Non-invasive health risk screening and validation system

WO2026217016A3PCT designated stage Publication Date: 2026-11-19ABULABAN AMANI
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
PCT/SA2026/050027
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2025-12-14
Filing Date
2026-04-07
Publication Date
2026-11-19

AI Technical Summary

Technical Problem

Existing non-invasive screening technologies are limited to single disease types or sensor outputs and do not effectively integrate non-invasive physiological evidence with adaptive questioning to support comprehensive chronic disease screening across multiple disease classes.

Method used

A configurable system that combines non-invasive physiological acquisition with adaptive questioning, conflict resolution, and score fusion to generate differentiated outputs for both providers and subjects, using a unified screening architecture deployable through various platforms.

Benefits of technology

Enables efficient screening for multiple chronic diseases by reducing redundant questioning and generating tailored outputs, supporting broad chronic disease screening without architectural redesign.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure SA2026050027_19112026_PF_FP_ABST
    Figure SA2026050027_19112026_PF_FP_ABST
Patent Text Reader

Abstract

The invention relates to a computer-implemented system and method for non-invasive screening of a subject for one or more target health conditions. The system may be deployed through a smart kiosk, mobile device, wearable interface, smart minor, or related platform and may obtain physiological data using photoplethysmography, remote photoplethysmography, computer vision, optical sensing, or combinations thereof. A condition-model registry stores a plurality of condition profiles associated with chronic diseases and disease classes, including endocrine diseases, metabolic diseases, cardiovascular diseases, and additional conditions that may be added through configuration. A gate controller may execute a sequential workflow including onboarding, sentinel questions, physiological acquisition, and questionnaire validation. A pre-score registry stores earlier-stage evidence to reduce redundant questioning, and a conflict-resolution engine suppresses mutually inconsistent condition pathways before further questioning.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] NON-INVASIVE HEALTH RISK SCREENING AND VALIDATION SYSTEM CROSS-REFERENCE TO RELATED APPLICATIONS

[0002] The present application claims the benefit of U.S. Provisional Patent Application No.

[0003] 63 / 786,327, filed April 10, 2025, entitled "CAREVISION COGNITIVE Al HEALTH SCREENING AND ENGAGEMENT SYSTEM VIA SMART KIOSKS AND HEALTH COMPANION," under 35 U.S.C. § 119(e). The entirety of the provisional application is incorporated herein by reference.

[0004] In nonprovisional practice, the benefit claim to the above provisional application should also be presented in an Application Data Sheet in accordance with 37 C.F.R. § 1.78 and USPTO practice.

[0005] BACKGROUND OF THE INVENTION

[0006] The present invention relates to digital health technologies, non-invasive physiological screening, adaptive assessment systems, and automated health-risk stratification. More particularly, the invention concerns a configurable system and method for screening a subject for one or more target health conditions by combining non-invasive physiological acquisition, staged intake logic, adaptive questioning, score accumulation, and output generation within a unified screening architecture deploy able through kiosks, mobile devices, wearables, smart mirrors, and related computing platforms.Existing non-invasive screening technologies, including remote photoplethysmography (rPPG), contact photoplethysmography, camera-based heart-rate variability estimation, respiratory monitoring, and facial micro-circulation analysis, are capable of extracting physiological parameters such as heart rate, heart rate variability, respiratory rate, perfusion indicators, and related biomarker signals. Such systems are useful for measurement, but they typically function as stand-alone sensing or estimation tools rather than as extensible condition-screening architectures configured to support multiple disease classes through a shared computational pipeline.

[0007] Questionnaire-driven health-risk systems are also known. Certain prior systems dynamically adapt questions based on prior answers, segmentation logic, or predictive analytics. However, such systems generally begin from subjective input and do not use non-invasive physiological evidence as an integrated trigger, constraint, or directional control signal within a multi-stage screening architecture directed to multiple chronic disease classes.

[0008] Clinical decision support systems and health kiosks are likewise known. Some systems use historical electronic records to suggest clinical actions, while others provide self-service patient intake, check-in, or preventive-care workflows. Yet such systems do not disclose a non-invasive chronic-disease screening architecture in which evidence is accumulated across sequential gates, stored in an intermediate score memory, used to suppress redundant questions, and transformed into both clinician-facing and subject-facing outputs from a common computational state.Cardiovascular rhythm detection devices and consumer cardiac-alert systems are also part of the prior-art environment. Products and systems exist for atrial fibrillation or arrhythmia notification, including consumer wearable irregular-rhythm alerts and dedicated ECG-based arrhythmia tools. Such systems may indicate possible rhythm irregularities, but they do not provide the broader chronic-disease screening framework disclosed herein, nor do they teach a configurable diseasemodel registry spanning endocrine, metabolic, cardiovascular, and other chronic conditions through a shared architecture.

[0009] Accordingly, a need exists for a unified, extensible screening system that is not limited to a single disease type or sensor output, and that instead supports screening for multiple target health conditions, including chronic diseases, by combining non-invasive physiological acquisition with adaptive questioning, conflict resolution, score fusion, and differentiated output generation. OBJECTS AND ADVANTAGES OF THE INVENTION

[0010] It is an object of the invention to provide a non-invasive screening platform capable of screening for a plurality of target health conditions using a shared acquisition and processing architecture. It is another object of the invention to extend screening beyond cardiovascular abnormalities to chronic diseases more broadly, including endocrine diseases, metabolic diseases, cardiovascular diseases, respiratory disorders, neurologic disorders, renal disorders, hepatic disorders, sleep-related disorders, inflammatory conditions, and related condition families, while permittingadditional disease models to be incorporated in the future through configuration rather than architectural redesign.

[0011] It is a further object of the invention to provide a staged screening process in which data gathered earlier in the workflow reduces later questioning, improves efficiency, and enables adaptive disease-specific assessment.

[0012] It is a further object of the invention to provide differentiated outputs for provider use and subject use from a common computational process, while allowing subject-facing wellness communications to be framed in a manner consistent with the FDA General Wellness Policy for low-risk products.

[0013] SUMMARY OF THE INVENTION

[0014] The invention provides a computer-implemented system and method for non-invasive screening of a subject for one or more target health conditions. In various embodiments, the system obtains physiological data through one or more non-invasive sensors, extracts physiological features, compares the features to a plurality of configurable condition profiles, generates one or more preliminary condition scores, presents adaptive questions responsive to candidate conditions, and generates validated risk classifications and differentiated outputs.

[0015] In some embodiments, the target health conditions comprise chronic diseases or chronic disease classes, including but not limited to endocrine diseases, metabolic diseases, cardiovascular diseases, respiratory diseases, neurologic diseases, renal diseases, hepatic diseases, sleep-relateddisorders, inflammatory disorders, and combinations thereof. In exemplary but non-limiting embodiments, endocrine and metabolic conditions may include hypothyroidism, hyperthyroidism, type 2 diabetes, prediabetes, metabolic syndrome, primary hyperaldosteronism, adrenal insufficiency, and Cushing's syndrome. In exemplary but non-limiting embodiments, cardiovascular conditions may include a cardiovascular disease category and illustrative subtypes such as arrhythmia, atrial fibrillation, supraventricular tachycardia, premature ventricular contractions, heart failure, and sleep apnea.

[0016] In preferred embodiments, the system is configurable through a condition-model registry comprising a plurality of condition profiles. Each condition profile may define one or more associated physiological indicators, symptom indicators, abnormality patterns, scoring rules, weighting rules, exclusion rules, question logic, conflict-resolution logic, and output policies. New condition profiles may be added or modified without altering the core acquisition and processing architecture.

[0017] In certain embodiments, the system executes a sequential screening workflow comprising an onboarding stage, a sentinel-question stage, a physiological acquisition stage, and a questionnaire stage. Data acquired at one stage may be stored and reused to influence later stage selection, question suppression, condition ranking, or output generation.

[0018] In certain embodiments, the system further includes a pre-score registry configured to accumulate scored items generated from earlier stages of the workflow, thereby reducingredundant questioning and preserving informational continuity across stages. In certain embodiments, a conflict-resolution mechanism operates before questionnaire presentation to identify mutually inconsistent condition hypotheses and suppress at least one branch based on directional physiological evidence or rule-based incompatibility.

[0019] In certain embodiments, the system generates both a provider-facing risk analysis output and a subject-facing wellness-language output from the same underlying computation. Subject-facing output may include wellness guidance, monitoring prompts, and referral encouragement framed in accordance with the FDA General Wellness Policy for low-risk products, while providerfacing output may include richer analytical detail, condition ranking, and follow-up rationale. BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a line illustration showing a non-invasive sensor capturing physiological signals from a subject and transmitting extracted signals to an automated processor for analysis.

[0020] FIG. 2 is a flow diagram illustrating core validation logic including digital biomarker acquisition, abnormality detection, threshold comparison, questionnaire triggering, weighted scoring, score fusion, and final risk classification.

[0021] FIG. 3 illustrates a system architecture including a data acquisition layer, core processing engine, validation module, risk classification module, and output and integration layer.

[0022] FIG. 4 is a flow diagram illustrating a multi-condition screening workflow in which a shared acquisition layer evaluates multiple candidate disease classes through a condition-model registry.FIG. 5 is a flow diagram illustrating a four-stage gate funnel comprising onboarding, sentinel questions, physiological acquisition, and questionnaire validation.

[0023] FIG. 6 illustrates an exemplary pre-score registry storing stage-derived evidence for reuse and redundancy suppression.

[0024] FIG. 7 illustrates an exemplary conflict-resolution mechanism for suppressing mutually inconsistent condition pathways before questionnaire generation.

[0025] FIG. 8 illustrates a dual-register output architecture generating provider-facing and subjectfacing outputs from a common computational state.

[0026] DEFINITIONS

[0027] As used herein, the term “target health condition” means any disease, disorder, syndrome, risk state, physiologic abnormality, wellness-relevant deviation, or disease class for which the system is configured to screen, stratify, or generate a follow-up assessment pathway. The term includes both named diseases and broader categories or families of related conditions.

[0028] As used herein, the term “condition profile” means a data structure, model, rule set, or registry entry associated with a target health condition and comprising one or more of physiological features, thresholds, classifier inputs, symptom indicators, question-selection logic, exclusion rules, conflict-resolution logic, score weights, and output-generation rules.

[0029] As used herein, the term “chronic disease” includes conditions that are persistent, recurrent, progressive, or long-duration, and may include endocrine, metabolic, cardiovascular, respiratory,neurologic, renal, hepatic, inflammatory, sleep-related, and other disease classes. Such examples are illustrative and non-limiting.

[0030] As used herein, the term “cardiovascular disease” may refer to a broad disease category represented by one or more condition profiles and is not limited to any particular cardiovascular subtype unless expressly stated.

[0031] DETAILED DESCRIPTION OF THE INVENTION

[0032] In one embodiment, the invention is implemented as a multi-condition screening architecture rather than a single-condition detector. The architecture is configured to screen for a plurality of target health conditions using shared sensing, processing, and assessment components. A subject need not be pre-assigned to a single disease pathway prior to data acquisition. Instead, the system may evaluate multiple candidate conditions, disease classes, or syndrome families concurrently or sequentially using a common physiological acquisition layer and a configurable conditionmodel registry.

[0033] In some embodiments, the plurality of target health conditions includes chronic diseases. Chronic diseases may include, without limitation, endocrine diseases, metabolic diseases, cardiovascular diseases, respiratory diseases, neurologic diseases, sleep-related disorders, renal diseases, hepatic diseases, inflammatory disorders, and combinations thereof. In some embodiments, endocrine and metabolic conditions include hypothyroidism, hyperthyroidism, type 2 diabetes, prediabetes, metabolic syndrome, primary hyperaldosteronism, adrenal insufficiency, and Cushing'ssyndrome. These examples are illustrative only, and additional disease models may be incorporated by adding or modifying condition profiles within the registry.

[0034] In some embodiments, the system includes a cardiovascular disease profile or profile family that functions as a broad category-level screening construct. Such a profile may encompass a plurality of subtypes or manifestations rather than being restricted to a single arrhythmia class. Illustrative embodiments may include arrhythmia, atrial fibrillation, supraventricular tachycardia, premature ventricular contractions, heart failure, and sleep apnea. However, the invention is not limited to those examples, and other cardiovascular conditions may be included.

[0035] 1. CONDITION-MODEL REGISTRY

[0036] In one embodiment, the system stores a condition-model registry in memory. The registry comprises a plurality of condition profiles, each associated with a respective target health condition, disease subtype, disease class, syndrome family, or risk state. Each condition profile may include one or more of: associated physiological features, abnormality signatures, symptom indicators, sentinel questions, questionnaire items, weighting parameters, conflict rules, exclusion rules, stage-trigger rules, score thresholds, referral logic, and output templates.

[0037] The condition-model registry enables the system to be extended beyond any single disease family. In some embodiments, a new disease or disease class is deployed by adding a new registry entry or modifying an existing registry entry, without requiring hardware redesign or structural modification of the shared acquisition, scoring, or output architecture.The registry may support hierarchical organization. For example, a higher-level cardiovascular profile may lead to one or more lower-level subtype profiles, or a broader endocrine / metabolic profile may branch into specific hypothyroid, hyperthyroid, insulin-resistance, or adrenal-related profiles. Such hierarchical organization allows the system to remain broad at early stages and progressively increase condition specificity as evidence accumulates.

[0038] 2. SEQUENTIAL GATE FUNNEL ARCHITECTURE

[0039] In one embodiment, the system executes a four-stage sequential screening workflow comprising: (1) onboarding, (2) sentinel questions, (3) physiological acquisition, and (4) questionnaire validation. The stages may be executed in fixed order, configurable order, or condition-dependent order, provided that information gathered at an earlier stage is made available to a later stage.

[0040] At the onboarding stage, the system may collect demographics, family history, medical history, medication information, lifestyle indicators, prior diagnoses, baseline metrics, and user preferences. At the sentinel-question stage, the system presents a limited subset of high-discrimination questions configured to separate broad disease classes or condition families. At the physiological acquisition stage, the system acquires non-invasive physiological data and derives features relevant to one or more condition profiles. At the questionnaire stage, the system presents adaptive follow-up questions only for unresolved or sufficiently ranked candidate conditions.This staged architecture is particularly useful when screening for chronic diseases spanning multiple condition families, because it allows broad intake and rapid narrowing before presenting condition- specific follow-up questions. As a result, the system can efficiently support cardiovascular, endocrine, metabolic, and other chronic disease pathways within a single platform.

[0041] 3. NON-INVASIVE SENSING AND DATA ACQUISITION

[0042] The system includes one or more non-invasive sensors such as an RGB camera, infrared sensor, photoplethysmographic imaging module, wearable sensor, or mobile-device sensor configured to capture physiologic data from a subject. In some embodiments, the system captures facial blood-flow variations, pulse-related optical signals, respiratory-related motion, thermal patterns, or other contactless physiologic indicators.

[0043] The system may further receive supplementary inputs from wearable devices, medical devices, electronic medical record systems, laboratory systems, and user input interfaces. This multisource data aggregation enables the system to form a rich biometric and contextual input set for downstream processing.

[0044] Depending on the target condition profile or disease class being evaluated, the extracted physiological features may further include one or more features indicative of autonomic balance, peripheral perfusion, vascular stiffness, pulse-wave morphology, thermal asymmetry, sleep-related breathing variation, stress-state proxies, activity-state proxies, or combinations thereof.The invention is not limited to a particular physiological parameter set and may operate on any non-invasive feature set suitable for differentiating among multiple chronic conditions or disease classes.

[0045] In certain embodiments, physiological features derived from shared sensor inputs may contribute to multiple condition profiles simultaneously. For example, a single camera-based acquisition event may produce features relevant to cardiovascular screening, metabolic risk screening, stress-related screening, or sleep-related screening. Accordingly, the same acquisition event may feed parallel or ranked condition hypotheses rather than a single narrow disease pathway.

[0046] 4. CORE PROCESSING ENGINE AND ANOMALY DETECTION

[0047] The core processing engine may include a signal processing module, a feature extraction module, and an anomaly detection engine. The signal processing module may perform preprocessing operations such as region-of-interest selection, temporal filtering, spatial averaging, waveform decomposition, and motion-artifact suppression.

[0048] The feature extraction module may derive one or more physiological metrics including heart rate, heart rate variability, respiratory rate, pulse-wave characteristics, perfusion-related indicators, stress proxies, vascular-age estimates, blood pressure estimates, and additional derived measures relevant to one or more condition profiles.

[0049] The anomaly detection engine evaluates the extracted feature set through normative thresholding, deviation from personalized baselines, rule-based pattern recognition, probabilistic inference,machine-learning models, or combinations thereof. The engine may generate one or more preliminary condition scores corresponding to one or more candidate target health conditions. If a preliminary condition score exceeds a gate-specific threshold or ranking criterion, the system advances the corresponding condition pathway for additional validation.

[0050] 5. DISEASE-SCOPE EXAMPLES AND CONDITION PATHWAYS

[0051] Illustrative disease-scope examples include, without limitation, cardiovascular disease pathways, endocrine disease pathways, metabolic disease pathways, and other chronic disease pathways. A cardiovascular pathway may include condition profiles associated with arrhythmia, atrial fibrillation, supraventricular tachycardia, premature ventricular contractions, heart failure, sleep apnea, or other cardiovascular conditions. An endocrine or metabolic pathway may include profiles associated with hypothyroidism, hyperthyroidism, type 2 diabetes, prediabetes, metabolic syndrome, primary hyperaldosteronism, adrenal insufficiency, Cushing's syndrome, or related disorders.

[0052] These examples are illustrative and are not intended to limit the invention to any enumerated disease list, disease family, symptom list, or condition pathway. The same architecture may be extended to additional disease classes by configuring the condition-model registry and associated scoring logic.6. ADAPTIVE QUESTION SELECTION AND VALIDATION

[0053] When one or more candidate condition pathways satisfy a threshold, ranking, or selection criterion, the questionnaire engine retrieves one or more sentinel or validation questions associated with the corresponding condition profiles. Question selection may be based on disease class, subtype ranking, feature pattern, prior responses, conflict rules, prior history, or combinations thereof.

[0054] Illustrative examples of question selection across disease classes include, without limitation: a cardiovascular profile associated with irregular rhythm or abnormal pulse variability, leading to questions regarding palpitations, dizziness, exertional intolerance, syncope, chest discomfort, or prior rhythm diagnoses; an endocrine or metabolic profile associated with a pattern suggestive of insulin resistance, autonomic imbalance, or altered metabolic state, leading to questions regarding polyuria, polydipsia, weight change, fatigue, sleep pattern, dietary habits, medication use, or laboratory history; a thyroid-related profile leading to questions regarding heat intolerance, cold intolerance, tremor, constipation, bowel frequency, weight fluctuation, menstrual change, heart-rate symptoms, and medication history; and an adrenal -related profile leading to questions regarding orthostatic symptoms, salt craving, fatigue, hyperpigmentation history, truncal weight gain, blood-pressure pattern, muscle weakness, and steroid exposure.The response analysis module evaluates one or more user responses according to severity, frequency, chronicity, temporal pattern, symptom clustering, red-flag weighting, or other evidence -based scoring criteria to generate one or more response-based scores.

[0055] 7. PRE-SCORE REGISTRY

[0056] In one embodiment, the system maintains a pre-score registry that stores scored items, inferred indicators, rule outcomes, or confidence-weighted observations generated before the main questionnaire stage. The pre-score registry may receive entries from onboarding, sentinel questions, physiological acquisition, prior sessions, device integrations, or external records. Each registry entry may include one or more of: a stage identifier, a source identifier, a condition identifier, a score value, a confidence value, a polarity or direction flag, a timestamp, a reuse indicator, and a redundancy indicator. The questionnaire engine may inspect the pre-score registry before selecting questions so as to suppress questions that would merely duplicate already established evidence.

[0057] In this manner, the system is able to support broader chronic-disease screening without proportionally increasing question burden, because informational overlap among condition pathways can be managed through score reuse and redundancy suppression.

[0058] 8. CONFLICT-RESOLUTION MECHANISM

[0059] In one embodiment, the system applies a pre-questionnaire conflict-resolution process to candidate conditions that are mutually inconsistent, directionally opposed, or otherwise unlikelyto coexist within the same evidence state. The process may use direction flags, polarity indicators, exclusion rules, confidence comparisons, or hierarchical precedence rules to suppress at least one condition branch before presentation of further questions.

[0060] For example, evidence favoring one physiologic direction may reduce or eliminate the need to ask questions directed to an opposite or incompatible branch. In this way, the invention conserves question burden and improves logical consistency across multiple chronic-disease pathways. The conflict-resolution process may operate between subtypes within a disease class or across distinct disease classes.

[0061] 9. SCORE FUSION AND WEIGHTED AGGREGATION

[0062] In one embodiment, the scoring framework is disease-agnostic at the architectural level and disease-specific at the profile level. Thus, a common scoring engine may operate across multiple disease classes, while score components, feature weights, symptom weights, threshold values, and fusion logic are selected according to the applicable condition profile.

[0063] In certain embodiments, the system generates preliminary scores for multiple candidate conditions and applies ranking, weighting, thresholding, or conflict-resolution logic to determine which conditions proceed to later validation. In some embodiments, a weighted average, weighted linear combination, probabilistic model, machine-learning model, or hybrid fusion model is used to combine evidence from onboarding, sentinel questions, physiological acquisition, and questionnaire responses.The resulting validated risk classification may be expressed as a categorical, ordinal, probabilistic, or ranked output, including low risk, intermediate risk, suspected risk, elevated risk, or high risk.

[0064] 10. DUAL-REGISTER OUTPUT ARCHITECTURE

[0065] In one embodiment, the system generates output through a dual-register architecture based on a common internal computational state. A first register generates a provider-facing output comprising one or more of condition rankings, abnormality attributions, score breakdowns, confidence indicators, longitudinal comparisons, and follow-up rationale. A second register generates a subject-facing output comprising wellness-language messaging, self-monitoring prompts, educational content, adherence encouragement, and referral prompts.

[0066] In certain embodiments, the subject-facing output is expressly framed as a general-wellness communication for low-risk-device use, including maintenance or encouragement of healthy lifestyle practices, awareness of wellness-relevant physiologic patterns, and encouragement to seek professional care when appropriate. In such embodiments, the subject-facing register may omit or transform diagnostic phrasing while still preserving the analytical benefit of the underlying computation.

[0067] In some embodiments, the subject-facing output is tier contingent. For example, a lower-risk tier may trigger lifestyle guidance and re-screen timing, an intermediate tier may trigger enhanced monitoring or educational prompts, and an elevated tier may trigger a referral encouragementprompt. Such outputs may be generated for chronic disease categories generally and are not limited to any one cardiovascular condition.

[0068] 11. REFERRAL PROMPT AND PERSONALIZED WELLNESS CONTENT GRAPH In one embodiment, the system includes a referral-prompt module and a personalized wellness content graph. The referral -prompt module generates a subject-facing prompt based on validated risk tier, condition class, symptom context, persistence over time, or combinations thereof. The personalized wellness content graph links a subject's screening output to one or more wellness guidance nodes comprising educational content, behavior- support content, monitoring reminders, lifestyle suggestions, or follow-up prompts.

[0069] The wellness content graph may be condition-sensitive while remaining non-diagnostic in subject-facing presentation. By way of example, different content pathways may be selected for a cardiovascular-oriented profile, an endocrine / metabolic profile, a sleep-related profile, or a stress-related profile, with each pathway emphasizing general wellness support and care-seeking encouragement rather than definitive diagnosis.

[0070] 12. PLATFORM DEPLOYMENT AND INTEGRATION

[0071] The disclosed system may be deployed through smart kiosks, smart mirrors, mobile applications, wearable interfaces, teleconsultation platforms, pharmacy systems, electronic medical record systems, health-information systems, insurer triage platforms, or combinations thereof. In certain embodiments, different front-end platforms communicate with a shared back-end processingarchitecture that manages the condition-model registry, score generation, conflict resolution, and output rendering.

[0072] The system may further support longitudinal monitoring by storing prior-session information, prior scores, trend data, or historical registry entries for reuse in a later screening event.

[0073] 13. EXEMPLARY OPERATION

[0074] In one example, the system receives onboarding data and a small number of sentinel responses from a subject at a kiosk. The system then acquires non-invasive physiologic data using a camera-based optical sensor. Features are extracted and compared across multiple condition profiles. Preliminary scores are generated for a cardiovascular profile and an endocrine / metabolic profile. The pre-score registry records onboarding-derived and physiologic evidence. The conflict-resolution engine suppresses one low-confidence branch. The questionnaire engine presents only a small set of additional questions directed to the unresolved candidate condition. A fusion engine combines physiologic and response -based evidence to generate a validated risk classification. A provider-facing output is then generated for clinical review, while a subject-facing wellness-language output is delivered through the kiosk or a mobile application.

[0075] In another example, a mobile implementation acquires camera-based physiologic data and combines the result with onboarding data, medication history, and prior-session records to screenfor a broader disease class rather than a single named disease. A new disease profile may be introduced into the registry without changing the shared sensing pipeline.

[0076] 14. NON-LIMITING CLAUSE

[0077] Although various embodiments are described with reference to particular disease examples, physiological features, sensor types, question sets, or deployment environments, such examples are illustrative and non-limiting. The invention may be implemented with additional or alternative chronic diseases, disease classes, biomarkers, scoring methods, rule systems, machine-learning models, interfaces, and output modalities without departing from the scope of the disclosure.

[0078] In particular, addition of a new disease model to the condition-model registry is contemplated as an implementation of the disclosed architecture rather than a departure from it.

Claims

NON-INVASIVE HEALTH RISK SCREENING AND VALIDATION SYSTEM CLAIMS1. A computer-implemented system for non-invasive screening of a subject for one or more target health conditions, the system comprising: a physiological data acquisition module configured to obtain non-invasive physiological data from the subject using one or more sensors; a feature extraction engine configured to derive one or more physiological features from the physiological data; a condition-model registry stored in memory and comprising a plurality of condition profiles, each condition profile corresponding to a respective target health condition and associating at least one of (i) one or more physiological features, (ii) one or more abnormality patterns, (iii) one or more question sets, (iv) one or more scoring rules, (v) one or more weighting rules, (vi) one or more conflict rules, and (vii) one or more output rules; an anomaly detection engine configured to compare the one or more physiological features to one or more rules, thresholds, baselines, models, or combinations thereof and generate, for at least one candidate target health condition, a preliminary condition score; a questionnaire engine configured to select, based on the at least one candidate target health condition, a subset of questions associated with the corresponding condition profile; a response analysis engine configured to process subject responses to the subset of questions and generate a response-based score; a fusion engine configured to combine the preliminary condition score and the response-based score to generate a validated risk classification for the at least one candidate target health condition; and an output interface configured to generate a subject-facing output and a providerfacing output based on the validated risk classification.

2. The system of claim 1 , wherein the one or more target health conditions comprise one or more chronic diseases.

3. The system of claim 2, wherein the one or more chronic diseases comprise one or more endocrine diseases, metabolic diseases, cardiovascular diseases, respiratory diseases, neurologic diseases, renal diseases, hepatic diseases, sleep-related disorders, inflammatory diseases, or combinations thereof.

4. The system of claim 3, wherein the one or more endocrine or metabolic diseases comprise one or more of hypothyroidism, hyperthyroidism, type 2 diabetes, prediabetes, metabolic syndrome, primary hyperaldosteronism, adrenal insufficiency, and Cushing's syndrome.

5. The system of claim 3, wherein the one or more cardiovascular diseases comprise a cardiovascular disease category represented by a condition profile that is not limited to any single cardiovascular disease subtype.

6. The system of claim 5, wherein illustrative embodiments of the cardiovascular disease category comprise one or more of arrhythmia, atrial fibrillation, supraventricular tachycardia, premature ventricular contractions, heart failure, and sleep apnea.

7. The system of claim 1, wherein the condition-model registry is configurable to add, remove, or modify a condition profile without changing a core architecture of the physiological data acquisition module, feature extraction engine, anomaly detection engine, questionnaire engine, response analysis engine, fusion engine, or output interface.

8. The system of claim 1, wherein at least one condition profile represents a disease class or syndrome family rather than a single named diagnosis.

9. The system of claim 1, wherein the anomaly detection engine is configured to generate preliminary condition scores for a plurality of candidate target health conditions and the questionnaire engine is configured to select the subset of questions based on a ranking, thresholding, or conflict-resolution process applied to the plurality of candidate target health conditions.

10. The system of claim 1 , wherein the one or more physiological features comprise one or more of heart rate, heart rate variability, respiratory activity, pulse wave characteristics, vascular parameters, perfusion parameters, thermal parameters, facial micro-circulation parameters, activity parameters, sleep-related parameters, or combinations thereof.

11. The system of claim 1, wherein the one or more sensors comprise one or more of an RGB camera, infrared camera, photoplethysmographic sensor, wearable sensor, mobile device sensor, kiosk-based sensor, contact sensor, contactless optical sensor, or combinations thereof.

12. The system of claim 1, further comprising a gate controller configured to execute a sequential screening workflow comprising an onboarding gate, a sentinel-question gate, a physiological acquisition gate, and a questionnaire gate.

13. The system of claim 12, wherein information obtained at an earlier gate is propagated forward to influence later gate selection, score generation, question suppression, condition ranking, or output generation.

14. The system of claim 1, further comprising a pre-score registry configured to store scored items, inferred indicators, or confidence-weighted observations generated prior to completion of the questionnaire engine.

15. The system of claim 14, wherein the questionnaire engine is configured to inspect the prescore registry and suppress presentation of one or more questions whose informational contribution has already been satisfied by earlier-stage evidence.

16. The system of claim 14, wherein each registry entry comprises one or more of a stage identifier, a source identifier, a condition identifier, a score value, a confidence value, a direction flag, a reuse flag, a redundancy flag, or a timestamp.

17. The system of claim 1 , further comprising a conflict-resolution engine configured to identify candidate target health conditions associated with mutually inconsistent physiological directions, symptom polarities, or exclusion rules, and to suppress at least one condition pathway before questionnaire presentation.

18. The system of claim 17, wherein the conflict -resolution engine applies one or more direction flags to resolve physiologically mutually exclusive disease pairs before the questionnaire engine presents further questions.

19. The system of claim 1, wherein the fusion engine applies a weighted average, weighted linear combination, probabilistic model, logistic model, machine-learning model, or hybrid fusion model to combine evidence from at least two stages of the screening workflow.

20. The system of claim 19, wherein the fusion engine applies configurable stage weights across onboarding, sentinel questions, physiological acquisition, and questionnaire response data.

21. The system of claim 1, wherein the output interface comprises a dual-register output architecture including a clinician register configured to generate a provider-facing risk analysis output and a subject register configured to generate a subject -facing wellness-language output from a common computational state.

22. The system of claim 21, wherein the provider-facing risk analysis output comprises one or more of condition ranking, abnormality attribution, score breakdown, confidence indicator, referral rationale, or recommended follow-up.

23. The system of claim 21, wherein the subject-facing wellness-language output comprises one or more of lifestyle guidance, educational guidance, self-monitoring prompts, adherence encouragement, wellness coaching prompts, or referral encouragement.

24. The system of claim 23, wherein the subject-facing wellness-language output is framed in compliance with the FDA General Wellness Policy for low -risk products.

25. The system of claim 1 , further comprising a referral -prompt module configured to generate a subject-facing referral prompt based on risk tier, condition class, symptom context, persistence over time, or combinations thereof.

26. The system of claim 25, further comprising a personalized wellness content graph configured to map the validated risk classification to one or more subject-facing content pathways associated with a corresponding disease class, syndrome family, or wellness objective.

27. The system of claim 1, wherein the validated risk classification comprises a classification selected from low risk, intermediate risk, suspected risk, elevated risk, high risk, or combinations thereof.

28. A computer-implemented method for non-invasive screening of a subject for one or more target health conditions, the method comprising: obtaining non-invasive physiological data from the subject; extracting one or more physiological features from the physiological data; comparing the one or more physiological features to a plurality of condition profiles corresponding to a plurality of target health conditions; generating one or more preliminary condition scores for at least a subset of the plurality of target health conditions; selecting, based on at least one preliminary condition score, a subset of questions associated with at least one corresponding target health condition; receiving subject responses to the subset of questions; generating aresponse-based score from the subject responses; combining the preliminary condition score and the response-based score to generate a validated risk classification; and outputting a subjectfacing result and a provider-facing result.

29. The method of claim 28, wherein the plurality of target health conditions comprises chronic diseases including endocrine diseases, metabolic diseases, and cardiovascular diseases.

30. The method of claim 29, wherein the endocrine or metabolic diseases comprise one or more of hypothyroidism, hyperthyroidism, type 2 diabetes, prediabetes, metabolic syndrome, primary hyperaldosteronism, adrenal insufficiency, and Cushing's syndrome.

31. The method of claim 29, wherein the cardiovascular diseases comprise a cardiovascular disease category and one or more illustrative embodiments selected from arrhythmia, atrial fibrillation, supraventricular tachycardia, premature ventricular contractions, heart failure, and sleep apnea.

32. The method of claim 28, further comprising executing, in sequence, an onboarding gate, a sentinel-question gate, a physiological acquisition gate, and a questionnaire gate, wherein information obtained at an earlier gate is propagated forward to reduce later questioning.

33. The method of claim 28, further comprising storing one or more scored items in a pre-score registry before completion of the questionnaire gate and suppressing one or more subsequent questions based on the stored scored items.

34. The method of claim 28, further comprising identifying candidate target health conditions associated with mutually inconsistent physiologic directions or symptom polarities and resolving the candidate target health conditions based on direction flags before questionnaire presentation.

35. The method of claim 28, wherein the subject-facing result comprises a wellness-language output framed in compliance with the FDA General Wellness Policy for low-risk products and the provider-facing result comprises a clinician-facing risk analysis output derived from the same underlying computation.

36. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform the method of claim 28.

37. A computer-implemented data structure stored in non-transitory memory for controlling adaptive screening of a subject for one or more target health conditions, the data structure comprising a pre-score registry having a plurality of entries, each entry comprising: a stage identifier; a source identifier; a condition identifier; a score value; a confidence value; a direction flag; and at least one of a reuse flag, a redundancy flag, or a timestamp; wherein the pre-score registry is accessible by a questionnaire engine to suppress redundant later-stage questioning.

38. The data structure of claim 37, wherein the condition identifier corresponds to a disease class, named disease, syndrome family, risk state, or physiologic abnormality pattern.

39. The data structure of claim 37, wherein the direction flag is used by a conflict-resolution engine to suppress one of a pair of mutually inconsistent disease pathways before a questionnaire is presented.

40. A computer-implemented output generation system comprising: a common computational state generated from a non-invasive health screening workflow; a clinician register configured to transform the common computational state into a provider-facing output including at least one of abnormality attribution, condition ranking, score breakdown, confidence indicator, or referral rationale; and a subject register configured to transform the common computational state into a subject-facing wellness-language output including at least one of lifestyle guidance, educational guidance, self-monitoring prompts, or referral encouragement, without requiring separate data collection from that used to generate the provider-facing output.

41. The output generation system of claim 40, wherein the subject-facing wellness-language output is generated in a form compliant with the FDA General Wellness Policy for low-risk products.

42. The output generation system of claim 40, further comprising a personalized wellness content graph configured to select content pathways based on condition class, symptom context, validated risk tier, or persistence over time.End of Claims