as amended claim 1 requires. Examiner respectfully disagrees. First, reference Aguirre is introduced for the modification of including a Cross-Attention layer in a learning model disclosed by the other references. It is the disclosed learning model, but not the Cross-Attention layer (see a new 112(a) rejection in later section), that receives the two inputs. Second, in reference Aguirre, similar to the claim, two inputs are fused into one model, i.e. Xl (PPG signals) and XDI (age and gender demographic information vector) as shown in Fig. 6 of Aguirre.
On Pages 13-14 of Remarks at the point of B), Applicant argues that Aguirre provides no motivation to fuse a non-invasive-blood-pressure-derived feature stream, and specifically, Aguirre’s method only needs raw PPG from the finger. Examiner respectfully disagrees. Aguirre does not use cuff-based blood pressure measurement as Applicant pointed out, but proposes a similar approach as the claim, i.e. measure PPG signals or waveform, and combine it with an additional parameter (blood pressure measurement in the claim, and demographic information in Aguirre) using a model with cross-attention layer, with the same goal of estimating arterial blood pressure with high accuracy. Because of such high similarity between Aguirre and the application, one of ordinary skill in the art would modify with Aguirre.
On Page 14 of Remarks at the point of C), Applicant argues that modification of reference Wang with reference Aguirre would change the principle of operation of Wang, and specifically, Wang uses a linear-regression estimator but Aguirre uses a deep learning model with a Cross-Attention layer. Examiner respectfully disagrees. Linear regression method used by Wang and deep learning model used by Aguirre are the same in that both the methods require a learning or training procedure, in which training data are fed into an optimization to determine optimal values for model parameters (neural network parameters for deep learning model, and slope and y-axis intercept for linear regression model). Deep learning or neural network model is advantageous over a linear regression model, mostly due to the nonlinear property and much higher number of model parameters (thus model complexity) of the former, in performing complex tasks such as classification and prediction with high accuracy. Modification of Wang and other references with Aguirre would not change the principle of Wang, but enhance the performance of the method.
On Pages 14-15 of Remarks at the point of D), Applicant argues that the rejection relies on impermissible hindsight, and specifically, the references Wang, Addison and Aguirre are three architecturally distinct systems, and none of them discloses the specific pipeline of the application. Examiner respectfully disagrees. Both Wang and Addison use a similar pipeline as the application, and Aguirre’s so-called sequence-to-sequence approach combines both PPG signals and demographic information, resembling the claimed pipeline.
On Page 15 of Remarks at the point of E), Applicant argues that the claimed derived variables and second feature value are not taught, and specifically, one derived variable of the plurality of derived variables are generated for each time point of the plurality of time points in the prediction section, but Addison predicts blood pressure point by point. In current office action, rejection of the features of “a derived variable extraction unit configured to …” and “a non-invasive blood pressure unit configured to …” in amended Claim 1 relies on a new ground based on references Wang, Addison and Aguirre: briefly, Wang discloses generating a plurality of derived variables and inputting, to a learning model, the derived variables and the most recently measured non-invasive blood pressure measurement value; Addison discloses, for one time point to be predicted or calibrated, one derived variable, the derived value being generated based on a time elapsed from the most recently measured non-invasive blood pressure measurement value to the one time point, and inputting, to a learning model, the one derived variable and an equal number of copies of the most recently measured non-invasive blood pressure measurement value; Aguirre discloses generating a plurality of derived variables (predicted outputs at an earlier time-step as shown in Fig. 6) and an equal number of copies of calibration factor (demographic information vector as shown in Fig. 6).
On Page 15 of Remarks at the point of F), Applicant argues that the claimed architecture provides a demonstrated technical improvement, specifically by considering non-invasive blood pressure and PPG waveform to produce a continuous waveform. As discussed above, the claimed architecture is similar to that of Wang, Addison and Aguirre.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitations are:
“a photoplethysmography analysis unit configured to …” in Claim 1 and in Claim 5.
A review of the Specification discloses that the corresponding structure for “a photoplethysmography analysis unit” is formed of “a deep learning-based algorithm such as Resnet algorithm, U-net algorithm, Attention algorithm, and Self-Attention algorithm” (Para 0042).
“a derived variable extraction unit configured to generate …” in Claim 1 and in Claim 5.
A review of the Specification discloses indicates that sufficient structural information of the “derived variable extraction unit” is not provided in the Specification (see Para 0046; also refer to the section of 35 U.S.C 112(a)).
“a non-invasive blood pressure analysis unit configured to input …” in Claim 1 and in Claim 5.
A review of the Specification discloses that the corresponding structure for “a non-invasive blood pressure analysis unit” is formed of “one or more of several known algorithms such as the LSTM algorithm” (Para 0050).
Because these claim limitations are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, they are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof.
If applicant does not intend to have these limitations interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitations to avoid them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitations recite sufficient structure to perform the claimed function so as to avoid them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
The following is a quotation of the first paragraph of 35 U.S.C. 112(a):
(a) IN GENERAL.—The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor or joint inventor of carrying out the invention.
The following is a quotation of the first paragraph of pre-AIA 35 U.S.C. 112:
The specification shall contain a written description of the invention, and of the manner and process of making and using it, in such full, clear, concise, and exact terms as to enable any person skilled in the art to which it pertains, or with which it is most nearly connected, to make and use the same, and shall set forth the best mode contemplated by the inventor of carrying out his invention.
Claims 1 and 5-6 are rejected under 35 U.S.C. 112(a) or 35 U.S.C. 112 (pre-AIA ), first paragraph, as failing to comply with the written description requirement. The claim contains subject matter which was not described in the specification in such a way as to reasonably convey to one skilled in the relevant art that the inventor or a joint inventor, or for applications subject to pre-AIA 35 U.S.C. 112, the inventor(s), at the time the application was filed, had possession of the claimed invention.
Claim 1, Lines 21-29, and Claim 5, Lines 16-24, recite “a derived variable extraction unit configured to generate …”, and Claim 6, Lines 14-21, recites “generating … a plurality of derived variables …”. Specification, Para 0031 and Para 0046, describes how derived variables are generated, but it merely species that such derived variables relate to the time elapsed from the time point of the most recent non-invasive blood pressure measurement to the time point to be predicted, which is not adequate or clear enough for one person of ordinary skill in the art to perform the generating process.
Claim 1, Lines 37-39, Claim 5, Lines 32-34, and Claim 6, Lines 27-29, recite the Cross-Attention layer receiving “the primary arterial blood pressure waveform … and the second feature value …”. Specification, Para 0053, specifies that the third learning model includes the Cross-Attention layer, and the third learning model receives the primary arterial blood pressure waveform and the second feature value, which is NOT equivalent to that the Cross-Attention layer directly receives the waveform and the value.
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 and 5-6 are rejected under 35 U.S.C. 103 as being unpatentable over Wang et al (US 20230293117 A1; hereafter Wang), in view of Addison et al (US 20210193311 A1; hereafter Addison), Aguirre et al (Sensors 2021, 21, 2167; hereafter Aguirre), and Barak et al (US 20200253564 A1; hereafter Barak).
With regard to Claim 1, Wang discloses a deep learning-based (Wang, Para 0087; “… after the deep learning processing of Approximation Network and Refinement Network”) continuous arterial blood pressure monitoring system (Wang, Para 0008; “The method or system is comfortable, continuous, and suitable for all-day use, and the accuracy of estimation of BPs and/or CBPs are sufficiently improved.”), comprising:
a photoplethysmography sensor configured to measure a photoplethysmography measurement value of a patient (Wang, Para 0035; “an upper-arm wearable apparatus including a PPG sensor and sensing modeling-used PPG waveform signals …”);
a non-invasive blood pressure measurement device including a cuff (Wang, Abstract; “A system for estimating BPs … comprises … a cuff-based BP measuring apparatus …”; Para 0065; “The cuff-based BP measuring apparatus 12 has been calibrated so as to provide … an accurate blood pressure.”) configured to obtain a most recently measured non-invasive blood pressure measurement value of the patient by applying a pressure to a blood vessel while the cuff is worn on a body part of the patient, the most recently measured non-invasive blood pressure measurement value including a systolic blood pressure value and a diastolic blood pressure value (Wang, Fig. 1 shows that “a cuff 121 included in a cuff-based BP measuring apparatus 12” (Para 0065) is worn on an upper arm of a subject, and generates “Real SBP/DBP” (i.e. systolic blood pressure and diastolic blood pressure). The disclosed “Real SBP/DBP” is measured directly from the subject and fed into the system, so corresponds to “most recently measured” values);
a data receiving unit (the PPG signal receiver and analyzer 13) configured to receive the photoplethysmography measurement value and the most recently measured non-invasive blood pressure measurement value (Wang, Fig. 1 shows that the PPG to BP subsystem (101) includes the PPG signal Receiver and analyzer (13) and also receives “Real SBP/DBP”);
a photoplethysmography analysis unit (PPG to PVR transformer 15) configured to normalize the photoplethysmography measurement value (Wang, Para 0081; “… the normalization value (Maxima and minima) for upper-arm PPG and PVR waveform signals were calculated from the upper-arm PPG database in this embodiment.”), to input a normalized photoplethysmography measurement value to a first learning model (Wang, Para 0067; “a PPG to PVR transformer 15 converts the PPG waveform signals to refined PVR waveforms using deep machine learning”), and to output a primary arterial blood pressure waveform (refined PVR waveforms), the first learning model including an encoding region and a decoding region, the encoding region including a plurality of convolutional layers configured to extract a plurality of features in a plurality of encoding steps, the first learning model being configured to store the plurality of features extracted in the plurality of encoding steps, and the decoding region being configured to output the primary arterial blood pressure waveform based on the stored plurality of features (Wang, Para 0084; “U-Net comprises a network constructed using only convolutional layers … The network structure is constructed using a symmetric pair of Encoder Network and Decoder Network.” The disclosed “U-Net” inherently comprises the claimed structures of the first learning model);
a derived variable extraction unit (the PPG signal receiver and analyzer 13) configured to generate a plurality of derived variables (Wang, Para 0071; “… the characteristic parameters including waveform parameters and time-related parameters extracted by the PPG signal receiver and analyzer 13. The waveform parameters include …”);
a non-invasive blood pressure analysis unit (PPG to BP estimator and calibrator 14) configured to input, to a second learning model (Wang, Para 0066; “PPG to BP estimator and calibrator 14 … it uses the age grouping method and further calibrates preliminary values by machine learning (ML) algorithms or other algorithms (e.g. regression algorithms, artificial neural networks (ANN), fuzzy logic, and support vector machine) …”), the plurality of derived variables and the most recently measured non-invasive blood pressure measurement value (“Real SBP/DBP” in Fig. 1), and to output a second feature value (calibrated-estimated SBP/DBP);
a blood pressure waveform prediction unit (CBP estimator) configured to input the primary arterial blood pressure waveform and the second feature value to a third learning model (Wang, Para 0088; “… the CBP estimator 16 further uses linear regression equations to estimate CBPs in clinic use, and substitutes the calibrated-estimated BPs and waveform parameters of the user’s refined PVR waveform into the linear regression equation to have estimated CBPs.” The disclosed “linear regression equations” is a learning model, with their regression coefficients learnt from training data), and to output a continuous arterial blood pressure waveform (Wang, Para 0006; “The translation of PPG signals into arterial blood pressure (ABP) is described in another non-patent literature entitled “PPG2ABP: Translating Photoplethysmogram (PPG) Signals to Arterial Blood Pressure (ABP) Waveforms using Fully Convolutional Neural Networks” (written by Nabil Ibtehaz, M. Sohel Rahman; Electrical Engineering and Systems Science: Signal Processing, published on May 5, 2020), the teachings of which are incorporated herein by reference in their entirety.”. The reference of Ibtehaz et al discloses “… a method to estimate the continuous arterial blood pressure (ABP) waveform through a non-invasive approach using Photoplethysmogram (PPG) signals” (Abstract)); and
a blood pressure state prediction unit configured to determine a systolic blood pressure and a diastolic blood pressure from the continuous arterial blood pressure waveform (Wang, Para 0006; “PPG2ABP … written by Nabil Ibtehaz … the teachings of which are incorporated herein by reference in their entirety”. Ibtehaz et al discloses in Page 6 that SBP and DBP are calculated as maximum and minimum of continuous arterial blood pressure (ABP) waveform, respectively), and to classify a blood pressure state of the patient as one blood pressure state selected from low blood pressure, normal, prehypertension, and high blood pressure based on the determined systolic blood pressure and the determined diastolic blood pressure (Wang, Para 0006; “Jia-Wei Chen et al … the teachings of which are incorporated herein by reference in their entirety”. Chen et al discloses in Page 8 “… actual BP and PPG data and information from participants who were diagnosed with normotension (BP < 130/90 mm Hg), prehypertension, and hypertension …”, and in Page 14 “The abnormal BP levels were defined as hypertension (SBP/DBP ≥ 130/80 mmHg) and hypotension (SBP/DBP < 90/60 mmHg) …” ).
Wang does not clearly and explicitly disclose:
receiving data through a serial port or a LAN port of a commercial patient monitoring device;
generating, for a prediction section to be predicted and having a length corresponding to a length of the photoplethysmography measurement value for the precision section, a plurality of derived variables;
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points;
each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to a corresponding time point of the plurality of time points within the prediction section;
inputting, to a second learning model, the plurality of derived variables and the equal number of copies of the most recently measured non-invasive blood pressure measurement value; and
a learning model including a Cross-Attention layer and a forward neural network.
Addison in the same field of endeavor discloses:
for one time point, each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to the one time point (Addison, Para 0083; “Features 604 may be a value that corresponds [to] the time elapsed since the most recent calibration point.” The disclosed elapsed time corresponds to the claimed derived variable); and
inputting, to a learning model, the one derived variable (feature 604, or elapsed time) and one copy of the most recently measured non-invasive blood pressure measurement value (feature 606, or BP at Calibration Point) (Addison, Fig. 6B shows the features as input for determining blood pressure for a particular time t, which include feature 604 (corresponding to the claimed “derived variable” as discussed above) and feature 606 (“BP at Calibration Point”));
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, as suggested by Addison, in order for one time point of prediction to input a copy of blood pressure measured at calibration point and derive a variable based on elapsed time. One of ordinary skill in the art would have been motivated to make the modification of improving prediction accuracy for the estimated blood pressure (Addison, Para 0086; “… by using the calibration data received at the most recent calibration point as additional inputs for training CNIBP model 124, CNIBP model 124 may improve its accuracy in determining the blood pressure of patients.”).
Wang and Addison do not explicitly and clearly disclose:
receiving data through a serial port or a LAN port of a commercial patient monitoring device;
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable and an equal number of copies of calibration factor; and
a learning model including a Cross-Attention layer and a forward neural network.
Aguirre in the same field of endeavor discloses:
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable (predicted outputs at an earlier time-step as shown in Fig. 6, i.e. [y0, y1, …, yT-1]) and an equal number of copies of calibration factor (demographic information vector XDI as shown in Fig. 6) (Aguirre, Page 8, Para 1; “Age and gender demographic information vector (XDI) is concatenated at each time-step with yi to conform the decoder inputs (yi&DI).” As shown in Fig. 6, a number T of yi values and equal number of copies of XDI are input into the model); and
a learning model including a Cross-Attention layer (Aguirre, Abstract; “this methodology is capable of transforming PPG into an ABP pulse …”; Page 6, Para 2; “The proposed deep learning architecture is inspired by seq2seq encoder-decoder [36] models with attention mechanism.” Fig. 6 of Aguirre shows that the block “Attention” combines the outputs of both the encoder (from input Xl) and the decoder (from output of each time step concatenated with demographic information vector), so is a typical cross-attention operation) and a forward neural network (Aguirre, Fig. 6 show that the model also includes multiperceptron layers (MPLv and MPLc), which are feed-forward neural network layers).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang and Addison, as suggested by Aguirre, in order to prepare a plurality of derived variables paired with calibration factor for one implementation of model prediction and comprise a cross-attention component and a feed-forward component in the network. One of ordinary skill in the art would have been motivated to make the modification of preparing a plurality of derived variables for the benefit of improving prediction efficiency by producing a continuous waveform of blood pressure in a single prediction, and make the modification of using a model with a cross-attention layer for the benefit of achieving a cross-attention mechanism that is known to enable higher efficiency and accuracy as compared to regular neural networks.
Wang, Addison and Aguirre do not clearly and explicitly disclose receiving data through a serial port or a LAN port of a commercial patient monitoring device.
Barak in the same field of endeavor discloses receiving data through a serial port or a LAN port of a commercial patient monitoring device (Barak, Para 0023; “Typically, PCMs (Patient Care Monitors) obtain signals from PPG sensors …”; Para 0256; “Cable 107 connects tracking transducer 102 to PCM 103.” Para 0279; “Interface circuit 304 may format data for transmission serial RS232 connection, USB, Ethernet (LAN), or any other data format in which PCM 103 is capable of receiving BPV and SVV data.”). It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, Addison and Aguirre, as suggested by Barak, in order to receive data from clinical monitors using serial port or LAN port. One of ordinary skill in the art would have been motivated to make the modification for the benefit of reliably receiving patient data measured with clinical-standard device.
With regard to Claim 5, Wang discloses a deep learning-based (Wang, Para 0087; “… after the deep learning processing of Approximation Network and Refinement Network”) continuous arterial blood pressure monitoring system (Wang, Para 0008; “The method or system is comfortable, continuous, and suitable for all-day use, and the accuracy of estimation of BPs and/or CBPs are sufficiently improved.”) comprising:
a data receiving unit (the PPG signal receiver and analyzer 13) configured to receive a photoplethysmography measurement value of a patient (Wang, Para 0035; “an upper-arm wearable apparatus including a PPG sensor and sensing modeling-used PPG waveform signals …”) and a most recently measured non-invasive blood pressure measurement value of the patient (Wang, Abstract; “A system for estimating BPs … comprises … a cuff-based BP measuring apparatus …”; Para 0065; “The cuff-based BP measuring apparatus 12 has been calibrated so as to provide … an accurate blood pressure.”), the most recently measured non-invasive blood pressure measurement value including a systolic blood pressure value and a diastolic blood pressure value (Wang, Fig. 1 shows that “a cuff 121 included in a cuff-based BP measuring apparatus 12” (Para 0065) is worn on an upper arm of a subject, and generates “Real SBP/DBP” (i.e. systolic blood pressure and diastolic blood pressure). The disclosed “Real SBP/DBP” is measured directly from the subject and fed into the system, so corresponds to “most recently measured” values);
a photoplethysmography analysis unit (PPG to PVR transformer 15) configured to normalize the photoplethysmography measurement value (Wang, Para 0081; “… the normalization value (Maxima and minima) for upper-arm PPG and PVR waveform signals were calculated from the upper-arm PPG database in this embodiment.”), to input a normalized photoplethysmography measurement value to a first learning model (Wang, Para 0067; “a PPG to PVR transformer 15 converts the PPG waveform signals to refined PVR waveforms using deep machine learning”), and to output a primary arterial blood pressure waveform (refined PVR waveforms), the first learning model including an encoding region and a decoding region, the encoding region including a plurality of convolutional layers configured to extract a plurality of features in a plurality of encoding steps, the first learning model being configured to store the plurality of features extracted in the plurality of encoding steps, and the decoding region being configured to output the primary arterial blood pressure waveform based on the stored plurality of features (Wang, Para 0084; “U-Net comprises a network constructed using only convolutional layers … The network structure is constructed using a symmetric pair of Encoder Network and Decoder Network.” The disclosed “U-Net” inherently comprises the claimed structures of the first learning model);
a derived variable extraction unit (the PPG signal receiver and analyzer 13) configured to generate a plurality of derived variables (Wang, Para 0071; “… the characteristic parameters including waveform parameters and time-related parameters extracted by the PPG signal receiver and analyzer 13. The waveform parameters include …”);
a non-invasive blood pressure analysis unit (PPG to BP estimator and calibrator 14) configured to input, to a second learning model (Wang, Para 0066; “PPG to BP estimator and calibrator 14 … it uses the age grouping method and further calibrates preliminary values by machine learning (ML) algorithms or other algorithms (e.g. regression algorithms, artificial neural networks (ANN), fuzzy logic, and support vector machine) …”), the plurality of derived variables and the most recently measured non-invasive blood pressure measurement value (“Real SBP/DBP” in Fig. 1), and to output a second feature value (calibrated-estimated SBP/DBP);
a blood pressure waveform prediction unit (CBP estimator) configured to input the primary arterial blood pressure waveform and the second feature value to a third learning model (Wang, Para 0088; “… the CBP estimator 16 further uses linear regression equations to estimate CBPs in clinic use, and substitutes the calibrated-estimated BPs and waveform parameters of the user’s refined PVR waveform into the linear regression equation to have estimated CBPs.” The disclosed “linear regression equations” is a learning model, with their regression coefficients learnt from training data), and to output a continuous arterial blood pressure waveform (Wang, Para 0006; “The translation of PPG signals into arterial blood pressure (ABP) is described in another non-patent literature entitled “PPG2ABP: Translating Photoplethysmogram (PPG) Signals to Arterial Blood Pressure (ABP) Waveforms using Fully Convolutional Neural Networks” (written by Nabil Ibtehaz, M. Sohel Rahman; Electrical Engineering and Systems Science: Signal Processing, published on May 5, 2020), the teachings of which are incorporated herein by reference in their entirety.”. The reference of Ibtehaz et al discloses “… a method to estimate the continuous arterial blood pressure (ABP) waveform through a non-invasive approach using Photoplethysmogram (PPG) signals” (Abstract));
a blood pressure state prediction unit configured to determine a systolic blood pressure and a diastolic blood pressure from the continuous arterial blood pressure waveform (Wang, Para 0006; “PPG2ABP … written by Nabil Ibtehaz … the teachings of which are incorporated herein by reference in their entirety”. Ibtehaz et al discloses in Page 6 that SBP and DBP are calculated as maximum and minimum of continuous arterial blood pressure (ABP) waveform, respectively), and to classify a blood pressure state of the patient as one blood pressure state selected from low blood pressure, normal, prehypertension, and high blood pressure based on the determined systolic blood pressure and the determined diastolic blood pressure (Wang, Para 0006; “Jia-Wei Chen et al … the teachings of which are incorporated herein by reference in their entirety”. Chen et al discloses in Page 8 “… actual BP and PPG data and information from participants who were diagnosed with normotension (BP < 130/90 mm Hg), prehypertension, and hypertension …”, and in Page 14 “The abnormal BP levels were defined as hypertension (SBP/DBP ≥ 130/80 mmHg) and hypotension (SBP/DBP < 90/60 mmHg) …”); and
transmitting the determined results to at least one of a central monitoring device and an electronic medical record system (Wang, Para 0006; “Jia-Wei Chen et al … the teachings of which are incorporated herein by reference in their entirety”. Chen et al discloses in Page 5 “… the measured PPG-based BP and heart rate (HR), as well as oxygen saturation, could be transmitted wirelessly to a mobile phone for display and recording of multimodal physiological data in the mobile application (APP)”).
Wang does not clearly and explicitly disclose:
receiving data through a serial port or a LAN port of a commercial patient monitoring device,
generating, for a prediction section to be predicted and having a length corresponding to a length of the photoplethysmography measurement value for the precision section, a plurality of derived variables;
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points;
each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to a corresponding time point of the plurality of time points within the prediction section;
inputting, to a second learning model, the plurality of derived variables and the equal number of copies of the most recently measured non-invasive blood pressure measurement value,
a learning model including a Cross-Attention layer and a forward neural network, and
a data transmitting unit configured to convert the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and to transmit the voltage to a blood pressure measurement module of the commercial patient monitoring device, or to transmit the continuous arterial blood pressure waveform in digital form.
Addison in the same field of endeavor discloses:
for one time point, each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to the one time point (Addison, Para 0083; “Features 604 may be a value that corresponds [to] the time elapsed since the most recent calibration point.” The disclosed elapsed time corresponds to the claimed derived variable); and
inputting, to a learning model, the one derived variable (feature 604, or elapsed time) and one copy of the most recently measured non-invasive blood pressure measurement value (feature 606, or BP at Calibration Point) (Addison, Fig. 6B shows the features as input for determining blood pressure for a particular time t, which include feature 604 (corresponding to the claimed “derived variable” as discussed above) and feature 606 (“BP at Calibration Point”));
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, as suggested by Addison, in order for one time point of prediction to input a copy of blood pressure measured at calibration point and derive a variable based on elapsed time. One of ordinary skill in the art would have been motivated to make the modification of improving prediction accuracy for the estimated blood pressure (Addison, Para 0086; “… by using the calibration data received at the most recent calibration point as additional inputs for training CNIBP model 124, CNIBP model 124 may improve its accuracy in determining the blood pressure of patients.”).
Wang and Addison do not explicitly and clearly disclose:
receiving data through a serial port or a LAN port of a commercial patient monitoring device;
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable and an equal number of copies of calibration factor;
a learning model including a Cross-Attention layer and a forward neural network; and
a data transmitting unit configured to convert the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and to transmit the voltage to a blood pressure measurement module of the commercial patient monitoring device, or to transmit the continuous arterial blood pressure waveform in digital form.
Aguirre in the same field of endeavor discloses:
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable (predicted outputs at an earlier time-step as shown in Fig. 6, i.e. [y0, y1, …, yT-1]) and an equal number of copies of calibration factor (demographic information vector XDI as shown in Fig. 6) (Aguirre, Page 8, Para 1; “Age and gender demographic information vector (XDI) is concatenated at each time-step with yi to conform the decoder inputs (yi&DI).” As shown in Fig. 6, a number T of yi values and equal number of copies of XDI are input into the model); and
a learning model including a Cross-Attention layer (Aguirre, Abstract; “this methodology is capable of transforming PPG into an ABP pulse …”; Page 6, Para 2; “The proposed deep learning architecture is inspired by seq2seq encoder-decoder [36] models with attention mechanism.” Fig. 6 of Aguirre shows that the block “Attention” combines the outputs of both the encoder (from input Xl) and the decoder (from output of each time step concatenated with demographic information vector), so is a typical cross-attention operation) and a forward neural network (Aguirre, Fig. 6 show that the model also includes multiperceptron layers (MPLv and MPLc), which are feed-forward neural network layers).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang and Addison, as suggested by Aguirre, in order to prepare a plurality of derived variables paired with calibration factor for one implementation of model prediction and comprise a cross-attention component and a feed-forward component in the network. One of ordinary skill in the art would have been motivated to make the modification of preparing a plurality of derived variables for the benefit of improving prediction efficiency by producing a continuous waveform of blood pressure in a single prediction, and make the modification of using a model with a cross-attention layer for the benefit of achieving a cross-attention mechanism that is known to enable higher efficiency and accuracy as compared to regular neural networks.
Wang, Addison and Aguirre do not explicitly and clearly disclose:
receiving data through a serial port or a LAN port of a commercial patient monitoring device; and
a data transmitting unit configured to convert the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and to transmit the voltage to a blood pressure measurement module of the commercial patient monitoring device, or to transmit the continuous arterial blood pressure waveform in digital form.
Barak in the same field of endeavor discloses:
receiving data through a serial port or a LAN port of a commercial patient monitoring device (Barak, Para 0023; “Typically, PCMs (Patient Care Monitors) obtain signals from PPG sensors …”; Para 0256; “Cable 107 connects tracking transducer 102 to PCM 103.” Para 0279; “Interface circuit 304 may format data for transmission serial RS232 connection, USB, Ethernet (LAN), or any other data format in which PCM 103 is capable of receiving BPV and SVV data.”); and
a data transmitting unit (the combination of interface circuit 604 and D/A converter 505 in Fig. 5) configured to convert the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and to transmit the voltage to a blood pressure measurement module of the commercial patient monitoring device (Barak; Para 0265; “Wires 204 provide arterial pressure values to PCM 103, for example to be displayed.” Fig. 5 shows that the values first go through block 505 (D/A converter) before being transmitted by wires 204), or to transmit the continuous arterial blood pressure waveform in digital form (Barak; Para 0265; “D/A converter 505 are optional”; when such converter is not used, the transmitted data is in digital form).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, Addison and Aguirre, as suggested by Barak, in order to transmit data among acquisition apparatus, processing device and patient monitor using standard wired methods. One of ordinary skill in the art would have been motivated to make the modification for the benefit of transferring data and results in a reliable way and enabling efficient monitoring of patient’s vital information.
With regard to Claim 6, Wang discloses a continuous arterial blood pressure monitoring method (Wang, Para 0008; “The method or system is comfortable, continuous, and suitable for all-day use, and the accuracy of estimation of BPs and/or CBPs are sufficiently improved.”), comprising:
receiving a photoplethysmography measurement value of a patient from a photoplethysmography sensor (Wang, Para 0035; “an upper-arm wearable apparatus including a PPG sensor and sensing modeling-used PPG waveform signals …”) and receiving a most recently measured non-invasive blood pressure measurement value of the patient from a non-invasive blood pressure measurement device including a cuff (Wang, Abstract; “A system for estimating BPs … comprises … a cuff-based BP measuring apparatus …”; Para 0065; “The cuff-based BP measuring apparatus 12 has been calibrated so as to provide … an accurate blood pressure.”), the most recently measured non-invasive blood pressure measurement value including a systolic blood pressure value and a diastolic blood pressure value (Wang, Fig. 1 shows that “a cuff 121 included in a cuff-based BP measuring apparatus 12” (Para 0065) is worn on an upper arm of a subject, and generates “Real SBP/DBP” (i.e. systolic blood pressure and diastolic blood pressure). The disclosed “Real SBP/DBP” is measured directly from the subject and fed into the system, so corresponds to “most recently measured” values);
normalizing the photoplethysmography measurement value (Wang, Para 0081; “… the normalization value (Maxima and minima) for upper-arm PPG and PVR waveform signals were calculated from the upper-arm PPG database in this embodiment.”) and inputting a normalized photoplethysmography measurement value to a first learning model (Wang, Para 0067; “a PPG to PVR transformer 15 converts the PPG waveform signals to refined PVR waveforms using deep machine learning”) including an encoding region and a decoding region, extracting a plurality of features through a plurality of convolutional layers in a plurality of encoding steps, storing the plurality of features extracted in the plurality of encoding steps (Wang, Para 0084; “U-Net comprises a network constructed using only convolutional layers … The network structure is constructed using a symmetric pair of Encoder Network and Decoder Network.” The disclosed “U-Net” inherently comprises the claimed structures of the first learning model), and outputting a primary arterial blood pressure waveform in a decoding step based on the stored plurality of features (refined PVR waveforms);
generating a plurality of derived variables (Wang, Para 0071; “… the characteristic parameters including waveform parameters and time-related parameters extracted by the PPG signal receiver and analyzer 13. The waveform parameters include …”);
inputting, to a second learning model (Wang, Para 0066; “PPG to BP estimator and calibrator 14 … it uses the age grouping method and further calibrates preliminary values by machine learning (ML) algorithms or other algorithms (e.g. regression algorithms, artificial neural networks (ANN), fuzzy logic, and support vector machine) …”), the plurality of derived variables and non-invasive blood pressure measurement values (“Real SBP/DBP” in Fig. 1), and extracting a second feature value (calibrated-estimated SBP/DBP);
inputting the primary arterial blood pressure waveform and the second feature value to a third learning model (Wang, Para 0088; “… the CBP estimator 16 further uses linear regression equations to estimate CBPs in clinic use, and substitutes the calibrated-estimated BPs and waveform parameters of the user’s refined PVR waveform into the linear regression equation to have estimated CBPs.” The disclosed “linear regression equations” is a learning model, with their regression coefficients learnt from training data), and outputting a continuous arterial blood pressure waveform (Wang, Para 0006; “The translation of PPG signals into arterial blood pressure (ABP) is described in another non-patent literature entitled “PPG2ABP: Translating Photoplethysmogram (PPG) Signals to Arterial Blood Pressure (ABP) Waveforms using Fully Convolutional Neural Networks” (written by Nabil Ibtehaz, M. Sohel Rahman; Electrical Engineering and Systems Science: Signal Processing, published on May 5, 2020), the teachings of which are incorporated herein by reference in their entirety.”. The reference of Ibtehaz et al discloses “… a method to estimate the continuous arterial blood pressure (ABP) waveform through a non-invasive approach using Photoplethysmogram (PPG) signals” (Abstract));
determining a systolic blood pressure and a diastolic blood pressure from the continuous arterial blood pressure waveform (Wang, Para 0006; “PPG2ABP … written by Nabil Ibtehaz … the teachings of which are incorporated herein by reference in their entirety”. Ibtehaz et al discloses in Page 6 that SBP and DBP are calculated as maximum and minimum of continuous arterial blood pressure (ABP) waveform, respectively), and classifying a blood pressure state of the patient as one blood pressure state selected from low blood pressure, normal, prehypertension, and high blood pressure based on the determined systolic blood pressure and the determined diastolic blood pressure (Wang, Para 0006; “Jia-Wei Chen et al … the teachings of which are incorporated herein by reference in their entirety”. Chen et al discloses in Page 8 “… actual BP and PPG data and information from participants who were diagnosed with normotension (BP < 130/90 mm Hg), prehypertension, and hypertension …”, and in Page 14 “The abnormal BP levels were defined as hypertension (SBP/DBP ≥ 130/80 mmHg) and hypotension (SBP/DBP < 90/60 mmHg) …”);
transmitting the determined results to at least one of a central monitoring device and an electronic medical record system (Wang, Para 0006; “Jia-Wei Chen et al … the teachings of which are incorporated herein by reference in their entirety”. Chen et al discloses in Page 5 “… the measured PPG-based BP and heart rate (HR), as well as oxygen saturation, could be transmitted wirelessly to a mobile phone for display and recording of multimodal physiological data in the mobile application (APP)”).
Wang does not clearly and explicitly disclose:
generating, for a prediction section to be predicted and having a length corresponding to a length of the photoplethysmography measurement value for the precision section, a plurality of derived variables;
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points;
each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to a corresponding time point of the plurality of time points within the prediction section;
inputting, to a second learning model, the plurality of derived variables and the equal number of copies of the most recently measured non-invasive blood pressure measurement value; and
a learning model including a Cross-Attention layer and a forward neural network, and
converting the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and transmitting the voltage to a blood pressure measurement module of the commercial patient monitoring device, or transmitting the continuous arterial blood pressure waveform in digital form.
Addison in the same field of endeavor discloses:
for one time point, each derived variable being generated based on a time elapsed from a measurement time point of the most recently measured non-invasive blood pressure measurement value to the one time point (Addison, Para 0083; “Features 604 may be a value that corresponds [to] the time elapsed since the most recent calibration point.” The disclosed elapsed time corresponds to the claimed derived variable); and
inputting, to a learning model, the one derived variable (feature 604, or elapsed time) and one copy of the most recently measured non-invasive blood pressure measurement value (feature 606, or BP at Calibration Point) (Addison, Fig. 6B shows the features as input for determining blood pressure for a particular time t, which include feature 604 (corresponding to the claimed “derived variable” as discussed above) and feature 606 (“BP at Calibration Point”));
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, as suggested by Addison, in order for one time point of prediction to input a copy of blood pressure measured at calibration point and derive a variable based on elapsed time. One of ordinary skill in the art would have been motivated to make the modification of improving prediction accuracy for the estimated blood pressure (Addison, Para 0086; “… by using the calibration data received at the most recent calibration point as additional inputs for training CNIBP model 124, CNIBP model 124 may improve its accuracy in determining the blood pressure of patients.”).
Wang and Addison do not explicitly and clearly disclose:
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable and an equal number of copies of calibration factor;
a learning model including a Cross-Attention layer and a forward neural network; and
converting the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and transmitting the voltage to a blood pressure measurement module of the commercial patient monitoring device, or transmitting the continuous arterial blood pressure waveform in digital form.
Aguirre in the same field of endeavor discloses:
the prediction section including a plurality of time points, one derived variable of the plurality of derived variables being generated for each time point of the plurality of time points; inputting, to a learning model, the plurality of derived variable (predicted outputs at an earlier time-step as shown in Fig. 6, i.e. [y0, y1, …, yT-1]) and an equal number of copies of calibration factor (demographic information vector XDI as shown in Fig. 6) (Aguirre, Page 8, Para 1; “Age and gender demographic information vector (XDI) is concatenated at each time-step with yi to conform the decoder inputs (yi&DI).” As shown in Fig. 6, a number T of yi values and equal number of copies of XDI are input into the model); and
a learning model including a Cross-Attention layer (Aguirre, Abstract; “this methodology is capable of transforming PPG into an ABP pulse …”; Page 6, Para 2; “The proposed deep learning architecture is inspired by seq2seq encoder-decoder [36] models with attention mechanism.” Fig. 6 of Aguirre shows that the block “Attention” combines the outputs of both the encoder (from input Xl) and the decoder (from output of each time step concatenated with demographic information vector), so is a typical cross-attention operation) and a forward neural network (Aguirre, Fig. 6 show that the model also includes multiperceptron layers (MPLv and MPLc), which are feed-forward neural network layers).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang and Addison, as suggested by Aguirre, in order to prepare a plurality of derived variables paired with calibration factor for one implementation of model prediction and comprise a cross-attention component and a feed-forward component in the network. One of ordinary skill in the art would have been motivated to make the modification of preparing a plurality of derived variables for the benefit of improving prediction efficiency by producing a continuous waveform of blood pressure in a single prediction, and make the modification of using a model with a cross-attention layer for the benefit of achieving a cross-attention mechanism that is known to enable higher efficiency and accuracy as compared to regular neural networks.
Wang, Addison and Aguirre do not explicitly and clearly disclose converting the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and transmitting the voltage to a blood pressure measurement module of the commercial patient monitoring device, or transmitting the continuous arterial blood pressure waveform in digital form.
Barak in the same field of endeavor discloses:
converting the continuous arterial blood pressure waveform into a voltage using a digital-to-analog converter and transmitting the voltage to a blood pressure measurement module of the commercial patient monitoring device (Barak; Para 0265; “Wires 204 provide arterial pressure values to PCM 103, for example to be displayed.” Fig. 5 shows that the values first go through block 505 (D/A converter) before being transmitted by wires 204), or transmitting the continuous arterial blood pressure waveform in digital form (Barak; Para 0265; “D/A converter 505 are optional”; when such converter is not used, the transmitted data is in digital form).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Wang, Addison and Aguirre, as suggested by Barak, in order to transmit data between processing device and patient monitor using standard wired methods. One of ordinary skill in the art would have been motivated to make the modification for the benefit of transferring results in a reliable way and enabling timely monitoring of patient’s vital information.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LEI ZHANG whose telephone number is (571)272-7172. The examiner can normally be reached Monday-Friday 8am-5pm E.T..
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, Pascal Bui-Pho can be reached at (571) 272-2714. 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.
/L.Z./Examiner, Art Unit 3798
/PASCAL M BUI PHO/Supervisory Patent Examiner, Art Unit 3798
