US-20260128141-A1 - INTELLIGENT HEALTH MANAGEMENT SYSTEMS AND METHODS WITH REAL-TIME DATA INTEGRATION AND PERSONALIZED REPORTING

US20260128141A1US 20260128141 A1US20260128141 A1US 20260128141A1US-20260128141-A1

Abstract

A method for intelligent health management with real-time data integration and personalized reporting includes the steps of: collecting user-specific data from a plurality of sources, wherein the user-specific data comprises biometric data, clinical data, psychological data, lifestyle data, and social support data; processing the collected data using a hybrid AI engine that integrates quantitative analytical methods with large language models (LLMs); generating a reconfigurable hierarchical decision matrix that dynamically incorporates both qualitative and quantitative decision-making processes; producing personalized health reports through an output module, wherein the reports comprise actionable insights, risk assessments, and individualized recommendations tailored to user preferences and medical history; and, automatically reconfiguring decision guidelines and thresholds in real-time based on, new medical research data; user-specific health variations; and, healthcare provider inputs and feedback.

Inventors

Kian Hong Quah
Wee Lip NG

Assignees

Kian Hong Quah
Wee Lip NG

Dates

Publication Date: 20260507
Application Date: 20251107
Priority Date: 20241107

Claims (16)

1 . A method for intelligent health management with real-time data integration and personalized reporting, said method comprising the steps of: collecting user-specific data from a plurality of sources, wherein said user-specific data comprises biometric data, clinical data, psychological data, lifestyle data, and social support data; processing said collected data using a hybrid AI engine that integrates quantitative analytical methods with large language models (LLMs), wherein said hybrid AI engine, applies natural language processing to extract contextual insights from unstructured data; employs pattern recognition algorithms to perform real-time analysis of health metrics; and, implements explicit decision guidelines to mitigate AI hallucinations; generating a reconfigurable hierarchical decision matrix that dynamically incorporates both qualitative and quantitative decision-making processes, wherein said decision matrix, maps user inputs to health risk factors using binary relevance indicators; integrates new health metrics and prognostic factors; and, evolves based on advancements in healthcare knowledge; producing personalized health reports through an output module, wherein said reports comprise actionable insights, risk assessments, and individualized recommendations tailored to user preferences and medical history; and, automatically reconfiguring decision guidelines and thresholds in real-time based on, new medical research data; user-specific health variations; and, healthcare provider inputs and feedback.
2 . The method of claim 1 , wherein said plurality of parameters further comprises one or more of laboratory test results, blood test results, urine test results, stool test results, imaging reports, scan reports, endoscopy results, DNA testing results, genetic testing results, sleep study data, electrophysiological testing data, histopathology reports, and cytology reports.
3 . The method of claim 1 , wherein said processing step further comprises: cross-validating outputs from said large language models with said quantitative analytical methods to reduce AI overgeneralization; aggregating multiple health risk factors and prognostic metrics into a structured summary; and modifying decision pathways based on user-specific preferences and healthcare provider inputs to customize recommendations for individual physiological and psychological profiles.
4 . The method of claim 1 , further comprising: adjusting said individualized recommendations using a personalization module based on individual user preferences, prior health history, and lifestyle factors; and providing a review system for healthcare providers to review and approve AI-generated reports, wherein said review system includes an interface for provider feedback to enable additional adjustments and ensure compliance with medical standards.
5 . The method of claim 1 , wherein said reconfigurable hierarchical decision matrix includes a feedback loop that captures user responses and incorporates new data points into future assessments for continuous improvement and adaptation of decision criteria based on real-time health insights and advancements in medical research.
6 . The method of claim 1 , further comprising: generating an automated follow-up plan that provides users with reminders, alerts, and suggested actions based on said personalized health reports; and adapting said follow-up plan to changes in user health metrics to modify suggestions accordingly.
7 . The method of claim 1 , further comprising: incorporating a weighted decision criteria process to mitigate AI hallucinations at each stage of analysis, wherein each criterion is associated with a confidence score for enhancing accuracy and reliability of AI-generated health assessments; and quantifying reliability and confidence levels of AI-generated insights, wherein said reconfigurable hierarchical decision matrix includes a scoring system based on real-time physiological data providing users with metrics representing accuracy of medical recommendations.
8 . The method of claim 1 , further comprising: collecting real-time biometric data and user preferences related to body sculpting using a non-invasive body contouring recommendation module; analyzing said collected biometric data using said hybrid AI engine to generate tailored body contouring recommendations; and dynamically adjusting said body contouring recommendations based on updated user data and progress measurements.
9 . An intelligent health management system with real-time data integration and personalized reporting, said system comprising: a data collection module configured to gather user-specific data from a plurality of sources, wherein said user-specific data comprises biometric data, clinical data, psychological data, lifestyle data, and social support data; a hybrid AI engine that integrates quantitative analytical methods with large language models (LLMs), wherein said hybrid AI engine is configured to apply natural language processing to extract contextual insights from unstructured data, employ pattern recognition algorithms to perform real-time analysis of health metrics, and implement explicit decision guidelines to mitigate AI hallucinations; a reconfigurable hierarchical decision matrix that dynamically incorporates both qualitative and quantitative decision-making processes, wherein said decision matrix is configured to map user inputs to health risk factors using binary relevance indicators, integrate new health metrics and prognostic factors, and evolve based on advancements in healthcare knowledge; an output module configured to produce personalized health reports, wherein said reports comprise actionable insights, risk assessments, and individualized recommendations tailored to user preferences and medical history; and a reconfiguration mechanism configured to automatically reconfigure decision guidelines and thresholds in real-time based on new medical research data, user-specific health variations, and healthcare provider inputs and feedback.
10 . The system of claim 9 , wherein said plurality of parameters further comprises one or more of laboratory test results, blood test results, urine test results, stool test results, imaging reports, scan reports, endoscopy results, DNA testing results, genetic testing results, sleep study data, electrophysiological testing data, histopathology reports, and cytology reports.
11 . The system of claim 9 , wherein said hybrid AI engine is further configured to cross-validate outputs from said large language models with said quantitative analytical methods to reduce AI overgeneralization, aggregate multiple health risk factors and prognostic metrics into a structured summary, and modify decision pathways based on user-specific preferences and healthcare provider inputs to customize recommendations for individual physiological and psychological profiles.
12 . The system of claim 9 , further comprising a personalization module configured to adjust said individualized recommendations based on individual user preferences, prior health history, and lifestyle factors, and a review system for healthcare providers configured to review and approve AI-generated reports, wherein said review system includes an interface for provider feedback to enable additional adjustments and ensure compliance with medical standards.
13 . The system of claim 9 , wherein said reconfigurable hierarchical decision matrix includes a feedback loop configured to capture user responses and incorporate new data points into future assessments for continuous improvement and adaptation of decision criteria based on real-time health insights and advancements in medical research.
14 . The system of claim 9 , further comprising an automated follow-up plan generator configured to provide users with reminders, alerts, and suggested actions based on said personalized health reports, and to adapt said follow-up plan to changes in user health metrics to modify suggestions accordingly.
15 . The system of claim 9 , further comprising a weighted decision criteria process configured to mitigate AI hallucinations at each stage of analysis, wherein each criterion is associated with a confidence score for enhancing accuracy and reliability of AI-generated health assessments, and wherein said reconfigurable hierarchical decision matrix includes a scoring system based on real-time physiological data providing users with metrics representing accuracy of medical recommendations.
16 . The system of claim 9 , further comprising a non-invasive body contouring recommendation module configured to collect real-time biometric data and user preferences related to body sculpting, analyze said collected biometric data using said hybrid AI engine to generate tailored body contouring recommendations, and dynamically adjust said body contouring recommendations based on updated user data and progress measurements.

Description

CROSS-REFERENCE TO RELATED APPLICATION This application claims the benefit of and priority to Malaysian patent application no. PI2024006397 filed on 7 Nov. 2024, which are herein incorporated by reference in its entirety. BACKGROUND OF INVENTION As it is well known, over the past decade, the integration of digital technologies in healthcare has led to substantial improvements in diagnostics, remote monitoring, and preventive care. The modern wearable devices now provide continuous monitoring capabilities, as these devices collect physiological data including heart rate, oxygen saturation, sleep patterns, and physical activity levels, enabling real-time health insights and personalized treatment recommendations. Also, the healthcare providers increasingly rely on decision support systems powered by artificial intelligence to analyze electronic health records (EHRs), imaging data, and diagnostic results. Generally, AI technologies in medicine exist in many forms, from purely virtual systems for health information management to cyber-physical systems used to assist surgeons and provide active guidance in treatment decisions. The contemporary AI-based decision engines often employ machine learning and large language models (LLMs) to generate insights from structured and unstructured datasets. The large language models have shown promise in clinical contexts, with applications ranging from clinical note generation to medical text summarization, though they face significant challenges including hallucination rates of 1.47% in some implementations. In this aspect, the medical hallucinations are defined as instances where models generate misleading medical content, including incorrect dosages, drug interactions, or diagnostic criteria that can directly lead to life-threatening outcomes. In addition, LLM-augmented clinical decision support systems have been tested in randomized controlled trials, with participants expressing concerns that “the chatbot will hallucinate, which is particularly bad in medicine,” highlighting significant trust and reliability issues. Herein, the wearable medical devices and complementary applications provide medical professionals with holistic pictures of patients' health states, enabling continuous monitoring for chronic conditions such as diabetes, cardiovascular disease, and respiratory disorders. The integration of artificial intelligence within wearable devices is revolutionizing healthcare by enabling proactive management through real-time feedback and comprehensive health monitoring. However, current systems face significant challenges in real-time data integration. Issues related to data quality, interoperability, health equity, and fairness have been identified as major concerns, with many systems struggling to handle continuous, multi-source input streams effectively. The existing approaches are often restricted to static, limited-domain datasets and are not generalizable across various healthcare scenarios. Furthermore, existing AI healthcare solutions demonstrate several technical and practical limitations that constrain their broader utility in clinical and consumer-facing environments. Traditional decision trees used by physicians are effectively tied to initial tree structures and are thus somewhat static, while deep learning models are less easily interpretable and may make it difficult to establish causal links. As it has been observed, AI-based clinical tools face accountability and safety challenges, with systems potentially ignoring previous states when making decisions and going against usual clinical practice by recommending sudden changes in treatment protocols. Consequently, current decision support systems often lack systematic approaches to AI implementation, resulting in niche roles rather than comprehensive integration. In addition to the above, AI-generated outputs, particularly those involving large language models, are susceptible to generating incorrect or unsupported information, commonly referred to as “AI hallucinations,” which may lead to misinterpretations of clinical significance or inappropriate treatment recommendations. The recent studies show that large language models repeat or elaborate on planted errors in up to 83% of cases in clinical vignettes, with simple mitigation prompts halving the rate but not eliminating the risk, as published in Communications Medicine's study titled “Multi-model assurance analysis showing large language models are highly vulnerable to adversarial hallucination attacks during clinical decision support”. The phenomenon of hallucination, where LLMs generate outputs that deviate from factual accuracy or context, poses critical challenges in high-stakes healthcare domains, requiring specialized techniques for mitigation including retrieval-augmented generation, iterative feedback loops, and supervised fine-tuning. The lack of transparent mechanisms to validate or cross-check these outputs against clinical protocols exac