DETAILED ACTION
Claims 1-2, 4-6, 9-11, 13-15, and 18-21 are presented for examination.
Claims 1, 2, 9-11, and 18-19 have been amended.
Claims 3, 7, 8, 12, and 16-17 have been cancelled.
Claim 21 is new.
This office action is in response to the amendment submitted on 9-Apr-2026.
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 Arguments – 35 USC 103
Applicant’s arguments with respect to the 103 rejections have been considered, but are moot in view of the new ground(s) of rejection provided below.
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 1-2, 4-6, 9-11, 13-15, and 18-21 are rejected under 35 U.S.C. 103 as being unpatentable over Anderson et al. (Modeling and prediction of pressure injury) in view of Drakos (US11978558B1) and further in view of Tschannen et al (The pressure injury predictive model) and further in view of Cha et al (US20200005900A1).
Regarding Claim 1, Anderson teaches collecting data relating to a first patient’s health comprising symptoms experienced by the first patient (Page 2, PI risk prediction using EHR “Widespread use of the electronic health record (EHR) has led to the collection of vast amounts of clinical data”).
wherein the data indicates that the first patient is flagged for previous wounds
sustained by the first patient (Page 2, Risk prediction complexity and PIPM “the Pressure Injury Prediction Model (PIPM) was developed representing 6 constructs: pressure, tissue tolerance, friction and shear, as well as three new constructs—patient characteristics, environment, and episode of care.” The PIPM model defined by Tschannen at al. and incorporated into Anderson includes, within the patient characteristics pg. 7
PNG
media_image1.png
257
638
media_image1.png
Greyscale
and episode of care pg. 14-16, previous injuries/wounds as defined by Tschannen et al.).
in response to the first patient being flagged for previous wounds: a machine learning model that is trained to generate output data indicating a likelihood that the first patient will develop a wound at one or more healthcare facilities, wherein the input data is based, at least in part, on the data relating to the first patient’s health (Pg. 1, Methods, “The AI training was based on large, incongruent, incomplete, heterogeneous, and time-varying data of hospitalized patients. Both model-based statistical methods and model-free AI strategies were used to forecast PI outcomes and determine the salient features that are highly predictive of the outcomes,” Pg. 1, Conclusions, “Our PI (Pressure Injury) prediction model provide a first generation of AI guidance to prescreen patients at risk for developing PIs,” and Pg. 11, Conclusion: “The app allows interactive forecasting of the probability of acquiring PI in different hospitalization settings”).
communicating, to a first healthcare facility of the one or more healthcare facilities, a message indicating an action that should be taken to reduce the first likelihood that the first patient will sustain the first wound while the first patient is at the first healthcare facility, facility (Pg. 11, Conclusion: “Findings from this study have identified risk profiles for various surgical services that must be considered when determining prevention intervention strategies to employ. The importance of getting this type of discrete information to the bedside nurse cannot be overstated. To meet this need, we are developing an interactive webapp that implements the RF model to predict PI within specific surgical services or globally for an entire hospital.” The prevention strategies reduce the risk profile).
However, Anderson does not appear to teach the symptoms experienced by the first patient necessitating in-patient care
wherein the action is tailored according to the first healthcare facility and the first patient
triggering execution of
wherein the triggering comprises providing input data to the machine learning model,
in response to symptoms experienced by a second patient not necessitating in- patient care or the second patient not being flagged for previous wounds, forego triggering execution of the machine learning model to process data relating to the second patient's health
Drakos teaches the symptoms experienced by the patient necessitating in-patient care (Col 45 Line 22-28 “When patients encountered by EMS meet certain criteria (i.e. there is some relatively high level of diagnostic certainty relative to the EMS medical providers' expertise) for the conditions mentioned above, those patients are transported directly to the relevant specialty center even if it means bypassing a closer medical facility”).
wherein the action is tailored according to the first healthcare facility and the patient (Col 19 line 41-46 “The PreDICT system determines that a patient at Hospital A with chest pain is experiencing an ST elevation myocardial infarction (STEMI) and requires a cardiac catheterization to relieve the coronary artery obstruction. Hospital A does not have this capability but Hospital B does”).
Anderson, Drakos, and Tschannen are analogous art because they are from the same field of endeavor in machine learning and healthcare. Anderson strictly applies machine learning to learn and predict probabilities of wounds, while Drakos creates a sophisticated diagnosis and routing application to minimize adverse effects and optimize outcomes for patients. Anderson incorporates the model and database defined by Tschannen. Before the effective filing date of the invention, it would have been obvious to a person of ordinary skill in the art, to combine Anderson, Drakos, and Tschannen to arrive at utilizing the predictive data models of wounds and injuries to optimally route the patient to the most equipped facility to minimize any adverse outcomes. “In this manner, users including expert and nonexpert users can provide information regarding a condition of a subject and receive timely and accurate information regarding risk stratification, treatment options and other medical evaluation information” (Drakos Col 6 line 41-45).
Cha teaches triggering execution of machine learning model ([0060] “At step 125, the system may generate a patient cohort comprising a subset of the preprocessed data records by, for example, filtering out any records that do not meet a cohorting criteria. Generally, cohorting criteria may relate to any information included in the preprocessed data records, such as various patient information. For example, cohorting criteria may relate to one or more of: demographic information (e.g., age, gender, race), DP information (e.g., presence of a specific DP code), laboratory tests information (e.g., presence of a value for a specific variable and/or specific values for a given test), genetics information (e.g., presence of a particular high-risk allele), medications information (e.g., usage of one or more particular medications), health program enrollment status, payer plan enrollment status, and/or other.” EN: Cha filters patients based on a set of criteria. Those that meet the criteria are passed down to the workflow which triggers the ML model. See Fig. 1 step 135, 140, and 145)
wherein the triggering comprises providing input data to the machine learning mode ([0063] “At step 130, various predictive features are created from the preprocessed patient information associated with each patient in the cohort. As discussed below, such features may be provided to the machine learning model to determine various risk information, such as a predictive value (i.e., a feature weight) of each feature and a risk score for each of the patients.” EN: after the filtering as per [0060], input data is provided to the ML model)
in response to symptoms experienced by a second patient not necessitating in- patient care or the second patient not being flagged for previous wounds, forego triggering execution of the machine learning model to process data relating to the second patient's health ([0060] EN: Cha teaches cohorting based on certain criteria. If the criteria are met, the ML model is triggered. If the criteria are not met, the ML model is not triggered. Drakos teaches the necessitating in-patient care as per Col 45 Line 22-28, Anderson teaches being flagged for previous wounds per Page 2, Risk prediction complexity and PIPM. )
Anderson, Drakos, Tschannen and Cha are analogous art because they are from the same field of endeavor in machine learning and healthcare. Before the effective filing date of the invention, it would have been obvious to a person of ordinary skill in the art, to combine Anderson, Drakos, Tschannen and Cha to incorporate Cha’s filtering of patient before triggering the model. A PHOSITA would be motivated to do so to achieve better efficiency, running resource intensive models only when necessary. “It will be appreciated that any number of features relating to the above-described groups may be created and employed by the machine learning models. Generally, it has been found that the creation of such groups provides enhanced interpretability of the various DP information found in the input data and allows for more efficient incorporation of such information into downstream operations.” (Cha, [0080])
Regarding Claim 2, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Drakos further teaches the output data comprises at least: first data indicating the likelihood that the first patient will develop the wound at the first healthcare facility; and second data score indicating the likelihood that the patient will develop the wound at a second health care facility of the one or more healthcare facilities, (Drakos Col 30 line 18-20 “For any possibility-set, there are multiple branches with different levels of probability that are often in dynamic interplay” and FIG. 6B. Showcases the comparison of probabilities and using the comparison in the time domain to determine the best course of action. Col 30 Line 47-56 “The underlying risk is progressing with time and the probability of realizing the terminal outcome (death) if appropriate intervention (to mitigate or avert the underlying risk, hemorrhage) is taken at time t is less than if appropriate intervention is taken at time (t+x), when the patient is experiencing more physiologic dysfunction. Stated differently, the patient has a higher probability of benefit (surviving) if intervention is accomplished at time t than at time (t+x)” Here there explicit probability branches being compared to each other. In the last example a (first) probability at time t is directly compared to a (second) probability at time (t+x).).
Regarding Claim 4, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Anderson further teaches collecting a dataset indicating (i) a plurality of past physical wounds sustained by different patients while the different patients were in the first healthcare facility (Page 4, Methods Data “Electronic health record (EHR) data for over 23,000 patient encounters discharged between June 1st, 2014 and June 26th, 2016 were obtained from the study site.” The data includes past wounds while at the first as well as other subsequent healthcare facilities).
(ii) a plurality of wound types of the plurality of past physical wounds (Page 2, Paragraph 1 “The authors note that likely contributors to their heterogeneous findings were that the studies varied on the PI stages that were included (e.g., Stage I, I & II, III & IV).” Each stage of PI is a different type of wound).
dividing the dataset into a training dataset and a validation dataset (Page 5, AI Modeling and analytics “The data was split into training: testing and a sequential network model was fit using the following parameter settings: units = 500, activation = relu or sigmoid, layer dropout rate in the range [0.3, 0.4], layer unit density between 2 and 128, loss function = binary cross-entropy, ADAM optimizer, accuracy metric, epochs = 100, batch size = 10, and validation split = 0.3”).
training the machine learning model using the training dataset (Page 3, Col 2 “The model built using random forest was the strongest, with precision of 0.998, and the average AUC of 1.00, in the training set”).
validating the machine learning model using the validation dataset after the machine learning model is trained (Page 3, Col 2 “however the AUC in the validation set the AUC was 0.864 with random forest still providing the best results”).
Regarding Claim 5, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Anderson further teaches removing portions of the dataset that were generated prior to a time threshold before training the machine learning model (Page 6, Results “After removal of cases that didn’t meet the study inclusion criteria (e.g. LOS ≥2 days, and undergoing a surgical procedure, and having staffing data).” This teaches removing data that doesn’t meet the cut off criteria).
Regarding Claim 6, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Anderson further teaches the first healthcare facility is flagged for previous wounds sustained at the first healthcare facility (Page 2, Risk prediction complexity and PIPM “the Pressure Injury Prediction Model (PIPM) was developed representing 6 constructs: pressure, tissue tolerance, friction and shear, as well as three new constructs—patient characteristics, environment, and episode of care.” The environment of care includes hospital flags such as hospital setting, type, certification status, teaching status and bed size as specified in the referenced study: “the sheer number of possible predictors and predictor combinations contributes to the complexity of PI risk assessment and prevention [4].” The PIPM is defined fully by Tschannen et al.
PNG
media_image2.png
150
651
media_image2.png
Greyscale
the environment variables are defined on pg. 12).
Drakos further teaches providing the input data to the machine learning model in response to determining that the first healthcare facility is within a distance threshold of the first patient (Col 19 Line 58-64 “The PreDICT system can evaluate historical transport data and real-time air and ground traffic data to determine that transport by helicopter will place the patient in the cardiac catheterization lab at Hospital B twelve minutes faster than transport by ground ambulance at this time of day due to heavy traffic volumes.” The training input was covered in Claim 1 above).
Regarding Claim 9, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Drakos further teaches communicating the message is in response to determining that the action reduces the likelihood that the patient will sustain the wound while the patient is at the first healthcare facility (Col 33 Line 37-47 Drakos teaches a sophisticated system for determining the probability of an incident and how to optimize/minimize its impact through a multi factor equation: “Considering tOR and its components yields the more specific function:
RB(tOR)=[DC(tDA)×PB(tOR)]−IR(tI) Equation 5:
Time of operational risk (tOR) components, or some of the components, will often be in dynamic interplay. For example, there may be several loops between tDX and tDA before a clear intervention and/or pathway to performing that intervention is identified. The components of tOR can be conceived of as a more comprehensive and detailed OODA (Observe, Orient, Decide, Act) Loop process that is not complete until the “act” is resulted”).
Regarding Claim 10, Drakos teaches An apparatus comprising: a memory; and a hardware processor communicatively coupled to the memory, the hardware processor configured to (Col 8 Line 27-28 “The illustrated user device 102 may include, for example, a smart phone, tablet computer or similar device”). The remaining limitations are similar to Claim 1 and are rejected for the same reason.
Regarding Claims 11, 13-15, and 18, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the apparatus of claim 10. Claims 11, 13-15, and 17-18 recite limitations similar to claims 2, 4-6, and 9 respectively and are rejected under the same rationale.
Claim 19 recites a non-transitory computer readable medium storing instructions: that, when executed by a processor, cause the processor to (Col 8 Line 28-29 “The user device 102 includes one or more sensors 110, a processor 112”).
The remaining limitations are similar to claim 1 and are rejected under the same rationale.
Claim 20, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the apparatus of claim 19. The remaining limitations are similar to claim 6 and are rejected under the same rationale.
Regarding Claim 21, Anderson in view of Drakos and further in view of Tschannen and further in view of Cha teaches the method of claim 1. Anderson teaches the input data further comprises historical wound data associated with the one or more healthcare facilities (Page 4, Methods Data “Electronic health record (EHR) data for over 23,000 patient encounters discharged between June 1st, 2014 and June 26th, 2016 were obtained from the study site.”).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US11410756B2: Discloses predicting health events from medical records.
US20200273581A1: Discloses post discharge risk prediction.
THIS ACTION IS MADE FINAL. 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 AMIR DARWISH whose telephone number is (571)272-4779. The examiner can normally be reached 7:30-5:30 M-Thurs.
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, Lewis Bullock can be reached on 571-272-3759. 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.
/A.E.D./Examiner, Art Unit 2199
/LEWIS A BULLOCK JR/Supervisory Patent Examiner, Art Unit 2199
Prosecution Insights