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
Applicant’s submission filed 01/31/2026 has been entered. Currently claims 1-13 are pending.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1-4 and 6-13 are rejected under 35 U.S.C. 103 as being unpatentable over Bashar et al., "Extraction of Heart Rate from PPG Signal: A Machine Learning Approach using Decision Tree Regression Algorithm," 2019 4th International Conference on Electrical Information and Communication Technology (EICT), Khulna, Bangladesh, 2019, pp. 1-6, doi: 10.1109/EICT48899.2019.9068845. (hereinafter “Bashar”) in view of Z. Zhang, "Photoplethysmography-Based Heart Rate Monitoring in Physical Activities via Joint Sparse Spectrum Reconstruction," in IEEE Transactions on Biomedical Engineering, vol. 62, no. 8, pp. 1902-1910, Aug. 2015, doi: 10.1109/TBME.2015.2406332 (hereinafter “Zhang”) and further in view of Chung et al., (US20240057881A1).
Regarding claim 1, Bashar teaches a method for estimating heart rate (HR) with high accuracy comprising (Abstract method to estimate heart rate):
(i) configuring a photoplethysmography (PPG) sensor of the wearable device to operate at a PPG sampling frequency (pg. 2 the dataset is sampled at 125 Hz)
(ii) obtaining signal from one or more photoplethysmography (PPG) sensor in contact with a subject (pg. 2 Data documentation the PPG data is measured from the wrist),
(iii) processing the signal using signal processing techniques to generate first HR estimations (pg. 2 III. Methodology the raw PPG signal is filtered; Primary processing of data an adaptive filter is used to preprocess the PPG signal and is used to identify features (Feature Engineering));
(iv) passing the first HR estimations through machine learning (ML) model generating second accurate HR estimations (pg. 2 III Methodology the ML learns from both noisy and non-noisy data to estimate HR; pg. 3 Heart Rate detection training data and actual HR is used for the noise/non noisy data, so the ML can learn and then estimate the HR) wherein the ML model is trained on data sampled to achieve a mean absolute percentage error (MAPE) of less than 5% to reference heart rate measurements (Abstract mean absolute error found was 1.18 beats per minute; looking at figs. 3a-6b the lowest BPM is a little below 80, so even if considering the lowest BPM was 60, the mean average prediction error would be 1.18/60, which would be less than 5%.) but is silent regarding operate at a PPG sampling frequency of 25 Hz or less to reduce power consumption and extend battery life of the wearable device; and data down sampled at 25 Hz or less.
In the same PPG field of endeavor, Zhang teaches operate at a PPG sampling frequency of 25 Hz or less to reduce power consumption and extend battery life of the wearable device (pg. 1907 the 125 hz samples are down sampled to 25 Hz, and would result in reduced power consumption and extend battery life); and data down sampled at 25Hz or less (pg. 1907 the 125 hz samples are down sampled to 25 Hz).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application to apply the technique of sampling at 25 Hz of Zhang with the method of Bashar, as both inventions relate to PPG signals that are initially sampled at 125 Hz, and would yield the predictable result of a method of using a raw ppg signal sampled at 125Hz to estimate a HR that is down sampled to 25Hz to one of ordinary skill. One of ordinary skill would be able to perform such an application, and the results of Bashar down sampling the PPG signals to 25 Hz are reasonably predictable. This down sampling would reduce the processing load for machine learning models and thus make them faster.
However, the combination of references are still silent regarding (v) and outputting the second HR estimations to a display of the wearable device for real-time monitoring.
In the same heart rate estimation field of endeavor, Chung teaches (v) and outputting the second HR estimations to a display of the wearable device for real-time monitoring ([0112] the final HR estimation is sent to a user device 202 such as a watch and [0113] is displayed;[0107] the HR estimates are in real time).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application to combine the method of modified Bashar with the display of the wearable device as taught by Chung, as both inventions relate to heart rate estimations, and would yield the predictable result of a heart rate estimation method that outputs to a display on a wearable device to one ordinary skill. One of ordinary skill would be able to perform such a combination, and the results of the method of modified Bashar outputting the heart rate estimations to a device of a wearable device are reasonably predictable.
Regarding claim 2, modified Bashar teaches the method of claim 1, wherein Bashar further teaches wherein the ML is selected from Decision Tree (DT) (Abstract Decision Tree is used).
Regarding claim 3, modified Bashar teaches the method of claim 1, wherein Bashar further teaches wherein the ML is a DT (Abstract Decision tree regression is used to fit and measure the HR).
Regarding claim 4, modified Bashar teaches the method of claim 3, wherein Bashar further teaches wherein the DT has 10-20 input features (pg. 2 Feature Engineering inputs include peak position after spectral subtraction of the first and acceleration signal, peak position of first PPG signa and Second PPG signal, acceleration signal’s peak power and position, PPG signal one and two peak power, X, Y, Z-axis acceleration signal’s peak power and position)
Regarding claim 6, modified Bashar teaches the method of claim 1, wherein Bashar further teaches placing the PPG on the skin of the subject (pg. 2 Data documentation the PPG data is measured from the wrist).
Regarding claim 7, modified Bashar teaches the method of claim 1, wherein Bashar further teaches wherein the PPG is placed on a subject’s wrist (pg. 2 Data documentation the PPG data is measured from the wrist).
Regarding claim 8, Bashar teaches a wearable device comprising (pg. 2 a pulse oximeter is used):
(i) a photoplethysmography (PPG) sensor configured to contact skin of a subject and operate at a PPG sampling frequency (pg. 2 the dataset is sampled at 125 Hz and is measured from the wrist)
(ii) a processor coupled to the PPG sensor and configured to process a signal obtained from the PPG sensor using signal processing techniques to generate first HR estimations (pg. 2 III. Methodology the raw PPG signal is filtered; Primary processing of data an adaptive filter is used to preprocess the PPG signal and is used to identify features, and would inherently have a processor (Feature Engineering)) and pass the first HR estimations through machine learning (ML) model generating second accurate HR estimations (pg. 2 III Methodology the ML learns from both noisy and non-noisy data to estimate HR; pg. 3 Heart Rate detection training data and actual HR is used for the noise/non noisy data, so the ML can learn and then estimate the HR) wherein the ML model is trained on data sampled to achieve a mean absolute percentage error (MAPE) of less than 5% to reference heart rate measurements (Abstract mean absolute error found was 1.18 beats per minute; looking at figs. 3a-6b the lowest BPM is a little below 80, so even if considering the lowest BPM was 60, the mean average prediction error would be 1.18/60, which would be less than 5%.) but is silent regarding operate at a PPG sampling frequency of 25 Hz or less to reduce power consumption and extend battery life of the wearable device; and data down sampled at 25 Hz or less.
In the same PPG field of endeavor, Zhang teaches operate at a PPG sampling frequency of 25 Hz or less to reduce power consumption and extend battery life of the wearable device (pg. 1907 the 125 hz samples are down sampled to 25 Hz, and would result in reduced power consumption and extend battery life); and data down sampled at 25Hz or less (pg. 1907 the 125 hz samples are down sampled to 25 Hz).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application to apply the technique of sampling at 25 Hz of Zhang with the method of Bashar, as both inventions relate to PPG signals that are initially sampled at 125 Hz, and would yield the predictable result of a method of using a raw ppg signal sampled at 125Hz to estimate a HR that is down sampled to 25Hz to one of ordinary skill. One of ordinary skill would be able to perform such an application, and the results of Bashar down sampling the PPG signals to 25 Hz are reasonably predictable. This down sampling would reduce the processing load for machine learning models and thus make them faster.
However, the combination of references are still silent regarding (iii) and a display coupled to the processor and configured to output the second HR estimations for real-time monitoring.
In the same heart rate estimation field of endeavor, Chung teaches (iii) and a display coupled to the processor and configured to output the second HR estimations for real-time monitoring ([0112] the final HR estimation is sent to a user device 202 such as a watch and [0113] is displayed;[0107] the HR estimates are in real time; [0091] processor within electronics 199).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application to combine the method of modified Bashar with the display of the wearable device as taught by Chung, as both inventions relate to heart rate estimations, and would yield the predictable result of a heart rate estimation method that outputs to a display on a wearable device to one ordinary skill. One of ordinary skill would be able to perform such a combination, and the results of the method of modified Bashar outputting the heart rate estimations to a device of a wearable device are reasonably predictable.
Regarding claim 9, modified Bashar teaches the device of claim 8, wherein Bashar further teaches wherein the ML is selected from Decision Tree (DT) (Abstract Decision Tree is used).
Regarding claim 10, modified Bashar teaches the device of claim 8, wherein Bashar further teaches wherein the ML is a DT (Abstract Decision tree regression is used to fit and measure the HR).
Regarding claim 11, modified Bashar teaches the device of claim 10, wherein Bashar further teaches wherein the DT has 10-20 input features (pg. 2 Feature Engineering inputs include peak position after spectral subtraction of the first and acceleration signal, peak position of first PPG signa and Second PPG signal, acceleration signal’s peak power and position, PPG signal one and two peak power, X, Y, Z-axis acceleration signal’s peak power and position)
Regarding claim 12, modified Bashar teaches the device of claim 8, wherein Bashar further teaches placing the PPG on the skin of the subject (pg. 2 Data documentation the PPG data is measured from the wrist).
Regarding claim 13, modified Bashar teaches the method of claim 8, wherein Bashar further teaches wherein the PPG is placed on a subject’s wrist (pg. 2 Data documentation the PPG data is measured from the wrist).
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Bashar in view of Zhang as applied to claim 3 above, and further in view of Yu et al., “Hybridizing Personal and Impersonal Machine Learning Models for Activity Recognition on Mobile Devices” Year 2016 MOBICASE ACM DOI: 10.4108/eai.30-11-2016.2267108 And Hong et al., “Nanosecond machine learning event classification with boosted decision trees in FPGA for high energy physics” 2021 JINST 16 P08016 (hereinafter “Hong”).
Regarding claim 5, modified Bashar the method of claim 3, but fails to explicitly disclose wherein DT models have a model size of about or less than 10 KB.
However in a reference pertinent to the problem faced by the inventor of reducing the ML model footprint, Yu teaches wherein DT models have a model size of about or less than 10 KB (Table 5 the DT model has a model size of 4.272KB ).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application apply the footprint of Yu to the method of modified Bashar, as both inventions relate to PPG signals and would yield the predictable result of a DT model having a small footprint of 4.272 KB to one of ordinary skill in the art. One of ordinary skill would be able to make such an application, and the results of modified Bashar having a DT model of 4.272 KB are reasonably predictable. This would be an improvement as it would reduce decision time making by reducing the load of the machine learning AI.
However, the combination of references are still silent regarding a shorter inference time of less than 3 microseconds (μs).
In the same machine learning field of endeavor, Hong teaches a shorter inference time of less than 3 microseconds (μs) (pg. 3 2 ML Training the algorithms make decisions in a fraction of a microsecond).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the instant application apply the decision time of Hong to the method of modified Bashar, as both inventions relate to DT machine learning techniques and would yield the predictable result of a DT model having a fraction of a microsecond decision time to one of ordinary skill in the art. One of ordinary skill would be able to make such an application, and the results of modified Bashar having a DT model that makes decisions in a fraction of a microsecond are reasonably predictable. This would be an improvement as it would facilitate the measuring of HR and reduce the time it takes for a user to receive the heart rate.
Response to Arguments
Applicant's arguments filed 01/31/2026 have been fully considered but they are not persuasive.
Regarding claim 1, Applicant has argued that Bashar, Zhang, or their combination do not teach retraining the machine learning model to down sampled 25 Hz or less PPG data in order to recover high accuracy, and argues that reducing the sampling rates to 25 Hz would destroy feature fidelity.
Examiner disagrees. Such arguments are unsupported attorney arguments because neither Bashar nor Zhang teaches that machine learning based HR estimations would be inoperable or incompatible with lower-rate sampled PPG data. Further, Zhang demonstrates that down sampling the data to 25Hz still achieves accurate HR estimates with an error percentage of 1.01%±2.29% (see Zhang pg. 1907). Thus, it would be obvious to one of ordinary skill in the art to combine Bashar and Zhang to achieve a ML model that is trained on 25 Hz or less.
Applicant further argues that neither reference teaches both extended battery life and <5% MAPE at ≤25 Hz.
Examiner disagrees. A recitation of the intended use of the claimed invention must result in a structural difference between the claimed invention and the prior art in order to patentably distinguish the claimed invention from the prior art. If the prior art structure is capable of performing the intended use, then it meets the claim. As such, since the combination of Zhang and Bashar already teaches a device that is capable of a sample rate of 25 Hz and has a MAPE of less than 5%, then it would be capable of reducing power consumption and extending battery life of the wearable device.
Applicant further argues that Burrello and Hong (in combination with Bashar and Zhang) fails to teach the claim limitations of claim 5, stating that Burrello teaches a deep temporal convolution network (that is far larger and more computationally intensive), and that Hong teaches boosted decision trees for FPGA for high-energy physics event classifications. However this argument has been rendered moot, as Burrello has not been used to teach, and instead Yu ( in combination with Bashar, Zhang, and Hong) has been used.
The remaining claims are rejected for substantially the same reasons as above.
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 MICHAEL Y FANG whose telephone number is (571)272-0952. The examiner can normally be reached Mon - Friday 9:30 am - 6:00pm.
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 5712722714. 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.
/MICHAEL YIMING FANG/Examiner, Art Unit 3798
/PASCAL M BUI PHO/Supervisory Patent Examiner, Art Unit 3798
Prosecution Insights