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 .
Status
Claims 1, 9, 12, 13 and 18 are amended, claims 2-8, 14-17 and 19-20 remain pending. All amended claims are considered below.
Respond to Arguments
Applicant’s arguments, see pages 8–11, filed in 11/25/2025, with respect to amended Claims 1–20 have been fully considered and are not persuasive.
Subject Matter Eligibility Applicant Arguments
The Applicant argues that the amended claim is directed to a specific improvement in technology related to diagnosis and treatment of a medical disorder named sleep apnea, not to an abstract idea. The Applicant contends that the claimed process uses machine learning technology to analyze physiological data for predicting sleep apnea severity and that this constitutes a concrete technical improvement rather than an abstract concept. The Applicant provides no documentary evidence, declaration, or technical comparison to support this characterization.
The Examiner respectfully disagrees and sustains the rejection because the Applicant's characterization does not match the actual claim language under its broadest reasonable interpretation. Stating that a claim is directed to a specific improvement in technology is a conclusion, not an argument supported by evidence. Under MPEP § 2106.05(a), a claim reflects a technological improvement only when it recites a specific technical mechanism that produces a measurable change in how the computer or technology functions. The amended claims recite collecting data, extracting features, performing arithmetic to generate compound features, predicting a score with a machine-learning model, comparing the score to a threshold, and recommending a breathing device. Each step describes what the process accomplishes, not how the technology itself is improved. The specification confirms: Any supervised machine learning model can be deployed such as Random Forest, Support Vector Machine, AdaBoost, and Regression models (Spec., para. 0008). A process that uses any interchangeable model to reach a medical recommendation does not improve the technology; it uses existing technology as a tool to perform an abstract analytical process.
The Applicant argues that the claim recites a specific technical solution that improves how computer systems generate an action strategy and automatically selects a device among breathing devices with different pressurized air-delivery characteristics, suited to subject's needs and preferences, to treat the sleep apnea. The Applicant characterizes this as a technical device-selection mechanism that uses machine-learning derived intensity scores to automatically choose a device with specific air-delivery characteristics. The Applicant provides no evidence beyond this characterization.
The Examiner respectfully disagrees and sustains the rejection because the Applicant argues features not found in the claims. The amended claim 1 recites: recommending a breathing device to automatically adjust breathing of the subject. The claim does not recite selecting from a plurality of breathing devices, does not recite different pressurized air-delivery characteristics, and does not recite suited to subject's needs and preferences. Under MPEP § 2111.01(II), limitations from the specification are not imported into the claims. The claim, as written, recommends one type of device. It does not select among multiple devices. It does not evaluate air-delivery characteristics. It does not match a device to a patient's preferences. Because the Applicant's entire argument rests on language absent from the claims, the argument addresses unclaimed features and even if the claims did recite these features, they would be part of the abstract idea itself. Picking out an appropriate treatment device customized to the patient’s condition is part of a treatment planning process that a clinician could perform by evaluating data and making judgements and would not amount to a technical improvement is not persuasive.
The Applicant argues that under Step 2A, Prong 1, the claim is not directed to an abstract concept but to a concrete technical process that cannot be performed mentally or with pen and paper. The Applicant contends that the amendments transform the claim from medical judgment to a machine-oriented control process. The evidence offered is the Applicant's own characterization of the claim language.
The Examiner respectfully disagrees and sustains the rejection because each claim step, read under its broadest reasonable interpretation, describes an activity a person can perform mentally or with pen and paper. MPEP § 2106.04(a)(2)(III) defines mental processes as concepts performed in the human mind, including observations, evaluations, judgments, and opinions. Walking through each claim limitation: a clinician gathers patient records (collecting a dataset), identifies health indicators like age and BMI (extracting a set of features), multiplies oxygen levels by age group on paper (generating a compound feature), estimates a severity score based on clinical experience (predicting an intensity score), compares the score against a clinical guideline number (determining exceedance of the intensity score above the pre-defined threshold), and writes in the patient chart that the patient needs an auto-adjusting pressure machine (recommending a breathing device to automatically adjust breathing). The recitation of a machine-learning model does not change this analysis because the model is the tool performing the mental process, not a specific technical mechanism that transforms the process. MPEP § 2106.04(a)(2)(III)(C) explains that a claim recites a mental process when it encompasses a human performing the step with a generic computer merely used as a tool.
The Applicant argues that the amended claim recites a technological control action, specifically automatically selecting, from a plurality of breathing devices having different pressurized air-delivery characteristics, a device to treat the sleep apnea. The Applicant states this is a technical device-selection mechanism and that the device characteristics are physical, machine-operational parameters and cannot be regarded as abstract concepts. The Applicant further states this converts the process into a technological process that improves a physical therapeutic outcome.
The Examiner respectfully disagrees and sustains the rejection because the claim does not recite the features the Applicant describes. The Applicant quotes language that does not appear in any claim. Amended claim 1 recites: recommending a breathing device to automatically adjust breathing of the subject. The phrase automatically selecting, from a plurality of breathing devices having different pressurized air-delivery characteristics is not in the claims and even if the claims did recite selecting a device, that would be part of the abstract idea because a person can make a judgement to select which device would be most appropriate for a patient. Under MPEP § 2111, patent rights are defined by the claim language, not by the Applicant's description of what the claim supposedly recites. The claim produces a recommendation. It does not select among multiple devices. It does not evaluate pressurized air-delivery characteristics. It does not control or interact with any physical device. The specification confirms this reading: Based on the intensity score or classification category, predicted by the machine-learning model, appropriate remedial action strategies are recommended to the user (Spec., para. 0032). The word recommended appears throughout the specification. The word select in the context of choosing among multiple breathing devices does not. A recommendation is an opinion, and forming an opinion is a mental process under MPEP § 2106.04(a)(2)(III).
The Applicant argues that the newly amended features transform the claim from a 'medical judgment' to a machine-oriented control process. The Applicant contends the claim requires machine-implemented threshold computation and triggers automated device selection, which is based on air-delivery characteristics. The Applicant concludes that these characteristics correspond to distinct mechanical operations that cannot be evaluated by a human mind. No documentary evidence accompanies these assertions.
The Examiner respectfully disagrees and sustains the rejection because the claim recites a threshold comparison and a recommendation, both of which a person can perform. The phrase air-delivery characteristics does not appear in any claim. It appears only in the Applicant's remarks. Under MPEP § 2111.01(II), that language cannot be read into the claims. As for the assertion that the process cannot be evaluated by a human mind, the specification itself describes the threshold mapping using the AHI index with numerical cutoffs of 5, 15, and 30 (Spec., para. 0044). A person compares a number against 5, 15, or 30 every day in clinical practice. The machine-implemented label does not change the nature of the underlying operation; it describes the tool used to perform it, not a transformation of the process itself. MPEP § 2106.04(a)(2)(III)(C) addresses this: a claim that encompasses performing the step mentally but adds only a generic computer as a tool still recites a mental process.
The Applicant argues that the claim maps ML-predicted intensity scores into sleep apnea categories using thresholds and uses that mapping to trigger device selection, where devices differ in air-delivery pressure mechanics. The Applicant concludes this leads to an improvement in the technical field of sleep-therapy device selection systems and is an improvement in device-control technology, not merely automation of a human process. No evidence beyond attorney argument supports this conclusion.
The Examiner respectfully disagrees and sustains the rejection because the claim does not recite controlling any device, and the argued improvement rests on unclaimed features.
The claim recites recommending a breathing device and outputting a result that represents the action strategy. The result is displayed on a graphical user interface of a health care provider. The claim outputs information to a human on a screen. It does not control, configure, activate, or connect to any breathing device. Under MPEP § 2106.05(a), an improvement to technology must be reflected in the claim language as a specific technical mechanism, not stated only in the remarks.
The Applicant argues that under Step 2A, Prong 2, even if any abstract concept were present, the claim integrates it into a practical application that improves the functioning of computer systems themselves. The Applicant states the method results in improved accuracy and interpretability of sleep apnea detection systems. No comparative data, benchmark results, or declaration accompanies this assertion.
The Examiner respectfully disagrees and sustains the rejection because the claim does not recite any mechanism that improves how a computer functions. Under MPEP § 2106.05(a), integration into a practical application through an improvement to technology requires a specific technical solution reflected in the claim language itself. The claim recites a generic computing environment (computer-implemented method), a generic model (machine-learning model), and a generic display (graphical user interface). None of these elements recites a specific improvement to how computers process data, store information, or communicate.
The Applicant argues that traditional methods to diagnose obstructive sleep apnea require a clinician to review polysomnography results and that the laboratories equipped to conduct polysomnography tests are limited and not easily accessible. The Applicant concludes that the claimed invention provides techniques to use machine-learning models to predict obstructive sleep apnea without using the results of a polysomnography diagnostic test, framing this as a practical application.
The Examiner respectfully disagrees and sustains the rejection because eliminating a diagnostic test describes the problem the abstract idea solves, not a technological improvement in how the claim works. Under MPEP § 2106.05(a), the relevant question is whether the claim language recites a specific technical mechanism that improves the functioning of the computer or another technology. Bypassing polysomnography describes a goal or benefit of the abstract analytical process. A clinician who reviews electronic health records, performs arithmetic on patient data, estimates a severity score, and writes a device recommendation in the chart also bypasses polysomnography. The abstract analytical steps remain the same whether a computer or a clinician performs them. The improvement the Applicant describes is in the outcome of the analysis (avoiding a lab test), not in the technology used to perform the analysis.
The Applicant argues that the amended claimed automates this entire technical workflow by generating compound physiological features, producing an ML-based intensity score, mapping the score to threshold-based apnea categories, and automatically selecting the appropriate breathing device. The Applicant characterizes this as a concrete technological improvement to respiratory assist systems. No technical evidence or comparison to prior systems accompanies this assertion.
The Examiner respectfully disagrees and sustains the rejection because automating an abstract process on a generic computer does not transform the abstract idea into a technological improvement.
The claim does not recite automatically selecting the appropriate breathing device. It recites recommending a breathing device to automatically adjust breathing of the subject. Under MPEP § 2111, the claim produces a recommendation for a type of device; it does not select a specific device from multiple options. Automating an abstract analytical process using generic computing tools does not integrate the exception into a practical application. The specification describes the computing environment as general purpose computers such as personal computers and laptops, workstation computers (Spec., para. 0080) and the model as any supervised machine-learning model (Spec., para. 0057). Running a mental process faster on general-purpose hardware does not change the character of the process. The claim does not improve any respiratory assist system because it does not interact with, configure, or control any such system. It displays a recommendation on a screen.
The Applicant argues that the specific recitation of the automated selection of a medical respiratory device based on ML-predicted intensity scores and displaying on a graphical user interface of a health care provider, and indicative of a category of the sleep apnea and the action strategy constitutes significantly more than generic computing. The Applicant concludes that the amended claims 1-20 are directed to significantly more than an abstract idea and requests withdrawal of the rejection.
The Examiner respectfully disagrees and sustains the rejection because the additional elements, individually and as an ordered combination, do not provide an inventive concept amounting to significantly more than the abstract idea.
Under MPEP § 2106.05, Step 2B evaluates whether the additional elements beyond the abstract idea provide significantly more. The additional elements in the claims are: a computer-implemented method, a machine-learning model, one or more data processors, a non-transitory computer readable storage medium, and a graphical user interface of a health care provider. The specification describes the computing environment as one or more general purpose computers, specialized server computers (Spec., para. 0082), the model as interchangeable (Any supervised machine-learning model can be deployed, Spec., para. 0057), and the display as a dashboard 800 may display the detailed record of each subject (Spec., para. 0072). Each element performs its standard function: the computer executes instructions, the model generates predictions, and the display shows results. The Applicant repeats the phrase automated selection of a medical respiratory device, but as explained in previously, the claim recites recommending a device, not selecting one. A recommendation displayed on a screen is output of the abstract process. MPEP § 2106.05(g) addresses this: insignificant post-solution activity, such as displaying results, does not amount to significantly more. The combination of generic hardware, an interchangeable model, and a standard display, when viewed together, does not transform the abstract idea into an inventive concept.
35 USC 103 Applicant Arguments
Applicant asserts that amendments to Claims 1, 12, and 13 overcome the rejection by introducing three limitations: (1) mapping an intensity score to a pre-defined threshold upon exceedance, (2) recommending a breathing device capable of automatically adjusting breathing of the subject, and (3) outputting a result representing an action strategy displayed on a graphical user interface for a healthcare provider indicating both sleep apnea category and the action strategy
Examiner respectfully disagreed, the applicant does not identify any deficiency in the prior art mapping supporting the rejection.
Long teaches threshold-based severity determination using predefined apnea–hypopnea index (AHI) ranges corresponding to severity categories. Kayyali teaches adjusting a CPAP device automatically based on diagnostic outputs and healthcare provider display data.
A person of ordinary skill in the art would have combined these teachings to produce a system that determines sleep apnea severity, recommends a treatment adjustment, and displays the result for a clinician. Refer to further details 35 usc 103 rejection below
Claim Rejections: 35 U.S.C. § 101
.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 1-20 are rejected under 35 U.S.C. § 101 because the claimed subject matter is directed to a judicial exception, specifically an abstract idea, without reciting elements that integrate the exception into a practical application or provide an inventive concept amounting to significantly more than the exception itself.
Step 1: Statutory Categories
The claims fall within recognized statutory categories under MPEP § 2106.03.
Process, Claims 1-11: The language reciting a computer-implemented method comprising: collecting a dataset ... extracting ... generating ... predicting ... generating an action strategy ... and outputting a result defines a series of acts or steps, placing these claims in the process category.
Machine, Claim 12: The language reciting a system comprising: one or more data processors; and a non-transitory computer readable storage medium containing instructions describes a concrete thing consisting of parts, placing this claim in the machine category.
Manufacture, Claims 13-20: The language reciting a computer-program product tangibly embodied in a non-transitory machine-readable storage medium describes a tangible article, placing these claims in the manufacture category.
All claims satisfy Step 1. The analysis proceeds to Step 2A to determine whether the claims recite a judicial exception.
Step 2A, Prong One: Abstract Idea Identification
Prong One asks whether the claim recites a judicial exception. The analysis evaluates each limitation under its broadest reasonable interpretation to determine whether it falls within an enumerated abstract idea category defined in MPEP § 2106.04(a)(2).
Invention Context
The invention relates to predicting sleep apnea severity using machine-learning models trained on compound features derived from patient health data. The specification describes a system that collects patient data from electronic health records and wearable sensors, extracts features, generates compound features through arithmetic operations, predicts an intensity score, maps the score to a severity category, and recommends action strategies to healthcare providers (Spec., para. 0005-0008). Refer to Figures 1, 6, and 9 for process flow details.
Under BRI per MPEP § 2111, amended independent claims 1, 12, and 13 recite a workflow that: (a) collects patient data, (b) identifies relevant indicators, (c) combines those indicators into a derived metric, (d) uses the metric to predict a risk score, (e) compares the score against a threshold, (f) recommends a breathing device when the score exceeds the threshold, and (g) displays the results on a healthcare provider interface. This sequence describes the cognitive workflow a clinician performs when evaluating a patient for sleep apnea: gathering information, identifying key indicators, synthesizing data, forming a prognostic opinion, and recommending a treatment pathway. The claims therefore describe a mental process of medical judgment combined with mathematical calculations for feature generation and threshold comparison.
Representative Independent Claim Markup with BRI Annotations
Claim 1 (as amended) is reproduced below. Non-bolded portions represent the abstract idea. Bolded portions represent additional elements evaluated in Prong Two and Step 2B. Below each limitation, a BRI annotation explains its scope.
Claim 1. A computer-implemented method comprising:
collecting a dataset from one or more data sources for a subject;
extracting a set of features from the dataset;
generating a compound feature from two or more features of the set of features;
predicting, for the subject, an intensity score for a sleep apnea by processing an input dataset that includes the compound feature using a machine-learning model;
generating an action strategy based on the intensity score of the sleep apnea predicted by the machine-learning model, the intensity score being mapped to a pre-defined threshold upon determining exceedance of the intensity score above the pre-defined threshold, recommending a breathing device to automatically adjust breathing of the subject; and
outputting a result that represents the action strategy, the result being:
displayed on a graphical user interface of a health care provider, and
indicative of a category of the sleep apnea and the action strategy.
Breathing-Device BRI Analysis:
The amended limitation recommending a breathing device to automatically adjust breathing of the subject is grammatically ambiguous. Under BRI per MPEP § 2111, the examiner must adopt the broadest reasonable interpretation consistent with the specification. Under BRI, the method generates a recommendation that the subject should use a type of breathing device (such as an APAP machine) that has the built-in capability of automatically adjusting breathing pressure. The method produces a recommendation; it does not control or interact with any physical device. Refer to par. 0003, 0045 – 0046, 0068 and figure 6.
Abstract Idea Classification
Under BRI per MPEP § 2111, the independent claims 1, 12, and 13 recite the abstract idea of forming a predictive medical judgment about sleep apnea severity and recommending a corresponding treatment strategy. This abstract idea falls within two enumerated categories.
Mental Processes per MPEP § 2106.04(a)(2)(III): A mental process includes concepts performed in the human mind, such as observations, evaluations, judgments, and opinions. The amended independent claims recite: collecting a dataset from one or more data sources for a subject (observation of patient information), extracting a set of features from the dataset (identifying relevant patient indicators), generating a compound feature from two or more features (synthesizing data points into a derived metric), predicting, for the subject, an intensity score for a sleep apnea (forming a prognostic evaluation), the intensity score being mapped to a pre-defined threshold upon determining exceedance of the intensity score above the pre-defined threshold (comparing a number against a benchmark), recommending a breathing device to automatically adjust breathing of the subject (forming a treatment opinion about what device to use), and outputting a result that represents the action strategy (expressing the judgment).
Each of these steps can be performed in the human mind or with pen and paper. A clinician gathers patient data (collecting a dataset), identifies relevant indicators like age, BMI, and oxygen levels (extracting features), mentally combines those indicators into a composite assessment (generating a compound feature), weighs the combined indicators to estimate severity (predicting an intensity score), compares the severity estimate against clinical guidelines (mapping to a threshold and determining exceedance), and writes in the chart that the patient should be fitted for an auto-adjusting pressure device (recommending a breathing device). The specification confirms this cognitive nature: Based on the intensity score or classification category, predicted by the machine-learning model, appropriate remedial action strategies are recommended to the user (Spec., para. 0032). The system outputs a recommendation; it does not itself control, activate, or connect to any breathing device.
Mathematical Concepts per MPEP § 2106.04(a)(2)(I): The claims also recite mathematical calculations. The limitation generating a compound feature from two or more features of the set of features describes performing arithmetic operations on extracted data values. The specification explains that compound features are calculated by measuring amount of O2 level (blood oxygen saturation) across different types of blood tests and multiplying by the subject age groups across median O2 levels (Spec., para. 0031). The limitation the intensity score being mapped to a pre-defined threshold upon determining exceedance of the intensity score above the pre-defined threshold describes a numerical comparison: mapping one value against a fixed reference number. These are mathematical calculations a person could perform with a calculator or pen and paper.
Human Analogy Scenario
The abstract nature of the claims is reinforced because the process mirrors fundamental human activities. Even if a computer performs these steps faster, the underlying process remains abstract per MPEP § 2106.04(a) because speed and efficiency do not transform an abstract idea into patent-eligible subject matter.
A sleep specialist reviews a patient chart, pulling together medical records and test results (collecting a dataset). She identifies key health indicators such as age, BMI, and blood oxygen levels (extracting features). She mentally calculates a composite risk metric by multiplying oxygen levels by the patient's age group (generating a compound feature). Based on her clinical experience, she estimates a severity score, checks it against a clinical guideline threshold, and finding the score exceeds the threshold, writes in the chart that the patient should be fitted for an auto-adjusting pressure machine (predicting an intensity score, determining exceedance, and recommending a breathing device).
Dependent Claims: Abstract Idea Analysis
Claims 2-3: Under BRI, these claims recite the dataset includes data from an electronic health record (claim 2) and data from a sensor in a wearable device (claim 3). These limitations specify data sources. Specifying where data comes from just narrows the abstract data-collection step without changing its character.
Claims 4, 14: Under BRI, these claims recite extracting from the dataset the set of features comprising of one or more demographic features, one or more comorbidities features, one or more anthropometric features or one or more sleep history features. These limitations specify feature categories. Specifying what types of indicators to identify narrows the abstract observation step.
Claims 5-6, 15-16: Under BRI, these claims recite generating a second compound feature from two or more features ... by performing arithmetic operations (claims 5, 15) and multiplying an O2 proportion in hemoglobin feature and a weight class feature (claims 6, 16). These limitations describe additional mathematical calculations per MPEP § 2106.04(a)(2)(I).
Claims 7, 20: Under BRI, these claims recite preprocessing the dataset, wherein the preprocessing comprises handling one or more missing values or converting a categorical set of features to a numerical set of features. Data preprocessing and format conversion are data-manipulation steps a person could perform with pen and paper. These are mental processes per MPEP § 2106.04(a)(2)(III).
Claims 8, 17: Under BRI, these claims recite predicting the sleep apnea using the machine-learning model trained using training dataset ... wherein the machine-learning model includes a Random Forest model, a Support Vector Machine model, or an AdaBoost model. Specifying model types narrows the abstract prediction to particular analytical techniques without changing the abstract nature.
Claims 9, 18: Under BRI, these claims recite mapping the intensity score of the sleep apnea to category of set of categories of the sleep apnea, wherein the set of categories includes a controlled apnea, a mild apnea, a moderate apnea and a severe apnea. Categorizing a numeric score into severity levels is a mental evaluation per MPEP § 2106.04(a)(2)(III).
Claims 10-11, 19: Under BRI, these claims recite generating the action strategy of one or more preventive recommendation strategies (claims 10, 19) and generating the action strategy of oral or breathing devices recommendation strategies (claim 11). Selecting a recommendation category based on severity is a human judgment per MPEP § 2106.04(a)(2)(III).
All dependent claims recite abstract ideas. The analysis proceeds to Prong Two.
Step 2A, Prong Two: Practical Application
Prong Two asks whether the claim as a whole integrates the judicial exception into a practical application by evaluating whether the additional elements, beyond the abstract idea, apply, rely on, or use the exception in a manner that imposes a meaningful limit per MPEP § 2106.04(d). Integration requires a specific technical mechanism reflecting a technological improvement, not merely a stated goal or result.
Independent Claims: Additional Element Evaluation
Computing Environment (computer-implemented, data processors, storage medium): The recitation of a computer-implemented method (claim 1), one or more data processors and a non-transitory computer readable storage medium containing instructions (claims 12-13) does not integrate the exception into a practical application. These elements describe a generic computing environment used to execute the abstract analytical process. The claims do not specify any particular processor configuration, hardware architecture, or memory arrangement. Per MPEP § 2106.05(f), reciting generic computing components that merely carry out the abstract idea amounts to instructions to apply the exception on a computer.
Machine-Learning Model: The recitation of a machine-learning model does not integrate the exception into a practical application. The claims do not specify any particular model architecture, training protocol, hyperparameter configuration, or technical improvement to the model itself. The specification confirms: Any supervised machine learning model can be deployed such as Random Forest, Support Vector Machine, AdaBoost, and Regression models (Spec., para. 0008). The model functions as a generic analytical tool that performs the abstract prediction step. Using a generic tool to perform the abstract idea faster does not impose a meaningful limit on the exception per MPEP § 2106.05(f).
Graphical User Interface: The recitation of displayed on a graphical user interface of a health care provider constitutes insignificant post-solution activity per MPEP § 2106.05(g). Displaying the output of an analysis on a screen is a necessary step in any computer implementation of the abstract idea. The specification describes a generic dashboard: A dashboard 800 may display the detailed record of each subject after the analysis of the data has been effectively performed by using machine-learning models (Spec., para. 0072). The dashboard displays information fields without any specific technical improvement to display technology. Removing this display step would not change what the claim fundamentally accomplishes; it would only change how results are communicated.
Combination as a Whole: When viewed together, the combination of a generic computing environment, a generic machine-learning model, and a generic display forms an input-process-output computing workflow. The system collects data, processes it through an analytical model, compares results against a threshold, generates a device recommendation, and displays the recommendation. This combination does not produce a synergistic technical effect. Each element performs its ordinary computing function within the combination, and the ordered sequence follows the inherent logic of the abstract idea rather than imposing external technical constraints. The combination does not transform the abstract idea into a practical application.
Dependent Claims: Prong Two
The dependent claims do not pass Prong Two.
Claims 2-4, 14: Specifying data sources and feature types is a field-of-use limitation per MPEP § 2106.05(h) that does not alter the abstract analysis.
Claims 5-6, 15-16: Adding mathematical operations narrows the abstract idea without adding new additional elements.
Claims 7, 20: Data preprocessing is insignificant pre-solution activity per MPEP § 2106.05(g).
Claims 8, 17: Specifying model types narrows the tool selection but does not add a new additional element.
Claims 9-11, 18-19: Mapping scores to categories and specifying recommendation types further detail the abstract judgment without adding new additional elements.
When viewed as a whole, the combination of elements in the independent and dependent claims does not integrate the judicial exception into a practical application. The analysis proceeds to Step 2B.
Step 2B: Inventive Concept
Step 2B asks whether the additional elements, individually and as an ordered combination, provide an inventive concept that amounts to significantly more than the judicial exception itself per MPEP § 2106.05. This analysis is consistent with but distinct from Prong Two.
Independent Claims: Additional Element Evaluation
The additional elements evaluated are: computer-implemented method, a machine-learning model, one or more data processors, a non-transitory computer readable storage medium, and a graphical user interface of a health care provider.
Computing Environment: The specification describes these components as: In some embodiments, a system is provided that includes one or more data processors and a non-transitory computer-readable storage medium containing instructions which, when executed on the one or more data processors, cause the one or more data processors to perform part or all of one or more methods disclosed herein (Spec., para. 0009). The specification further describes the computing devices as general purpose computers such as personal computers and laptops, workstation computers, wearable devices, gaming systems, thin subjects, various messaging devices, sensors or other sensing devices, and the like (Spec., para. 0080), and the server as one or more general purpose computers, specialized server computers (Spec., para. 0082). These applicant admissions confirm the computing environment consists of general-purpose hardware performing its ordinary function. Mere instructions to apply an abstract idea using generic computer components do not amount to significantly more per MPEP § 2106.05(f).
Machine-Learning Model: The specification describes: Any supervised machine learning model can be deployed such as Random Forest, Support Vector Machine, AdaBoost, and Regression models (Spec., para. 0008). The specification also states: Any supervised machine-learning model can be deployed. Examples of supervised machine-learning models that can be used include Random Forest model, Support Vector machine model, AdaBoost model, KNN, Regression, etc. (Spec., para. 0057). These applicant admissions confirm the machine-learning model is a generic, interchangeable analytical tool. Using a generic tool to execute the abstract analysis does not provide significantly more per MPEP § 2106.05(f).
Graphical User Interface: The specification describes: A dashboard 800 may display the detailed record of each subject after the analysis of the data has been effectively performed by using machine-learning models (Spec., para. 0072). Displaying analytical results on a dashboard is a post-solution output step and does not amount to significantly more per MPEP § 2106.05(g).
Combination as a Whole: The ordered combination of additional elements does not provide an inventive concept. The specification describes the architecture as a distributed system with computing devices coupled to a server via networks (Spec., para. 0076), where the server includes one or more general purpose computers, specialized server computers (Spec., para. 0082). This general-purpose computing arrangement does not transform the abstract idea into something significantly more.
Dependent Claims: Step 2B
Claims 2-4, 14: Specifying data sources and feature types narrows the field of use and does not provide an inventive concept per MPEP § 2106.05(h).
Claims 5-6, 15-16: Additional mathematical operations do not introduce new additional elements and do not provide an inventive concept.
Claims 7, 20: The specification describes data preparation as tasks like handling one or more missing values, converting a categorical set of features to a numerical set of features, correcting data anomalies, and removing noise (Spec., para. 0042). These generic data-cleaning steps do not provide an inventive concept.
Claims 8, 17: The specification describes model types as interchangeable: Any supervised machine-learning model can be deployed (Spec., para. 0057). Specifying a particular interchangeable model does not provide an inventive concept.
Claims 9-11, 18-19: Mapping scores to categories and specifying recommendation types further detail the abstract judgment without introducing new additional elements or an inventive concept.
Whole-Claim Determination
The ordered combination of elements in all claims does not transform the abstract idea into an inventive concept. The claims describe a process of collecting patient data, analyzing it to predict sleep apnea severity, and recommending treatment strategies including breathing devices. The additional elements serve only as tools to automate the abstract process without providing significantly more than the exception itself.
Claims 1-20 are directed to the abstract idea of forming a predictive medical judgment about sleep apnea severity and recommending a corresponding treatment strategy. This abstract idea encompasses mental processes (clinical observation, evaluation, judgment, and opinion) and mathematical concepts (arithmetic feature calculations and threshold comparisons).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claim(s) 1-5, 7-15, 17 and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over EP4298989A1- Long in combination with Kayyalie US8172766 -A1.
Claim 1.
Long teaches, A computer-implemented method comprising: collecting a dataset from one or more data sources for a subject; (Long, par. 0008, 0013, 0018, 0033-0034 )
Long, describe a system where used different data set first breathing detector, a first time-varying signal, from different devices; sleep-monitoring, breathing detectors and user inputs.
extracting a set of features from the dataset; (Long, par. 0079-0084, 0101)
Long describe an array of training inputs and known outputs, each training input corresponding to a respective known output; each training input comprises at least one or more snore rate variability characteristics for a sleep session for a subject; and each known output comprises a measure of a severity of sleep disordered breathing for the subject.
generating a compound feature from two or more features of the set of features; (Long, par. 0008, 0039, 0079-0084, 0101, 0035, 0039-0042, 0055)
Long describe the use of different types of data, for example: demographic, breathing (biometric data by wearable device) and Sleep quality data.
predicting, for the subject, an intensity score for a sleep apnea by processing an input dataset that includes the compound feature using a machine-learning model; (Long, par. 0041-0045, 0073-0078)
Long describe a machine-learning algorithm that processes input data in order to produce or predict output data. Here, the input data comprises at least the one or more snore rate variability characteristics and the output data comprises a measure of SDB severity.
, the intensity score being mapped to a pre-defined threshold upon determining exceedance of the intensity score above the pre-defined threshold, (Long, par. 0041-0045, 0073-0078)
Long describe a machine-learning algorithm that processes input data in order to produce or predict output data. Here, the input data comprises at least the one or more snore rate variability characteristics and the output data comprises a measure of SDB severity.
. (Long, 0037]-0046, 0041-0045, 0068, 0074-0075).
Long teaches determining a category of sleep apnea severity from patient features. In particular, Long discloses determining “a measure of SDB severity” that may be “selected from a plurality of predefined severity levels,” including normal, mild, moderate, and severe, which teaches determining and classifying the severity category of the patient’s sleep apnea condition from patient-derived inputs.
35 U.S.C 103 rational per amendment to claim1, the intensity score being mapped to a pre-defined threshold upon determining exceedance of the intensity score above the pre-defined threshold, recommending a breathing device to automatically adjust breathing of the subject
Long discloses a processing system that uses a machine-learning algorithm to output a measure of sleep disordered breathing severity. That measure can be an estimated apnea hypopnea index - a numerical score - which is then mapped to predefined severity levels anchored to fixed AHI thresholds: below 5 for normal, 5 to 15 for mild, 15 to 30 for moderate, and above 30 for severe. This mapping satisfies the limitation's requirement that a machine-learning model produce a severity score that is compared against predefined thresholds to determine a severity category, including determining when the score exceeds a given threshold. (Long, See at least, The processing system may use a machine learning algorithm to determine the measure of SDB severity, par. 0012; the determined measure of the severity of sleep disordered breathing is a severity level selected from a plurality of predefined severity levels, an estimated apnea hypopnea index, an estimated obstructive apnea index, an estimated hypopnea index, an estimated central apnea index, an estimated oxygen desaturation index, or an estimated respiratory disturbance index; the predetermined severity levels may comprise "mild and normal", "moderate" and "severe"; In clinical practice, an OSA severity is typically considered normal in subjects having an AHI < 5, mild in subjects having an AHI greater than or equal to 5 and less than 15, moderate in subjects having an AHI greater than or equal to 15 and less than 30, and severe in subjects having an AHI greater than or equal to 30, par. 0041-0044).
Long does not describe recommending a breathing device to automatically adjust breathing of the subject.
Kayyali teaches recommending a breathing device to automatically adjust breathing of the subject of Claim 1, that required, any system output that identifies or recommends a type of breathing device having the built-in capability to automatically adjust breathing pressure as part of an action strategy generated upon threshold exceedance. Kayyali discloses a system that produces a recommended positive pressure setting in response to a severity output and further requires human intervention prior to changing the positive pressure delivered to the subject (Kayyali, Claim 1). This embodiment reads directly on the limitation: the system recommends a self-adjusting breathing device, and the healthcare provider makes the final treatment decision. (Kayyali, See at least, One particular embodiment of the present invention involves diagnosing the level of severity of a Subject's sleep apnea and with or without human intervention adjusting an apparatus for providing positive airway pressure to the Subject, Col. 3. Ll.20-49; a continuous positive airway pressure (CPAP) device for treating the subjects sleep apnea, whose airflow rate can be automatically adjusted using the output, Col.3, ll. 53-67; delivery of a physical treatment such as CPAP can be adjusted during the treatment based on the sleep diagnosis results, Col. 21, ll. 55-67; The CPAP machine would preferably be set up to receive some type of signal, which would cause an adjustment in the flow rate or pressure of the breathing gas being delivered to the Subject, Col.22, ll. 1-15; the CPAP device is adjusted in real time, Claim 8, claim 1).
For example a skilled artisan in sleep disorder diagnostics and automated respiratory treatment with a background in biomedical engineering or clinical informatics, and direct experience developing machine-learning diagnostic tools and integrated therapeutic sleep devices, who read Long's application would combine Kayyali with Long because a practitioner deploying Long's severity-scoring system in a clinical setting would immediately recognize that a severity determination without a corresponding treatment response is clinically incomplete. A POSITA working in this field understands that sleep apnea severity scores exist to drive treatment decisions, this is clinical reality in sleep medicine. Searching the prior art for systems that connect severity diagnosis to treatment delivery, the POSITA would find Kayyali, which operates in the same clinical domain and provides exactly that connection: a system that takes a severity output and uses it to automatically drive CPAP adjustment. (See at least, Long: There is therefore a need for a less obtrusive and less costly method for determining the severity of OSA and other sleep disordered breathing (SDB) conditions, par. 0006; Kayyali: the sleep disorder treatment system of the present invention can use a diagnosis device to perform various forms of analysis to determine or diagnose a subjects sleeping disorder or symptoms, and using this analysis or diagnosis can with or in some embodiments without human intervention treat the subject either physically or chemically to improve the sleeping disorder or the symptoms of the sleeping disorder, Col. 2. Ll.45-67).
The combination of Long + Kayyali makes obvious the full limitation above, because Long produces a machine-learning-derived severity score mapped to predefined AHI thresholds supplying the scoring, mapping, and threshold-exceedance elements and Kayyali teaches that when sleep apnea severity is determined, the known clinical response is to automatically adjust a CPAP breathing device using that severity output. A POSITA would implement Kayyali's automatic CPAP adjustment as the action-generation step within Long's severity-scoring pipeline, requiring no redesign of either system and yielding the specific predictable result of a workflow that moves from ML-derived score, to threshold comparison, to automatic breathing device adjustment as a single integrated clinical loop.
A POSITA in sleep disorder diagnostics and automated respiratory treatment would be motivated to integrate Kayyali's automatic CPAP adjustment into Long's ML-severity pipeline because Long determines severity without acting on it a gap that a POSITA practicing in clinical deployment would recognize as limiting the system's real-world value. Kayyali's technique of using severity output to drive automatic CPAP adjustment was a known, proven approach operating on the same type of diagnostic input in the same clinical domain. Combining them requires no modification to either system's core function, Long's ML model continues to score severity, and Kayyali's CPAP adjustment continues to respond to severity output, making the combination predictable rather than speculative, and the routine next step a POSITA would take to produce a clinically complete system.
35 U.S.C 103 rational per amendment to claim1, outputting a result that represents the action strategy, the result being: displayed on a graphical user interface of a health care provider, and indicative of a category of the sleep apnea and the action strategy
Long teaches determining a category of sleep apnea severity from patient features. In particular, Long discloses determining “a measure of SDB severity” that may be “selected from a plurality of predefined severity levels,” including normal, mild, moderate, and severe, which teaches determining and classifying the severity category of the patient’s sleep apnea condition from patient-derived inputs. However, Long does not expressly teach outputting a result that represents the action strategy, where the result is displayed on a graphical user interface of a health care provider and is indicative of a category of the sleep apnea and the action strategy. (Long, 0037]-0046, 0041-0045, 0068, 0074-0075).
Kayyali discloses that a CPAP device can “produce a recommended positive pressure setting” in response to diagnostic output from the data acquisition system, and further discloses that the system can “display the data” on a viewing device “for further input by the Subject’s physician or another clinician.” These disclosures teach outputting and displaying a treatment-related result to a health care provider interface, where the recommended pressure setting represents the action strategy. (Kayyali, Col. 23, ll.20-30, claim 1-2; col. 9, ll. 15-27; col. 11, ll. 55-67; col. 19, ll. 1-15, Col. 11, ll. 55-67).
A POSITA would have been motivated to combine Long with Kayyali because the references address adjacent parts of the same sleep-apnea workflow and are technically compatible: Long provides the severity/category determination, and Kayyali uses diagnostic output to generate and display a treatment recommendation for clinician review. A POSITA would have understood that Long’s severity output could be used as the input to Kayyali’s recommendation logic, so that the determined severity category would drive the displayed recommended therapy setting. This integration would have predictably yielded a more clinically useful diagnosis-to-treatment system by linking known severity classification with known therapy recommendation and display functions. Under the broadest reasonable interpretation, the displayed recommended therapy setting is indicative of both the category of the sleep apnea and the action strategy because the recommendation is generated from, and thus reflects, the determined severity. Therefore, the combination of Long and Kayyali teaches or suggests the full limitation. (Long, 0041-0045, 0074-0075; Kayyali, claim 1; col. 9, ll. 15-27; col. 11, ll. 55-67; col. 19, ll. 1-15)
Claim 2.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, wherein the dataset includes data from an electronic health record. (Long, par. 0030-0031, 0040, 0094)
Long describe electronic health record, because use demography data as specify by the applicant in paragraph 0006.
Claim 3.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, wherein the dataset includes data from a sensor in a wearable device. (Long, par. 0008, 0013, 0039, 0079-0084, 0101, 0035, 0039-0042, 0055, 0097)
Long describe input data comprises at least the one or more snore rate variability characteristics and the output data comprises a measure of SDB severity, that include sleep quality data through sleep-monitoring devices.
Claim 4.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, further comprising: extracting from the dataset the set of features comprising of one or more demographic features, one or more comorbidities features, one or more anthropometric features or one or more sleep history features. (Long, par. 0029, 0031 , 0019, 0008, 0039, 0079-0084, 0101, 0035, 0039-0042, 0055)
Claim 5.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, further comprising: generating a second compound feature from two or more features of the set of features by performing arithmetic operations on the two or more features of the set of features. (Long, 0026, 0068, 0091-0093)
Long, demonstrates the use of arithmetic operations, including multiplication, addition, and statistical distribution parameter calculations (mean, variance, standard deviation, skewness, kurtosis, interquartile range), applied to snore energy and snore time stamp features to characterize snoring patterns and assess SDB severity.
Claim 7.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, further comprising: preprocessing the dataset, wherein the preprocessing comprises handling one or more missing values or converting a categorical set of features to a numerical set of features. (Long, par. 0079-0085, 0022, 0067)
Long describes a neural network, each neuron performs a mathematical operation, which is a weighted combination of a transformation on the input data, to sequentially produce a numerical output that is fed into subsequent layers until the final output is provided.
Claim 8.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, further comprising: predicting the sleep apnea using the machine-learning model trained using training dataset that includes the compound feature associated with each subject of a set of subjects wherein the machine-learning model includes a Random Forest model, a Support Vector Machine model, or an AdaBoost model. (Long, par. 0080)
Claim 9.
Long in combination with Kayyali teaches, The computer-implemented method of claim 1, further comprising: mapping the intensity score of the sleep apnea to a category of a set of categories of the sleep apnea, wherein the set of categories includes a controlled apnea, a mild apnea, a moderate apnea and a severe apnea. (Long, par. 0041-0043, 0075, 0101)
Note: Claim 12-15, 17 and 20 are rejected with the same analysis to be very similar.
Claim(s) 10-11 and 18-19 are rejected under 35 U.S.C. 103 as being unpatentable over EP4298989A1- Long in combination with Kayyalie US8172766 -A1 in further view of WO2007014429- Schindhelm.
Claim 10.
Long in combination with Kayyali teaches, The computer-implemented method of claim 9, wherein the category of the set of categories is the mild apnea, and wherein the computer-implemented method further comprises: generating the action strategy of one or more preventive recommendation strategies for the subject. (Long, par. 0041-0043, 0075, 0101; Schindhelm, par. 0003, 0005, 0008, 0026, 0034)
The Long and Schindhelm systems collaboratively address obstructive sleep apnea (OSA) by treating or managing its symptoms to mitigate severity. For instance, the Schindhelm system uses the Apnea-Hypopnea Index (AHI) as a treatment trigger and employs indicator lights to show OSA severity. This feedback facilitates ongoing management and allows for treatment adjustments to prevent worsening conditions. A key physical intervention described is the use of a mandibular advancement device, which repositions the patient's lower jaw forward to maintain an open airway and prevent collapse, a primary cause of OSA.
Claim 11.
Long in combination with Kayyali teaches, The computer-implemented method of claim 9, wherein the category of the set of categories is the moderate apnea, and wherein the computer-implemented method further comprises: generating the action strategy of oral or breathing devices recommendation strategies for the subject. (Long, par. 0041-0043, 0075, 0101; Schindhelm, par. 0003, 0005, 0008, 0026, 0034)
The Long and Schindhelm systems collaboratively manage obstructive sleep apnea by using the Apnea-Hypopnea Index (AHI) to trigger treatment and employing a mandibular advancement device for airway maintenance, thereby preventing worsening conditions.
Note: Claim 18-19, and 10-11 are rejected with the same analysis to be very similar.
Claim(s) 6 and 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over EP4298989A1- Long in combination with Kayyalie US-20140332010-A1, and further in view of Carl E. Rhodes Physiology, Oxygen Transport refer to PTO-892-U.
Claim 6.
Long in combination with Kayyali describe,, The computer-implemented method of claim 5, wherein generating the second compound feature includes 2 proportion in hemoglobin feature and a weight class feature of the subject. (Long, par. 0004, 0041, 0045, 0076 and 0026, 0042, 0075)
Long describe the severity of sleep-disdisordered breathing, particularly Obstructive Sleep Apnea (OSA), can be determined using various oxygen-related parameters like the Oxygen Desaturation Index (ODI) as a surrogate for AHI, or the more comprehensive Respiratory Disturbance Index (RDI) which includes respiratory-effort related arousals (RERAS), along with other indices like AHI, obstructive apnea, hypopnea, central apnea, and oxygen desaturation indices, all of which can be characterized by statistical distribution parameters. Also, Long include the function to predetermined severity levels may comprise "mild and normal", "moderate" and "severe. However does not explicitly disclosed how it is calculated the oxygen proportion in hemoglobin feature by the specific method of use multiplication.
However, missing element it is disclosed in Rhodes it is explained how develop the calculation process by use multiplication, CaO = 1.34 * [Hgb] * (SaO/100) + 0.003 * PaO2. This equation directly shows a calculation (generation) of the total oxygen concentration (CaO) in the blood. It includes a multiplication of hemoglobin concentration ([Hgb]), which serves as a proxy for a "weight class feature" in the context of oxygen-carrying capacity, and hemoglobin saturation (SaO/100), which represents the "O2 proportion in hemoglobin." The constant 1.34 mL O2/g Hbg represents the oxygen-carrying capacity per gram of hemoglobin, acting as a critical factor in this generation. Refer to Related Testing in Rhodes Refer PTO-892-U
It is obvious to combine Carl E. Physiology, Oxygen Transport with Long and Schindhelm because hemoglobin primary function is to transport oxygen from the lungs to the rest of the body, and to carry carbon dioxide back to the lungs to be exhaled important parameter in apnea disease as disclosed by Long different parameters in relationship of oxygen levels.
A PHOSITA will be motivates to include calculation using multiplication of the oxygen proportion in relationship of hemoglobin because, Hemoglobin, primarily within red blood cells, serves as the main carrier, with "Approximately 98% of total oxygen transported in the blood is bound to hemoglobin". Refer to Rhodes - Mechanism
Note: Claim 16 it is rejected with the same analysis to be very similar.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure. Javed (US 20240145085)-par. 0106, because shows that a CPAP device can be recommended specifically if a patient’s AHI is above a threshold.
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JOSHUA DAMIAN RUIZ whose telephone number is (571)272-0409. The examiner can normally be reached 0800-1800.
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, Shahid Merchant can be reached at (571) 270-1360. 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.
/JOSHUA DAMIAN RUIZ/Examiner, Art Unit 3684
/Shahid Merchant/Supervisory Patent Examiner, Art Unit 3684