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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed 1/28/2026 has been entered.
Status of Claims
This action is a non-final rejection
Claims 1-6, 9-12, 14-15, 26-30 are pending
Claim 1, 26 was amended
Claims 7, 8, 13, 16-25 were cancelled
Claims 27-30 were added
Claims 1-6, 9-12, 14-15, 26-30 are rejected under 35 USC § 101
Claims 1-6, 9-12, 14-15, 26-30 are rejected under 35 USC § 103
Priority
Acknowledgement is made of Applicant’s claim for a domestic priority date of 4-9-2021
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 11-9-2022, 1-9-2025 and 1-28-2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 101
35 U.S.C. 101 reads as follows:
Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title.
Claims 1-6, 9-12, 14-15, 26-30 are not patent eligible because the claimed invention is directed to an abstract idea without significantly more.
Analysis
First, claims are directed to one or more of the following statutory categories: a process, a machine, a manufacture, and a composition of matter. Regarding claims 1-6, 9-12, 14-15, 26-30 the claims recite an abstract idea of “evaluating patient treatment predictions”.
Independent Claims 1, 26 are rejected under 35 U.S.C 101 based on the following analysis.
-Step 1 (Does the claim fall within a statutory category? YES): claims 1 and 26 recite a method and one or more non-transitory computer-readable storage media respectively of evaluating patient treatment predictions.
-Step 2A Prong One (Does the claim fall within at least one of the groupings of abstract ideas?: YES):
Obtaining patient profiles for a plurality of patients, wherein each patient profile corresponds to a particular patient and includes (i) values for an expanded set of predictive features about the particular patient and (ii) a target response classification that indicates whether the particular patient responded to a particular medical treatment according to one or more criteria;
training, a treatment prediction model using the patient profiles, that causes the treatment prediction model to learn to predict, based on predictive features from a patient profile for a given patient, a likelihood that the given patient will respond to the particular medical treatment according to the one or more criteria;
identifying, a first subset of predictive features from the expanded set of predictive features that are most predictive of whether a given patient will respond to the particular medical treatment according to the one or more criteria;
the first subset of predictive features including at least one predictive feature defining a personal attribute of the patient's mental health, physical health, or socioeconomic status;
wherein at least one predictive feature in the first subset of predictive features is a shared predictive feature that is utilized across a plurality of treatment prediction models corresponding to different medical treatment modalities, the shared predictive feature being identified across the plurality of treatment prediction models;
configuring the treatment prediction model to generate predictions from values for the first subset of predictive features that are most predictive, without requiring values for a second subset of predictive features that are less predictive than features from the first subset;
wherein the configuring includes, after initially training the treatment prediction model on the expanded set of predictive features, re-training the treatment prediction model only on the first subset of predictive features to exclusion of the second subset of predictive features;
wherein at least one of a size or a computational complexity of the treatment prediction model is reduced as a result of re-training the treatment prediction model
applying the treatment prediction model to generate a treatment response prediction for a new patient.
belong to the grouping of mental processes under concepts performed in the human mind as it recites “evaluating patient treatment predictions”. Alternatively, the selected abstract idea belongs to the grouping of certain methods of organizing human activity under managing personal behavior or relationships or interactions between people as it recites “evaluating patient treatment predictions”. (refer to MPP 2106.04(a)(2)). Accordingly this claim recites an abstract idea.
-Step 2A Prong Two (Are there additional elements in the claim that imposes a meaningful limit on the abstract idea? NO).
Claims 1, 26 recite:
system comprising one or more computers;
applying a machine-learning technique;
Amounting to additional elements that are recited at a high-level of generality such that it amounts to no more than mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. Support for this can be found in the specification, paragraphs (0011-0014). [0033-0040]. (refer to MPEP 2106.05(f)). Accordingly, the claim as a whole does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
-Step 2B (Does the additional elements of the claim provide an inventive concept?: NO. As discussed previously with respect to Step 2A Prong Two,
Claims 1, 26 recite:
system comprising one or more computers;
applying a machine-learning technique;
Amount to additional elements that are recited at a high-level of generality such that it amounts to no more than mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. Support for this can be found in the specification, paragraphs (0011-0014). [0033-0040]. (refer to MPEP 2106.05(f)) Accordingly, even in combination the additional elements of the claim do not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible.
Dependent Claims:
Step 2A Prong One: The following dependent claims recite additional limitations that further define the abstract idea of “evaluating patient treatment predictions”. These claim limitations include:
Claim 3: wherein the particular medical treatment is a treatment for alleviating chronic pain;
Claim 4: wherein the treatment for alleviating chronic pain is one of a multimodal treatment, a class of medication, a particular medication, a class of injection, a particular type of injection, an implanted medical device, a behavioral medicine treatment, an integrative medicine treatment, a rehabilitation therapy, or an orthotic or assistive device;
Claim 5: wherein the expanded set of predictive features include features indicating at least one of: Charlson co-morbidity index, post-traumatic stress disorder (PTSD), pain experience duration, opioid use or misuse, neuropathic pain level, Medicaid status, socioeconomic factors in region of patient's residence, medical diagnoses, medication prescriptions, medical procedure history, body pain map, patient age, patient sex, patient education level, tobacco use, alcohol use, illicit drug use, anxiety level, depression level, global mental health level, global physical health level, pain interference, positive outlook level, or sleep disturbance level;
Claim 6: wherein configuring the treatment prediction model to generate predictions from values for the first subset of predictive features that are most predictive comprises establishing a recommendation or a requirement that, after training, patient profiles for new patients cannot include missing values for the first subset of predictive features as a condition of using the treatment prediction model to generate treatment response predictions for the new patients;
Claim 9;
identifying that the patient profile for a first particular patient is missing a value for a first feature in the expanded set of predictive features included in the patient profile; and
in response to identifying that the patient profile for the first particular patient is missing the value for the first feature, before using the patient profile for the first particular patient in training the treatment prediction model, imputing a value for the first feature.
Claim 10: wherein imputing the value for the first feature comprises;
if the first feature is a continuous variable, assigning an average value of the first feature from other patient profiles as the value for the first feature in the patient profile for the first particular patient;
or if the first feature is a discrete variable, assigning a null value as the value for the first feature in the patient profile for the first particular patient;
Claim 11:
obtaining, updated patient profiles that include data related to additional patients, additional predictive features, or both, which were not in the patient profiles on which the treatment prediction model was previously trained;
identifying, and based on the updated patient profiles, a third subset of predictive features that are most predictive of whether a given patient will respond to the particular medical treatment according to the one or more criteria; and
re-configuring the treatment prediction model to generate predictions from values for the third subset of predictive features that are most predictive;
Claim 12:
wherein the third subset of predictive features includes at least one predictive feature that is not among the predictive features in the first subset; and
re-configuring the treatment prediction model comprises adding a recommendation or requirement that patient profiles for new patients cannot include missing values for the at least one predictive feature as a condition of using the treatment prediction model to generate treatment response predictions for the new patient;s
Claim 14: wherein the one or more criteria comprise achieving at least one of:
(i) a threshold improvement in average pain intensity within a predetermined time interval;
(ii) a threshold improvement in physical function within a predetermined; time interval;
or (iii) a threshold improvement in patient's overall impression of change;
Claim 15: wherein training the treatment prediction model comprises achieving at least an area under receiver operating curve (AUROC) or selective area under receiver operating curve (SAUROC) score of 0.65;
Claim 27: wherein identifying the first subset of predictive features includes selecting a predetermined number of most predictive features in the expanded set of predictive features;
Claim 28: wherein identifying the first subset of predictive features includes determining which features in the expanded set of predictive features exceed a threshold level of predictive power;
Claim 29: wherein the shared predictive feature reduces a number of feature values to be derived when evaluating the plurality of treatment prediction models;
Claim 30: wherein the shared predictive feature reduces an amount of information collected from the new patient to generate treatment prediction responses using the plurality of treatment prediction models including the treatment response prediction using the treatment prediction model.
Step 2A Prong Two (Are there additional elements in the claim that imposes a meaningful limit on the abstract idea? NO). The following dependent claims recite additional elements that are recited at a high-level of generality such that it amounts to no more than mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, the claims as a whole do not integrate the abstract idea into a practical application because they do not impose any meaningful limits on practicing the abstract idea. The claims include:
Claim 2:
wherein the treatment prediction model is a random forest ensemble comprising a plurality of decision trees;
Claim 11:
by the system;
Step 2B (Does the additional elements of the claim provide an inventive concept?: NO). As discussed previously with respect to Step 2A Prong Two, the following dependent claims recite additional elements that are recited at a high-level of generality such that it amounts to no more than mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, even in combination the additional elements of the claim do not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible. The claims include:
Claim 2:
wherein the treatment prediction model is a random forest ensemble comprising a plurality of decision trees;
Claim 11:
by the system.
Claim Rejections - 35 USC § 102
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 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 the appropriate paragraphs of 35 U.S.C. 102 that form
the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale or otherwise available to the public before the effective filing date of the claimed invention.
Claims 1, 6, 9, 26-30 are rejected by 35 U.S.C. 102(a)(1) as being anticipated by De Bruin et.al (US 20110119212 A1) hereinafter “De Bruin”
Regarding claims 1 and 26 De Bruin teaches:
obtaining, by a system comprising one or more computers (medical digital expert system), patient profiles for a plurality of patients, wherein each patient profile corresponds to a particular patient and includes (i) values for an expanded set of predictive features about the particular patient (array of neuro-psycho-biological indicators) and (ii) a target response classification that indicates whether the particular patient responded to a particular medical treatment (response-probabilities associated with a range of possible treatments for the condition diagnosed) according to one or more criteria; (See at least [0069] via: “…As shown in FIG. 1, in one embodiment, the "medical digital expert system" operates in the following manner. The physician, or an assistant, collects as much relevant biometric, demographic, neurological, psychological, psychiatric, laboratory and clinical information (101) as possible regarding the patient. This information could include an array of neuro-psycho-biological indicators such as demographic information, past history, symptomatic presentation, a list of medical co-morbidities, results of laboratory testing, selected measures of personality and cognitive functioning, pharmacogenetic data, and biological data derived from electrophysiological, magnetic, electromagnetic, radiological, optical, infra-red, ultrasonic, acoustic, biochemical, medical imaging and other investigative procedures and attributes. The physician's presumptive diagnosis is also provided if available. This data is then either processed on-site using a computer algorithm pre-loaded into the user's computer or similar digital processing device, or sent electronically (103) to a remote central processing site. In both instances the data will be analyzed according to a machine learning and inference process (104-106). This process will generate a report consisting of the response-probabilities associated with a range of possible treatments for the condition diagnosed, and optionally, a list of diagnostic possibilities rank-ordered by likelihood. The list of recommended treatments with associated response probabilities, and optionally, a list of diagnostic possibilities rank-ordered by probability or likelihood, is then sent to the physician in a timely fashion (100-103)…”)
training, by the system, a treatment prediction model using the patient profiles, including applying a machine-learning technique (machine learning and inference) that causes the treatment prediction model to learn to predict, based on predictive features from a patient profile for a given patient, a likelihood that the given patient will respond to the particular medical treatment (estimate the probability of response to a range of treatment possibilities appropriate for the illness diagnosed) according to the one or more criteria; (See at least [0017] via: “…generating a first level training dataset comprising a plurality of records comprising patient related clinical, symptomatic and laboratory data from a large number of patients on a system database..”; in addition see at least [0021] via: “…using the treatment-planning and diagnosis datasets in a learning or training procedure to construct the processors and models and find their unknown parameters which are used for assessment of the patient, medical/clinical diagnosis and treatment planning models..”; in addition see at least [0022] via: “…the invention provides a method for predicting patient response, in which the method comprises generating, and storing on a system database, a first level training dataset comprising a plurality of records comprising measured patient related clinical and/or laboratory data feature information from a large number of patients, the data including data relating to patient treatment response; wherein the measured patient related clinical and/or laboratory data is processed to extract features from the measured data..”; n addition see at least [0045] via: “…The expert prediction system of the present invention is based on a machine learning methodology. The neuro-psycho-biological information obtained from the patient before starting the treatment, before treatment planning or before making clinical/medical diagnosis makes a multidimensional observation vector…”; in addition see at least [0069] via: “…the data will be analyzed according to a machine learning and inference process (104-106). This process will generate a report consisting of the response-probabilities associated with a range of possible treatments for the condition diagnosed, and optionally, a list of diagnostic possibilities rank-ordered by likelihood. The list of recommended treatments with associated response probabilities, and optionally, a list of diagnostic possibilities rank-ordered by probability or likelihood, is then sent to the physician in a timely fashion (100-103)…”; in addition see at least [0071] via: “…The system and methodology of the present invention is based on advanced "signal/information processing" and "machine learning and inference" techniques. This invention includes a digital automated medical expert system capable of integrating diverse sets of neurological, psychological, psychiatric, biological, demographic and other clinical data to enhance the effectiveness of the physician by using machine learning and inference methods to estimate the probability of response to a range of treatment possibilities appropriate for the illness diagnosed, and, optionally, to provide a list of diagnostic possibilities rank-ordered by likelihood/probability…”; in addition see at least [0082] via: “…part of the medical expert system which processes the data, extracts the critical and useful information, provides a list of diagnostic possibilities and predicts the treatment options. See 105 and 110. In doing so, this subsystem uses the training data containing information about the therapies used and clinical response in patients previously treated for this condition (106), employing machine learning and inference methodologies to find a list of the best treatment options that can be sent to the clinician/user. Here, the data base analysis duplicates and supplements the clinical acumen of the clinician. This subsystem could, optionally, identify key pieces of missing data and the communication and interface subsystem would send a prompt to the clinician requesting this further information. The expert system would then run its algorithm without this additional data then again if and when the missing data is provided. The data collection and data analysis results can be repeatedly calculated over several sessions until acceptable reliability is achieved or the medical expert system is not able to provide better results. This process is illustrated in FIG. 7..”)
identifying, by the system, a first subset of predictive features (a reduced feature data subset) from the expanded set of predictive features (whose cardinality is less than that of the first level training data set) that are most predictive (select features appearing to discriminate for a useful prediction) of whether a given patient will respond to the particular medical treatment according to the one or more criteria; (See at least [0023] via: “…processing the extracted feature dataset to select a reduced feature data subset whose cardinality is less than that of the first level training data set containing the clinical and/or laboratory data features; the reduced feature data subset being derived by means of processing the data to select features appearing to discriminate for a useful prediction..”)
the first subset of predictive features including at least one predictive feature defining a personal attribute of the patient's mental health, physical health, or socioeconomic status; (See at least [0204] via: “...in the case of a mental or psychiatric illness, as an example, in addition to EEG recordings, numerous clinical variables can be collected using questionnaires, depression and anxiety rating scales, symptom-checklists and personality and cognitive assessment. For MDD, the range of data collected could include co morbid diagnoses, age, sex, marital status, education level, social support measured using the Perceived Social Support from Friends and Family rating scale (PSS) or similar scales, quality of life measured using the SF-36 or similar quality of life rating instruments. Previous medication therapy and past compliance could be measured using a modified version of the Michigan Adequacy of (previous) treatment scale (MATS) or similar instrument. Psychological and personality attributes such as neuroticism and other might be measured using the NEO-PI and Minnesota Multiphasic personality Inventory (MMPI) (115 to 118, and 120, 122, 124)) or similar instrument. The measurements to assess depression severity and symptom profile could include the Montgomery Asberg (MA) and Beck Depression (BD) scales, while anxiety could be measured using the Spielberger State Trait Anxiety Index (STAI) or similar instruments..”; in addition see at least [0205] via: “...In the case of a mental or psychiatric illness, these clinical variables are combined with the features extracted from the EEG. This approach differentiates the disclosed method from the method of U.S. Pat. No. 7,177,675 (Suffin) which emphasizes that the EEG data alone are sufficient for prediction of anti-depressant response. In our early studies an improvement of 4% was found in average performance simply by including some very basic clinical information such as the scores on the Beck depression rating scales. Methods of efficiently reducing the feature set in combination with modern mathematically-based classifier and predictor structures and models are provided by the present invention..”; in addition see at least [0206] via: “...Various instruments and methods of medical imaging can be used as part of clinical and laboratory data. For example, in the case of mental and psychiatric illnesses and disorders, although the EEG is currently the mainstay of the brain activity measurement system in the described invention other technologies such as fMRI, MEG, positron emission tomography (PET), single photon emission computerized tomography (SPECT), MRI, magnetic resonance spectrometry (MRS), radiography, and various other medical imaging instruments and probes could also be used. The brain activity data derived from these methods could also be employed in addition to or instead of the data derived from the EEG..”; in addition see at least [0207] via: “...For example, for psychiatric and mental illnesses and disorders, the data of fMRI can be used. There are at least two reasons for using the fMRI: 1) the fMRI and EEG are complementary methods for the analysis of brain activity. The spatial resolution is in the range of millimeters with fMRI and the time resolution is in the range of milliseconds with EEG. Combining the two approaches will therefore yield a more complete assessment of brain activity than what is available with either method used in isolation, 2) During the update and improvement stage of the medical digital expert system, combining fMRI and EEG may help in identifying better features than those which can be extracted using the EEG alone. Ancillary techniques such as mood induction may further enhance predictive power of our method as these strategies may increase the probability that the brain regions activated during the fMRI scan are related to mood disorder, and not to extraneous and random variables or interferences..”)
wherein at least one predictive feature in the first subset of predictive features is a shared predictive feature (mutual information criterion) that is utilized across a plurality of treatment prediction models corresponding to different medical treatment modalities, the shared predictive feature being identified across the plurality of treatment prediction models (See at least [0098] via: “...After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “... Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained...”; in addition see at least [0100] via: “... the feature selection is based on a mutual information criterion. The features selected are those which exhibit the highest mutual information with the training set...’; in addition see at least [0102] via: “...the selected subset of features is the subset which is most useful for the estimator/predictor method. This criterion combines feature selection with the estimation/prediction method and the goal is to generate the best final result with the greatest possible efficiency...”)
configuring the treatment prediction model to generate predictions (expert system uses feature reduction and feature selection to provide a smooth estimator/predictor) from values for the first subset of predictive features that are most predictive (features selected for the reduced feature data subset), without requiring values for a second subset of predictive features that are less predictive (only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained) than features from the first subset; (See at least [0025] via: “…generating a subject patient dataset including the features selected for the reduced feature data subset..”; in addition see at least [0026] via: “…comparing the subject patient dataset to the feature data scheme or model to predict a response for the patient..”; in addition see at least [0099] via: “…Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained..”)
wherein the configuring includes, after initially training the treatment prediction model on the expanded set of predictive features, re-training the treatment prediction model (The data collection and data analysis results can be repeatedly calculated over several sessions until acceptable reliability is achieved) only on the first subset of predictive features to exclusion of the second subset of predictive features (only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained); (See at least [0058] via: “...FIG. 5 is a flow chart illustrating a gradual and adaptive update and improvement process in accordance with one embodiment of the invention..”; in addition see at least [0082] via: “...The data collection and data analysis results can be repeatedly calculated over several sessions until acceptable reliability is achieved or the medical expert system is not able to provide better results. This process is illustrated in FIG. 7...”; in addition see at least [0085] via: “...the treatment recommendation procedure, medical diagnosis, and the estimation/detection of critical parameters, level or severity of the illnesses, diseases or conditions is performed through an interactive, iterative, multi-stage procedure whereby the first of a possible series of diagnostic or treatment recommendation reports is accompanied by a request for further data from laboratory testing, clinical examination or symptomatic enquiry, and then the additional data requested will be used to further refine the detection/estimation/prediction result in a subsequent re-analysis and further information processing. An example is shown in FIG. 7 where the data collection and data analysis can be repeated many times. This bi-directional information transfer, interactive interface and communication method reduces ambiguity in the diagnosis and treatment planning decisions and reports..”; in addition see at least [0098] via: “…After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “…Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained..”; in addition see at least [0123] via: “...Based on FIG. 7, similar to the procedure used by a human physician, the user-interface software of the "digital clinician" system works as follows: First a first set of clinical, laboratory data and symptomatic information (which includes the most basic and general information) are input to the software (165). Then the "digital clinician" or the "medical digital expert system" processes the data (166) and produces diagnosis/treatment-planning report, which is accompanied with confidence values, or decision likelihood values (167). The report can be some numbers or can be of graphical forms that show the level of confidence and reliability that the modeling and data analysis have on the result. Then the user may need more decision reliability and therefore, if wanted, goes to the next stage, in which the "digital clinician" asks for more specific clinical and laboratory information (170). Upon getting the new set of data, the "medical digital expert system" uses the collection of old and new sets of data to generate better results (171). And this procedure iterates for as long as the digital expert system and the user can handle..”; in addition see at least [0124] via: “...The user-interface of the digital clinician system is based on a structure similar to a `decision tree` where as one goes down the branches of the decision tree moving from general information to more specific information as the level of uncertainty is decreased. In the second level and beyond, the algorithm asks more specific questions that enables it to differentiate among possible solutions. It is a knowledge-based system that uses a structure similar to troubleshooting systems/software. The information that the software requests of the user is initially designed by the expert human physicians, incorporating and duplicating their best clinical practice (based on their experience and published clinical/medical treatment and diagnosis guidelines). The user-interface is connected via the communication medium to the central/remote data management and databases. It uses machine learning procedures to update and improve the initial structure and information that were given by human physician experts, and to make better estimates of the unknowns (such as the decision weight of each rule, and the detection threshold in each rule, etc.). The type and form of the information requested of the user in the second and subsequent stages are learned by the algorithm based on the experience derived through the use of the training databases. By employing cognitive adaptation and training procedures the performance of the algorithm or "digital clinician" improves with use..”) The Examiner interprets the process as depicted in fig. 5 and 7 whereby the data feature space increases in reliability over each successive processing step. See steps 145, 146, 147 steps of fig. 5.
wherein at least one of a size or a computational complexity of the treatment prediction model is reduced as a result of re-training the treatment prediction model; (See at least [0046] via: “...The first step in a machine learning procedure is to transform the observed data into a discriminative feature space. The extraction of relevant features/attributes is always a critical issue in any machine learning application. Effective features depend greatly on the underlying problem, but typically they include an assortment of various attributes characterizing statistical, geometrical, temporal, hierarchical and dynamic model properties of the measured data. The number Nc of candidate features can therefore become very large. For optimum performance of the prediction/classification algorithm, it is necessary to reduce the dimensionality of the feature and to transform the candidate set containing Nc features into a subset containing only Ni features, where Ni ≤ Nc...”; in addition see at least [0047] via: “..The Ni features are extracted from the observed data and assembled into a vector xi εRNi The associated class, treatment responsiveness label or target value (assumed known) corresponding to the observed data sample i (of a patient) is denoted by the variable yiεN (classification) or yiεN (regression). The set Di= {(xi , yi ), i=1, . . . , My} is referred to as the training set, where Mt is the number of training samples available. Given a set of training patterns, the objective is to determine the function f. For classification, f establishes decision boundaries in the feature space which effectively separate patterns belonging to different classes. For regression, f is determined to give as close a fit as possible to the values yi corresponding to the input feature vector xi ...”; in addition see at least [0048] via: “...If Ni is relatively large compared to Mt, then numerical stability and over-training issues can arise, with the result that the estimated function f has large variance. If Ni is too small relative to Nc , then the selected features may not adequately describe the clustering behaviour of the observations. Both of these scenarios lead to poor performance..”; in addition see at least [0049] via: “...Therefore, it is apparent that only those features which provide the best discriminating power for the classifier/predictor should be retained..”; in addition see at least [0099] via: “…Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained..”) and
applying the treatment prediction model to generate a treatment response prediction (provide a smooth estimator/predictor) for a new patient. (See at least [0099] via: “… Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained..
Regarding claim 6 De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
wherein configuring the treatment prediction model to generate predictions from values for the first subset of predictive features that are most predictive comprises establishing a recommendation or a requirement that, after training, patient profiles for new patients cannot include missing values for the first subset of predictive features as a condition of using the treatment prediction model to generate treatment response predictions for the new patients. (See at least [0027] via: “… there are two steps to acquiring a set of reduced features: 1) the "raw" measured patient related clinical and/or laboratory data is processed to obtain a set of N.sub.c first-level features. These can be statistical quantities, can be transformed variables (e.g. Fourier coefficients at specific frequencies), and other types of coefficients that need some processing to extract. This step may be referred to as "feature extraction". 2) the next step is to reduce the N.sub.c first-level features to a smaller set containing only N.sub.i reduced features where N.sub.c>>N.sub.i. This step may be referred to as called feature reduction, feature compression, feature selection or dimensionality reduction..”; in addition see at least [0099] via: “… Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained…”; in addition see at least [0082] via: “…This subsystem could, optionally, identify key pieces of missing data and the communication and interface subsystem would send a prompt to the clinician requesting this further information. The expert system would then run its algorithm without this additional data then again if and when the missing data is provided. The data collection and data analysis results can be repeatedly calculated over several sessions until acceptable reliability is achieved or the medical expert system is not able to provide better results. This process is illustrated in FIG. 7…”)
Regarding claim 9: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
identifying that the patient profile for a first particular patient is missing a value for a first feature in the expanded set of predictive features included in the patient profile; and in response to identifying that the patient profile for the first particular patient is missing the value for the first feature, before using the patient profile for the first particular patient in training the treatment prediction model, imputing a value for the first feature. (See at least [0082] via: “…Another element is the central signal/information processing and central data management and analysis subsystem. This is the part of the medical expert system which processes the data, extracts the critical and useful information, provides a list of diagnostic possibilities and predicts the treatment options. See 105 and 110. In doing so, this subsystem uses the training data containing information about the therapies used and clinical response in patients previously treated for this condition (106), employing machine learning and inference methodologies to find a list of the best treatment options that can be sent to the clinician/user. Here, the data base analysis duplicates and supplements the clinical acumen of the clinician. This subsystem could, optionally, identify key pieces of missing data and the communication and interface subsystem would send a prompt to the clinician requesting this further information. The expert system would then run its algorithm without this additional data then again if and when the missing data is provided. The data collection and data analysis results can be repeatedly calculated over several sessions until acceptable reliability is achieved or the medical expert system is not able to provide better results. This process is illustrated in FIG. 7..”)
Regarding claim 27: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
wherein identifying the first subset of predictive features includes selecting a predetermined number of most predictive features in the expanded set of predictive features. (See at least [0098] via: “...After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “... Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained...”; in addition see at least [0100] via: “... the feature selection is based on a mutual information criterion. The features selected are those which exhibit the highest mutual information with the training set...’; in addition see at least [0102] via: “...the selected subset of features is the subset which is most useful for the estimator/predictor method. This criterion combines feature selection with the estimation/prediction method and the goal is to generate the best final result with the greatest possible efficiency...”)
Regarding claim 28: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
wherein identifying the first subset of predictive features includes determining which features in the expanded set of predictive features exceed a threshold level of predictive power. (See at least [0098] via: “...After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “... Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained...”; in addition see at least [0100] via: “... the feature selection is based on a mutual information criterion. The features selected are those which exhibit the highest mutual information with the training set...’; in addition see at least [0102] via: “...the selected subset of features is the subset which is most useful for the estimator/predictor method. This criterion combines feature selection with the estimation/prediction method and the goal is to generate the best final result with the greatest possible efficiency...”)
Regarding claim 29: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
wherein the shared predictive feature reduces a number of feature values to be derived when evaluating the plurality of treatment prediction models. (See at least [0098] via: “...After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “... Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained...”; in addition see at least [0100] via: “... the feature selection is based on a mutual information criterion. The features selected are those which exhibit the highest mutual information with the training set...’; in addition see at least [0102] via: “...the selected subset of features is the subset which is most useful for the estimator/predictor method. This criterion combines feature selection with the estimation/prediction method and the goal is to generate the best final result with the greatest possible efficiency...”)
Regarding claim 30: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin also teaches:
wherein the shared predictive feature reduces an amount of information collected from the new patient to generate treatment prediction responses using the plurality of treatment prediction models including the treatment response prediction using the treatment prediction model. (See at least [0098] via: “...After initial feature extraction, there is a feature reduction and feature selection phase in which only the most relevant features attributes are found and stored for further processing (146)...”; in addition see at least [0099] via: “... Feature reduction and feature selection are parts of the medical expert system of the present invention. The expert system uses feature reduction and feature selection to provide a smooth estimator/predictor with a good generalization capability. This procedure is based on the realization that not all features from the feature set contribute equally to the clustering of the data. The task of dimensionality reduction or feature selection provides for compact representations of high dimensional data. In a preferred embodiment of the method of the invention only those features which exhibit the strongest mutual information and/or correlation and/or relevance between the feature itself and the training set target values are retained...”; in addition see at least [0100] via: “... the feature selection is based on a mutual information criterion. The features selected are those which exhibit the highest mutual information with the training set...’; in addition see at least [0102] via: “...the selected subset of features is the subset which is most useful for the estimator/predictor method. This criterion combines feature selection with the estimation/prediction method and the goal is to generate the best final result with the greatest possible efficiency...”)
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 set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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
non-obviousness.
Claim 2 is rejected under 35 U.S.C. 103 as being un-patentable by De Bruin in view of Rubenstein et.al (WO 2017059022 A1) hereinafter “Rubenstein”
Regarding claim 2: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin is silent the following claim that is taught by Rubenstein:
wherein the treatment prediction model is a random forest ensemble comprising a plurality of decision trees. (See at least [0002] via: “…methods for predicting treatment-regimen-related outcomes using predictive models. The present disclosure describes a combination of machine learning prediction and patient preference assessment to enable informed consent and precise treatment decisions..”; in addition see at least [0047] via: “…the random forests (RF) algorithm is a tree-based method built on an ensemble of trees. A predictive model generated based on the RF algorithm does the following process many times: selects a bootstrap sample of the training dataset and builds a tree on the bootstrap sample. Within each tree, a randomly selected number of predictors are chosen and the optimal split is selected only from that sample. One or more tuning parameters for predictive model generated based on the RF algorithm include the number of randomly selected predictors for each split. Building an ensemble of trees in this way reduces the variance seen by using just a single tree.. “)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Rubenstein. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose using a random forest algorithm built on an ensemble of trees as a predicting model as taught by Rubenstein. Combining De Brun and Rubenstein is helpful in offering advantages to De Brun’s prediction model including higher accuracy, higher robustness against overfitting, and the ability to handle large datasets and missing values, making it a more powerful prediction model.
Claims 3-5 is rejected under 35 U.S.C. 103 as being un-patentable by De Bruin in view of Goetzke et.al (US 20020123670 A1) hereinafter “Goetzke”
Regarding claim 3: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. De Bruin is silent the following claim that is taught by Goetzke:
wherein the particular medical treatment is a treatment for alleviating chronic pain. (See at least [0057] via: “… potential chronic pain patients that can be efficiently treated. Some examples of categorization preferences for a primary care physician can include potential chronic pain patients that are suitable for treatment by the primary care physician and potential chronic pain patients that should be considered for referral to a specialist. Some examples of categorization preferences for a specialist physician can include potential chronic pain patients that are suitable for treatment by the specialist physician and potential chronic pain patients that should be considered for referral to a primary care physician...”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Goetzke. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose treatments for chronic pain patients as taught by Goetzke. Combining De Brun and Goetzke is helpful in offering treatment to a wider variety of patients including those that suffer from chronic pain.
Regarding claim 4: De Bruin teaches the invention as claimed and detailed above with respect to claim 1 and Goetzke teaches the invention as claimed and detailed above with respect to claim 3. De Bruin is silent the following claim that is taught by Goetzke:
wherein the treatment for alleviating chronic pain is one of a multimodal treatment, a class of medication, a particular medication, a class of injection, a particular type of injection, an implanted medical device, a behavioral medicine treatment, an integrative medicine treatment, a rehabilitation therapy, or an orthotic or assistive device. (See at least [0068] via: “…The combination of the lumbar diagnosis, the axial nature of the pain, the impact of the pain on the muskuloskeletal system, and the description of the pain as dull and throbbing in nature, leads to a low back diagnosis. In addition, a review of the pharmaceutical claims data establishing that Patient A received a prescription for an Opiate (Percocet, 8 per day) and a Nonsteroidal (Celecoxib, 4 caps per day), both for ≥ 91 days, establishes the chronicity of the back pain…”; in addition see at least [0074] via: “…The medical record also indicates that within the past 120 days from the last day of service patient received a trigger point injection. The record does not establish a pattern (≥91 days) of chronic use of trigger point injections. The pharmaceutical claims data indicates that Patient D received a prescription for a short acting opiate (Tylenol 3), and Dantrium (a muscle relaxant) both for less than 91 days within the past 120 days from the last day of service…”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Goetzke. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose treatments for chronic pain patients as taught by Goetzke. Combining De Brun and Goetzke is helpful in offering treatment to a wider variety of patients including those that suffer from chronic pain.
Regarding claim 5: De Bruin teaches the invention as claimed and detailed above with respect to claim 1 and Goetzke teaches the invention as claimed and detailed above with respect to claim 3. De Bruin also teaches:
Charlson co-morbidity index, post-traumatic stress disorder (PTSD), pain experience duration, opioid use or misuse, neuropathic pain level, Medicaid status, socioeconomic factors (demographic) in region of patient's residence, medical diagnoses (physician's .. diagnosis), medication prescriptions, medical procedure history (administration of a medication), body pain map, patient age (age), patient sex (sex), patient education level, tobacco use, alcohol use, illicit drug use, anxiety level, depression level (clinical depression), global mental health level (cognitive functioning), global physical health level, pain interference, positive outlook level, or sleep disturbance level. (See at least [0069] via: “… As shown in FIG. 1, in one embodiment, the "medical digital expert system" operates in the following manner. The physician, or an assistant, collects as much relevant biometric, demographic, neurological, psychological, psychiatric, laboratory and clinical information (101) as possible regarding the patient. This information could include an array of neuro-psycho-biological indicators such as demographic information, past history, symptomatic presentation, a list of medical co-morbidities, results of laboratory testing, selected measures of personality and cognitive functioning, pharmacogenetic data, and biological data derived from electrophysiological, magnetic, electromagnetic, radiological, optical, infra-red, ultrasonic, acoustic, biochemical, medical imaging and other investigative procedures and attributes. The physician's presumptive diagnosis is also provided if available…”; in addition see at least [0078] via: “…Ideally, the predictive accuracy of the medical digital expert system is optimal when the available neuro-psycho-biological data for a given test patient is maximized. However, in practice, because of time, cost, accessibility or other factors, patients do not receive every possible investigation and test. Therefore the disclosed system is designed to flexibly operate with incomplete data, provided that required minimum data requirements have been met (e.g. age, sex and EEG data in psychiatric illnesses and disorders). The set of available data and attributes for each patient is analyzed by the expert system, and the treatment response prediction, and optionally, the diagnostic estimation result, will be sent to the physician in electronic format. As an example, for a suspected mood disorder, in one of its simplest routines, a set of EEG data and a selected set of clinical depression rating scales are recorded and entered into the medical digital expert system..”; in addition see at least [0129] via: “…Collection of clinical and laboratory data and other measured attributes of body and brain functioning can be done in three modes: 1) in the normal and resting-awake state, 2) during stimulus-driven, mood-induced, event-related or activity-evoked states, 3) during sleep, or some other altered level of unconsciousness e.g. general anesthesia or, 4) following administration of a medication…”)
Claim 10 is rejected under 35 U.S.C. 103 as being un-patentable by De Bruin, in further view of Jiang et.al (WO 2004053659 A2) hereinafter “Jiang”
Regarding claim 10: De Bruin teaches the invention as claimed and detailed above with respect to claim 1&9. However De Bruin is silent the following claim that is taught by Jiang:
wherein imputing the value for the first feature comprises: if the first feature is a continuous variable, assigning an average value of the first feature from other patient profiles as the value for the first feature in the patient profile for the first particular patient; or if the first feature is a discrete variable, assigning a null value as the value for the first feature in the patient profile for the first particular patient. (See at least [0012] via: “… the method and system of the invention scans an entire data set and performs the following tasks: automatically distinguishes between continuous and categorical variables; automatically handles problem data, such as missing values ..”; in addition see at least [0043] via: “…It is a rare and fortuitous occasion when a data set exhibits no missing values. More often missing values are encountered for many if not all the fields or variables in a data set. Some fields may have only a few missing values, while for others more than half of the values may be missing. In one embodiment, the method of the invention deals with missing values in one of two ways, depending on whether the field is continuous or categorical. For continuous variables, the method substitutes for the missing values the mean value computed from the non-missing entries and reports the number of substitutions for each field. For categorical variables, the invention creates a new category that effectively labels the cases as "missing." In many applications, the fact that certain information is missing can be used profitably in model building and the invention can exploit this information. In one embodiment, in the case of incomplete datasets, the missing counts of the severely missing observations are presented to the user (perhaps in a rank order format). She or he then has the option to either eliminate those observations or variables from the design matrix or substitute corresponding mean values in their place..”) For the case of discrete variables (or categorical variables) the examiner interprets “ the invention creates a new category that effectively labels the cases as "missing."” Equivalent to assigning a null value
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Jiang. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose distinguishing between continuous and categorical variables in a data set and if there are some missing values replace with mean in case of the former or replace with a label in case of the latter. Combining De Brun and Jiang is helpful in providing continuity to parts of data set that has missing values so that the output of the predictive model is not impacted due to missing values in the input data set that has missing values.
Claims 11-12 are rejected under 35 U.S.C. 103 as being un-patentable by De Bruin, in view of Heldt et.al (US 20170042483 A1) hereinafter “Heldt”, and in further view of Burnett et.al (US 20210241871 A1) hereinafter “Burnett”:
Regarding claim 11: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. However De Bruin is silent the following claim that is taught by Heldt:
obtaining, by the system, updated patient profiles that include data related to additional patients, additional predictive features, or both, which were not in the patient profiles on which the treatment prediction model was previously trained; (See at least [0035] via: “…In some embodiments, the set of training input data may contain gaps in some segments. For example, for some patients in the training data, the demographic information may be incomplete or may be implausible (an 800 year old man, for example). In that case, the training input data may also include synthetic data…”)
identifying, by the system and based on the updated patient profiles, a third subset of predictive features that are most predictive of whether a given patient will respond to the particular medical treatment according to the one or more criteria; (See at least [0035] via: “…the training data may include solely synthetic data if the users associated with the training stage may wish to train the classifiers on a very specific set of characteristics and correlations that are not available in historic data of the first group of patients. For example, the user associated with the training stage may be a clinician (such as a doctor or a nurse) or a researcher at a hospital. In the process of importing the training input data, the user may identify gaps in the training data and might want to add some synthetic training data to fill these gaps…”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Heldt. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose adding data not available in the incoming historical data set used by the predictive model as taught by Heldt. Combining De Brun and Heldt is helpful for example to a clinician such as a doctor or nurse that has identified some gaps in the training data and might want to add some synthetic training data to fill some important gaps that are missing in the incoming historical data set.
However De Bruin and Heldt are silent the following limitation that is taught by Burnett:
re-configuring the treatment prediction model to generate predictions from values for the third subset of predictive features that are most predictive. (See at least [0016] via: “… the method includes periodically retraining the risk machine-learning system as new patient data, …, is collected. In some embodiments, the method includes, before receiving patient data for a plurality of patients, receiving historical patient data from at least one healthcare recording database. The method includes extracting training data from the historical patient data, utilizing the training data to train multiple risk machine-learning systems, selecting a risk machine-learning system of the multiple risk machine-learning systems that respectively determined risk scores above a predetermined accuracy rate, and storing the risk machine-learning system…”; in addition see at least [0069] via: “…the data selecting module 218 receives feedback from the machine-learning training module 220 that is used to adjust the historical patient data (e.g., adjust the number of patients and/or time period in which patients are included). In some embodiments, the data selecting module 216 adjusts the predetermined time period and/or the predetermined number of patients for the historical patient data such that a trained risk model meets a minimum accuracy rate. For instance, a trained risk model, as explained in detail below, may have an accuracy rate below the minimum accuracy rate and the machine-learning training system 300 can adjusts the predetermined time period and/or the predetermined number of patients for the historical patient data to retrain a risk model with an accurate rate above the minimum accuracy rate. The accuracy rate for a trained risk model is determined by machine-learning training module 220 ..”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun and Heldt with Burnett. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun and Heldt fail to disclose retraining the prediction model based on an adjusted subset of the original patient feature data as taught by Burnett. Combining De Brun and Burnett is helpful in increasing the accuracy of the prediction model.
Regarding claim 12: De Bruin teaches the invention as claimed and detailed above with respect to claim 1 and Heldt and Burnett teach the invention as claimed and detailed above with respect to claim 11. However, De Bruin and Heldt are silent the following claim that is taught by Burnett:
wherein the third subset of predictive features includes at least one predictive feature that is not among the predictive features in the first subset; and re-configuring the treatment prediction model comprises adding a recommendation or requirement that patient profiles for new patients cannot include missing values for the at least one predictive feature as a condition of using the treatment prediction model to generate treatment response predictions for the new patients. (See at least [0035] via: “…the training data may include solely synthetic data if the users associated with the training stage may wish to train the classifiers on a very specific set of characteristics and correlations that are not available in historic data of the first group of patients. For example, the user associated with the training stage may be a clinician (such as a doctor or a nurse) or a researcher at a hospital. In the process of importing the training input data, the user may identify gaps in the training data and might want to add some synthetic training data to fill these gaps…”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun and Burnett with Heldt. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun and Burnett fails to disclose adding data not available in the incoming historical data set used by the predictive model as taught by Heldt. Combining De Brun, Burnett and Heldt is helpful for example to a clinician such as a doctor or nurse that has identified some gaps in the training data and might want to add some synthetic training data to fill some important gaps that are missing in the incoming historical data set.
Claim 14 is rejected under 35 U.S.C. 103 as being un-patentable by De Bruin, in view of Klaerner et.al (WO 2019236636 A1) hereinafter “Klaerner”:
Regarding claim 14: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. However De bruin is silent the following claim that is taught by Klaerner:
wherein the one or more criteria comprise achieving at least one of:
(i) a threshold improvement in average pain intensity within a predetermined time interval;
(ii) a threshold improvement in physical function within a predetermined time interval; (See at least [001234] via: “…A method of improving the physical function of a patient afflicted with metabolic acidosis disease, the method comprising bicarbonate concentration by at least 1 mEq/L; (b) results in a sustained serum bicarbonate increase of at least 1 mEq/L over a period of at least twelve weeks; and (c) is sufficient to improve the physical function score of the patient compared to a placebo control group at the end of the period, wherein the improvement in the physical function score is statistically significant…”; in addition see at least [001235] via: “… Embodiment 861. A pharmaceutical composition for improving the physical function score of a human patient afflicted with chronic kidney disease and an acid-base disorder, the patient having a baseline serum bicarbonate level of <22 mEq/L prior to treatment, the composition being a nonabsorbable composition having the capacity to: (a) remove a target species from the patient selected from the group consisting of protons, strong acids, and conjugate bases of strong acids; and (b) improve the patient’s physical function score based on answers to question 3 of the Kidney Disease Quality of Life Short Form (KDQOL-SF) compared to a placebo control in a statistically significant manner at the end of at least a twelve-week period..”) or
(iii) a threshold improvement in patient's overall impression of change
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Klaerner. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose improving the physical function score of a patient wthin a given time period as taught by Klaerner. Combining De Brun and Klaerner is helpful in “improve the physical function score of the patient compared to a placebo control group at the end of the period, wherein the improvement in the physical function score is statistically significant [Klaerner 001234].
Claim 15 is rejected under 35 U.S.C. 103 as being un-patentable by De Bruin in view of Hagstrom et.al (WO 2012061821 A1) hereinafter “Hagstrom”
Regarding claim 15: De Bruin teaches the invention as claimed and detailed above with respect to claim 1. However, De Bruin is silent the following claim that is taught by Hagstrom:
wherein training the treatment prediction model comprises achieving at least an area under receiver operating curve (AUROC) or selective area under receiver operating curve (SAUROC) score of 0.65. (See at least [0067] via: “…FIG. 1 depicts a list of two-biomarker (TWOMRK) sets or panels. Predictive models were built, as described in certain embodiments of the present teachings, to estimate the rate of structural damage in subjects. Biomarker concentrations obtained at one timepoint were used to predict the change of total Van der Heijde Sharp Scores (DSS) that would occur over the next 12 months. The combinations of different biomarkers that provided model performance with Area under the ROC Curve (AUROC) of 0.6 or greater in 100 cross validations are reported in this figure…”; in addition see at least [0068] via: “…The detailed procedure for deriving TWOMRK sets is as follows. For all possible models with two-biomarker combinations, 70/30 cross validation performance was computed, as measured by AUROC. 70/30 means repeatedly training in a randomly selected 70% of the data, and testing in the remaining 30%. Because each randomly selected test set has different ranges of DSS, to ensure balanced groups, a median DSS threshold was used. The two-biomarker combinations (TWOMRK sets) with AUROC >= 0.6 are reported in this FIG. 1. This process was repeated for all combinations of 3, 4, 5, and 6 biomarkers (FIGS. 3- 6, respectively). To avoid redundancy, the n-biomarker combinations list contain only those marker sets with AUROC >=0.6, and do not contain any previously reported combination. For example, FIG. 3 describes 4-biomarker sets (FOURMRK), and does not list any set of 2 or 3 biomarkers that are already found in a TWOMRK or THREEMRK set…”; in addition see at least [0155] via: “…Use of statistical values such as the area under the curve (AUC), and specifically the AUC as it pertains to the area under the receiver operating characteristic (ROC) curve (AUROC curve), encompassing all potential threshold or cut-off point values, is generally used to quantify predictive model performance. As is known in the art, the ROC curve is a graphical plot of the sensitivity, or true positives, versus (1 - specificity), or false positives, for a binary (yes/no) classifier as its discrimination threshold is varied. Acceptable degrees of accuracy can be defined accordingly. In certain embodiments of the present teachings, for example, an acceptable degree of accuracy for a binary classifier predictive model can be one in which the AUROC curve is 0.60 or higher…”)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the invention to modify De Brun with Hagstrom. De Brun teaches A medical digital expert system to predict a patient's response to a variety of treatments (using pre-treatment information). The system utilizes data fusion, advanced signal/information processing and machine learning/inference methodologies and technologies to integrate and explore diverse sets of attributes, parameters and information that are available to select the optimal treatment choice for an individual or for a subset of individuals suffering from any illness or disease including psychiatric, mental or neurological disorders and illnesses. The methodology and system can also be used to determine or confirm medical diagnosis, estimate the level, index, severity or critical medical parameters of the illness or condition, or provide a list of likely diagnoses for an individual suffering/experiencing any illness, disorder or condition. However, De Brun fails to disclose providing model performance with Area under the ROC Curve (AUROC of 0.6 or greater as taught by Hagstrom. Combining De Brun and Hagstrom is helpful since it is “used to quantify predictive model performance” [Hagstrom 0155].
Prior Art Made of Record
The prior art made of record and not relied upon is considered pertinent to Applicant's disclosure, and is listed in the attached form PTO-892 (Notice of References Cited). Unless expressly noted otherwise by the Examiner, all documents listed on form PTO-892 are cited in their entirety.
Tucker (US 20070122824 A1) – Method And Kit For Assessing A Patient's Genetic Information, Lifestyle And Environment Conditions, And Providing A Tailored Therapeutic Regime- teaches: determining a personalised therapeutic regime, comprising: receiving genetic information relating to a patient; determining genetic criteria relevant to a personalised therapeutic regime for the patient using the genetic information; receiving personal information relating to the patient; determining personal criteria relevant to the personalised therapeutic regime using the personal information; and combining the genetic criteria and the personal criteria to determine the personalised therapeutic regime for the patient.
Wei (US 20040193019 A1) - Methods For Predicting An Individual's Clinical Treatment Outcome From Sampling A Group Of Patient's Biological Profiles – teaches: predict an individual's treatment outcome from a sampling of a group of patients' biological profiles. Biological profile information is received from patients who had a medical condition and who received a treatment. Treatment outcome information regarding the patients who had the medical condition and who received the treatment is also received. A discriminant analysis-based pattern recognition process is then performed on the biological profile information and the treatment outcome information, thereby generating a model that correlates between the biological profile information and the treatment outcome information. The model can be used for, among other things, predicting treatment outcome for the new patient for the treatment.
Response to Arguments
Applicant's arguments filed 1/28/2026 have been fully considered but they are not persuasive.
Applicant amended independent claims 1, 26; added dependent claims 27-30 and cancelled claims 7, 8, 13, 16-25 as posted in the above analysis with additions underlined.
In response to applicant's arguments regarding claim rejection under 35 U.S.C § 101:
Several steps are taken in the analysis as to whether an invention is rejected under 101. The first step is to determine if the claim falls within a statutory category. In this case it does for claims 1 and 26 since the claims recite a method and non-transitory computer-readable storage medium of “evaluating patient treatment predictions”.
The second step under 2A prong one is to determine if the claims recite an abstract idea, which would be the case if the invention can be grouped as either: a) mathematical concepts; (b) mental processes; or (c) certain methods of organizing human activity (encompassing (i) fundamental economic principles, (ii) commercial or legal interactions or (iii) managing personal behavior or relationships or interactions between people). The current invention is classified as an abstract idea since it may be grouped as a mental process as it recites “evaluating patient treatment predictions”. Alternatively the invention is classified as an abstract idea since it may be grouped as certain methods of organizing human activity under managing personal relationships or interactions between people as it recites “evaluating patient treatment predictions”
The third step under 2A Prong Two is to determine if additional elements in the claim imposes a meaningful limit on the abstract idea in order to integrate it into a practical idea. The current invention does not represent a practical idea since the additional elements amount to mere instructions to implement an abstract idea on a computer, or merely use a generic computer as a tool to implement the abstract idea.
the fourth step under 2B is to determine if additional elements of the claim provide an inventive concept. An invention may be classified as an inventive concept if a computer-implemented processes is determined to be significantly more than an abstract idea (and thus eligible), where generic computer components are able in combination to perform functions that are not merely generic, and non-conventional even if generic computer operations on a generic computing device is used to implement the abstract idea
Regarding Step 2A Prong One:
The Applicant argues that the 35 U.S.C § 101 rejection is moot based on the amendments of the claims. The Applicant argues that the claims do not recite an abstract idea. Accordingly, the Applicant requests that the 35 U.S.C § 101 rejection be withdrawn. The Applicant claims that the amendment to the independent claims 1 & 26 recites a specific technique for improving the performance of a computing system by including a shared predictive feature that is utilized across a plurality of treatment prediction models corresponding to different medical treatment modalities.
The Examiner disagrees with the Applicant’s argument since the Applicant’s argument is not persuasive. The amendment for claims 1 & 26 belong to an abstract idea.
The method to select the abstract idea is to strip the additional elements from the claims. As seen below the recited boldened words constitute the abstract idea after stripping the un-boldened additional elements of amended limitation of claims 1 and 26:
obtaining, by a system comprising one or more computers, patient profiles for a plurality of patients, wherein each patient profile corresponds to a particular patient and includes (i) values for an expanded set of predictive features about the particular patient and (ii) a target response classification that indicates whether the particular patient responded to a particular medical treatment according to one or more criteria;
training, by the system, a treatment prediction model using the patient profiles, including applying a machine-learning technique that causes the treatment prediction model to learn to predict, based on predictive features from a patient profile for a given patient, a likelihood that the given patient will respond to the particular medical treatment according to the one or more criteria;
identifying, by the system, a first subset of predictive features from the expanded set of predictive features that are most predictive of whether a given patient will respond to the particular medical treatment according to the one or more criteria,
the first subset of predictive features including at least one predictive feature defining a personal attribute of the patient's mental health, physical health, or socioeconomic status;
wherein at least one predictive feature in the first subset of predictive features is a shared predictive feature that is utilized across a plurality of treatment prediction models corresponding to different medical treatment modalities, the shared predictive feature being identified across the plurality of treatment prediction models;
configuring the treatment prediction model to generate predictions from values for the first subset of predictive features that are most predictive, without requiring values for a second subset of predictive features that are less predictive than features from the first subset;
wherein the configuring includes, after initially training the treatment prediction model on the expanded set of predictive features, re-training the treatment prediction model only on the first subset of predictive features to exclusion of the second subset of predictive features;
wherein at least one of a size or a computational complexity of the treatment prediction model is reduced as a result of re-training the treatment prediction model; and
applying the treatment prediction model to generate a treatment response prediction for a new patient.
The selected abstract idea (boldened limitations) of claims 1 and 26 can be implemented by pencil and paper and thus belong to the grouping of mental processes under concepts performed in the human mind (including an observation, evaluation, judgement, opinion) as it recites “evaluating patient treatment predictions”. (refer to MPP 2106.04(a)(2)). Alternatively the selected abstract idea belongs to certain methods of organizing human activity under managing personal relationships or interactions between people as it recites “evaluating patient treatment predictions”. (refer to MPP 2106.04(a)(2)). Accordingly independent claims 1 and 26 recite an abstract idea.
Regarding Step 2A Prong Two.
The Applicant argues that even if claims 1 and 26 are directed to an abstract idea under Step 2A Prong One, claims 1 and 26 as a whole integrates any such abstract idea into a practical application. The Applicant further argues that the claim as a whole provides a particular technological improvement by reducing the size and/or computational complexity of a treatment prediction model on a computing system by re-training the model in the specific manner as recited by the amendments of claim 1 and 26. In other words, claim 1 and 26 describes a technological improvement that allows a treatment prediction model to achieve specific technical advantages including smaller size and/or computational complexity as compared to a prior model.
The Examiner disagrees with the Applicant’s argument since the Applicant’s argument is not persuasive. The Applicant is interpreting practical application of the claims colloquially. The analysis is looking for Improvements to the functioning of a computer, or to any other technology or technical field which the claims do not demonstrate. Neither claims 1 or 26 recite additional elements that impose a meaningful limit on the abstract idea:
Claims 1 and 26 recite the following additional elements:
system comprising one or more computers;
applying a machine-learning technique
The additional elements as recited above amount to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, the claim as a whole does not integrate the abstract idea into a practical application because it does not impose any meaningful limits on practicing the abstract idea.
In order to integrate the abstract idea into a practical idea the Applicant could demonstrate at least one of the conditions enumerated below applies:
Improvements to the functioning of a computer, or to any other technology or technical field - see MPEP 2106.05(a)
Applying the judicial exception with, or by use of, a particular machine - see MPEP 2106.05(b)
Effecting a transformation or reduction of a particular article to a different state or thing - see MPEP 2106.05(c)
Applying or using the judicial exception in some other meaningful way beyond generally linking the use of the judicial exception to a particular technological environment, such that the claim as a whole is more than a drafting effort designed to monopolize the exception - see MPEP 2106.05(e) and Vanda Memo
The Applicant has not demonstrated any of the above listed conditions.
Regarding Step 2B
Similar to the analysis under Step 2A Prong Two, the additional elements amount to mere instructions to implement an abstract idea on a computer, or merely use a computer as a tool to implement the abstract idea. (refer to MPEP 2106.05(f)). Accordingly, the claim does not provide an inventive concept (significantly more than the abstract idea) and hence the claim is ineligible.
In order evaluate whether the claim recites additional elements that amount to an inventive concept what could be shown is:
Adding a specific limitation (unconventional other than what is well-understood, routine, conventional (WURC) activity in the field - see MPEP 2106.05(d)
The Applicant has not demonstrated the above listed condition.
In response to applicant's arguments regarding claim rejection under 35 U.S.C § 103.
The Applicant argues that the cited references fail to teach the following amendment to claims 1 & 26:
wherein at least one predictive feature in the first subset of predictive features is a shared predictive feature that is utilized across a plurality of treatment prediction models corresponding to different medical treatment modalities, the shared predictive feature being identified across the plurality of treatment prediction models.
Accordingly the Applicant requests that the 35 USC §103 rejection be withdrawn.
The Examiner disagrees with the Applicant since the Applicant’s arguments are not persuasive, since De Bruin teaches the above limitation.
The limitation as listed above is taught by De Bruin paragraphs: [0098], [0099], [0100].[0102].
Furthermore added claims 27, 28, 29 and 30 are also taught by De Bruin paragraphs: [0098], [0099], [0100].[0102]
As a result the rejection of independent claims based on 35 U.S.C § 102/103 is maintained.
In conclusion for reasons of record and as set forth above, the Examiner maintains the rejection of claims 1-6, 9-12, 14-15, 26-30 as being directed to a judicial exception without significantly more, and thereby being directed to non-statutory subject matter under 35 USC §101. In addition the Examiner maintains the 35 USC §103 rejection of claims 1-6, 9-12, 14-15, 26-30 based on prior art. In reaching this decision, the Examiner considered all evidence presented and all arguments actually made by Applicant.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to PIERRE L MACCAGNO whose telephone number is (571)270-5408. The examiner can normally be reached M-F 8:00 to 5:00.
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, Mamon Obeid can be reached at (571)270-1813. 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.
/PIERRE L MACCAGNO/Examiner, Art Unit 3687
/MAMON OBEID/ Supervisory Patent Examiner, Art Unit 3687
Prosecution Insights