Adaptive, Internet of Medical Things-based healthcare system for proactive monitoring and support

Abstract

A health monitoring system for the autonomous support of elderly people, the system comprising the following: a variety of Internet of Medical Things (IoMT) sensors, including at least one wearable physiological sensor, one or more environmental sensor units, a mobility tracking device and a medication intake module, wherein the IoMT sensors are configured to continuously collect physiological, environmental, behavioral and medication use data; an edge processing device that includes an embedded processor and a lightweight machine learning inference unit configured to perform real-time anomaly detection and generate local alerts in the absence of cloud connectivity; a cloud-based analytics engine comprising one or more deep learning models configured to perform longitudinal health modeling, population-level clustering, and predictive risk forecasting using historical and real-time data; and a multi-channel communication unit configured to securely transmit notifications, reports, and emergency triggers to nursing staff, clinical staff, family members, and emergency services, wherein the edge processing device includes a dedicated hardware accelerator selected from a field-programmable gate array (FPGA), the accelerator being optimized to perform low-latency computations, including fall detection, arrhythmia detection, and respiratory abnormality monitoring, within a processing window of less than 200 milliseconds;and wherein the wearable physiological sensor is designed as a wrist-worn or patch-based device comprising a set of sensor elements, including an electrocardiography (ECG) electrode, a photoplethysmography (PPG) sensor for SpO2 measurement, a three-axis accelerometer, a gyroscope, and a skin temperature probe, the device further being configured to integrate raw signal data streams through a multimodal fusion technique performed locally on the edge device to minimize noise, compensate for motion artifacts, and improve the specificity of anomaly detection across multiple concurrent parameters.

Description

Technical field of the invention

[0001] The present invention relates to the field of health monitoring systems. More specifically, it relates to an adaptive, Internet of Medical Things (IoMT)-enabled device and system architecture for the continuous, autonomous, and context-aware monitoring of elderly people in independent living environments. The invention integrates distributed sensors, adaptive AI analytics, and edge cloud orchestration to enable proactive, real-time support and multi-channel stakeholder communication while ensuring data privacy and security. Background of the invention

[0002] The demographic shift towards an aging population presents significant challenges for healthcare, particularly regarding ensuring the safety, health, and independence of older adults without constant care. Conventional systems offer fragmented monitoring solutions limited to specific physiological parameters or environmental factors. Most current devices either focus on single measurements such as heart rate or blood glucose levels or offer reactive alert mechanisms that require manual intervention. Furthermore, these systems often lack the adaptive learning capabilities, interoperability with medical records, and robust data privacy frameworks necessary for practical application.

[0003] Therefore, there is an urgent need for a unified, adaptive healthcare ecosystem that autonomously integrates multimodal data streams, performs predictive analytics, and provides nurses, clinicians, and emergency responders with actionable insights in real time. The invention described here overcomes these limitations through a device-integrated, scalable IoMT ecosystem that enables context-sensitive monitoring, proactive alerting, and seamless integration into healthcare infrastructure.

[0004] The healthcare industry has long recognized the growing challenge of effectively monitoring and supporting older adults, particularly those who wish to live independently. With the increasing aging of the global population and the steadily rising proportion of people over 65, research and commercial product development aimed at addressing the health and safety needs of older adults are accelerating. Despite significant progress, existing solutions have limitations that prevent them from meeting the complex, real-world demands of elderly care. A closer look at the technological landscape reveals how current systems have evolved, the approaches they employ, and why their shortcomings necessitate the development of a more adaptive and integrated solution, such as the present invention.

[0005] One of the first classes of solutions consisted of wearable devices for monitoring individual parameters. Products such as wrist-worn heart rate monitors, pedometers, and simple fall detection pendants were among the first attempts to enable older people to live independently with the help of technology. These devices typically use accelerometers, gyroscopes, or optical sensors to measure parameters such as steps taken, heart rate, or sudden changes in movement that might indicate a fall. Although these devices are affordable and widely available, their use is severely limited. They often generate false alarms, for example, by mistaking abrupt sitting movements for falls, and they cannot take context into account, such as whether a detected event actually indicates a critical medical emergency.Furthermore, the fact that the devices depend on regular charging, wearing, and manually triggering of help signals by the user hinders acceptance among older people who may have difficulty handling additional devices in everyday life.

[0006] A parallel category of solutions is home monitoring systems, often marketed as "aging-in-place" technologies. These systems typically include environmental sensors such as motion detectors, pressure-sensitive mats, or door sensors to detect movement within the living space. Some more advanced versions integrate video cameras or infrared sensors to detect unusual patterns of inactivity. While such systems can provide valuable insights into daily routines, they have several limitations. First, they are environment-dependent, meaning their effectiveness is limited to the home environment and does not extend to outdoor areas or when traveling. Second, they often lack sophisticated data fusion mechanisms, making it difficult to distinguish between harmless anomalies—such as prolonged bed rest—and serious emergencies.Finally, video surveillance raises privacy concerns that deter many older people from accepting such solutions in their homes.

[0007] Integrated telehealth platforms have gained prominence in recent years. These platforms typically rely on regular self-reporting by older adults, often via connected devices such as blood pressure cuffs, blood glucose meters, or digital scales. The data is transmitted to a cloud-based system where clinicians can review the information and adjust treatment plans accordingly. While these solutions represent an advancement in remote care, they remain largely reactive rather than proactive. Data is collected only sporadically, meaning that sudden adverse events such as arrhythmias, falls, or environmental hazards may go undetected between scheduled measurements. Furthermore, the reliance on user input when operating medical devices introduces human error, and compliance tends to decline over time as older adults forget or become fatigued by repetitive tasks.

[0008] Furthermore, medication adherence solutions have been introduced, ranging from simple reminder apps to smart pillboxes that track intake behavior. These technologies attempt to address one of the most pressing problems in elderly care: ensuring patients take their medications correctly and on time. However, the majority of these systems remain isolated, focusing solely on adherence without being integrated into broader health monitoring. While they may indicate whether a pillbox has been opened at the correct time, they cannot confirm administration or correlate medication intake with physiological outcomes such as improved vital signs or reduced symptoms.Although these devices can improve compliance rates to some extent, they do not provide the holistic health insights needed for effective, autonomous support of older people.

[0009] Artificial intelligence, in the form of anomaly detection techniques, predictive analytics, and natural language interfaces, is increasingly finding its way into elderly care. Commercial solutions such as AI-powered fall detection systems or predictive health platforms promise smarter and more adaptive monitoring. However, most of these implementations remain narrowly focused, applying AI to isolated data streams instead of achieving true multimodal integration. For example, an AI-powered wearable can predict fall risk based on gait patterns, but these patterns do not correlate with environmental hazards such as slippery floors or poorly lit rooms. Similarly, predictive health platforms can analyze the progression of chronic diseases, but are limited by sparse, self-reported data rather than continuous sensor data.This fragmentation underlines the lack of a unified ecosystem that integrates the various dimensions of older people's health and environment into a coherent monitoring and support framework.

[0010] Another critical drawback of existing systems is the lack of edge cloud orchestration. Traditional healthcare IoT solutions are often either entirely cloud-dependent or purely on-premises. While cloud-based systems are capable of performing extensive data analysis and longitudinal pattern recognition, they introduce latency and depend on a stable internet connection, which can delay emergency responses. On the other hand, purely on-premises systems lack the computing power and data resources to perform predictive modeling or population-level learning. This dichotomy leads to either delayed interventions or superficial analysis, neither of which is sufficient to protect older adults in high-risk scenarios. A hybrid approach that combines real-time edge processing with advanced cloud-based analytics is needed but is rarely implemented in commercially available solutions.

[0011] Security and data protection remain persistent challenges. Surveillance in elderly care naturally involves sensitive health data that must be protected by legal frameworks such as HIPAA and GDPR. Many existing systems lack a comprehensive privacy-by-design architecture. Data is often stored or transmitted without sufficient encryption, or third parties may have uncontrolled access to user information. Video-based surveillance systems are particularly invasive and raise ethical concerns regarding the monitoring of vulnerable individuals. Furthermore, the explainability of AI models is rarely prioritized, leaving caregivers and physicians unclear about the reasons for warnings or recommendations. This lack of transparency undermines trust and hinders acceptance.

[0012] Regarding interoperability, current solutions are typically proprietary and siloed, creating barriers to integration with third-party electronic health records (EHRs), telemedicine platforms, or healthcare systems. For example, while a wearable device may provide useful heart rate data, it cannot seamlessly share this information with a physician's dashboard or integrate it with a patient's existing medical record. Similarly, environmental monitoring systems often operate in isolation, without coordination with wearable or medication-administering technologies. The lack of open APIs, standards-based design, and modular architectures has severely limited the scalability and sustainability of existing products.

[0013] From a user-friendliness perspective, many solutions fail to consider the accessibility needs of older adults. User interfaces can be overly complex, requiring navigation through smartphone apps or dashboards that are difficult for seniors to understand. Devices can be bulky, uncomfortable to carry, or require frequent charging, impacting compliance and reliability. As a result, even technically sophisticated solutions often achieve low long-term acceptance due to a lack of adaptation to the daily habits, cognitive abilities, and physical limitations of older users.

[0014] Finally, the economic aspect must not be overlooked. Many modern systems are prohibitively expensive and restrict access to wealthy users or those with comprehensive health insurance. Subscription-based telemedicine platforms often incur ongoing costs that may be unsustainable for pensioners with a stable income. Furthermore, due to the fragmentation of current solutions, families often have to purchase multiple devices and services to even partially meet their needs. This results in a disjointed, costly, and inefficient system of elderly care.

[0015] In summary, existing solutions for the healthcare of older adults are limited by their isolated functionality, their reliance on cloud or local processing, a lack of interoperability, and insufficient consideration of data privacy, accessibility, and cost-effectiveness. They often fail to provide continuous, context-aware monitoring, leaving older adults vulnerable to undetected emergencies and a gradual decline in their health. The absence of a modular, scalable, and adaptive system that integrates multimodal data streams into a unified, proactive support ecosystem represents a critical gap in this field.These drawbacks highlight the urgent need for an invention that uniquely combines IoMT-based sensing, adaptive AI analytics, edge cloud orchestration, and privacy-friendly design into a comprehensive ecosystem tailored to enable dignified, independent living for older people. Objectives of the invention

[0016] The main objective of the invention is to provide a device and system that, through the integration of IoMT sensors, adaptive analytics, and automated response mechanisms, ensures proactive healthcare for elderly individuals living independently. A further objective is to create a modular and scalable ecosystem that enables seamless interoperability with external healthcare systems such as electronic health records (EHRs) and telemedicine platforms. Another objective is to enable context-sensitive, self-learning anomaly detection that minimizes false alarms while delivering timely, actionable alerts. Furthermore, the invention aims to integrate privacy-by-design principles that ensure user consent, encryption, and compliance with legal frameworks such as HIPAA and GDPR.

[0017] The objectives of the present invention focus on addressing the unresolved challenges in the health monitoring of older adults. This is achieved by creating a unified, adaptive, and secure ecosystem that enables a dignified, independent life while ensuring safety and well-being. A key objective of the invention is to provide a device and system that integrates multimodal IoMT sensors, including wearable, environmental, mobility, and medication intake modules, to deliver a continuous stream of health, behavioral, and contextual data for comprehensive monitoring. Another objective is the integration of adaptive artificial intelligence that learns from individual user patterns and longitudinal data, thereby providing predictive, context-sensitive anomaly detection and risk forecasting that minimizes false alarms and enables timely interventions.The invention also aims to establish an orchestrated edge-cloud computing framework where the edge layer guarantees low-latency emergency detection and immediate local response, while the cloud layer enables in-depth analytics, population-level learning, and longitudinal trend detection. A further objective is to ensure proactive, multi-channel alerting that seamlessly and in real time delivers actionable insights to caregivers, physicians, family members, and emergency services, thus closing the loop between monitoring, detection, and intervention. Furthermore, the invention aims to maintain user trust and ensure regulatory compliance through the integration of privacy-by-design principles such as end-to-end encryption, consent-based access, and explainable AI mechanisms.Another important goal is to create a modular, scalable, and standards-based architecture that can evolve with technological advancements and enables seamless interoperability with electronic health records, telemedicine platforms, and third-party devices. Finally, the invention aims to improve usability and accessibility through intuitive interfaces and ergonomic device designs, ensuring that older adults can use the system comfortably while receiving dignified, autonomous support in their daily lives. Summary of the invention

[0018] The invention discloses a multi-layered healthcare system consisting of distributed IoMT-enabled sensor devices, an adaptive AI engine, an edge-cloud framework for computational orchestration, and a multi-stakeholder communication module. Physiological, behavioral, environmental, and medication-related data are continuously collected and pre-processed by local edge devices. These edge devices perform anomaly detection and generate low-latency alerts, while cloud servers conduct longitudinal analyses, predictive modeling, and large-scale optimizations.

[0019] The AI ​​engine dynamically adapts to individual user routines and historical health data, providing personalized anomaly detection and predictive alerts. A modular architecture ensures interoperability with third-party devices, electronic health records, and telemedicine platforms. Data protection and security mechanisms are integrated at all levels, including encryption, user-controlled consent, and explainable, AI-driven audit trails. The invention is implemented both as a wearable monitoring device and as an integrated, home-embedded IoMT structure, forming a comprehensive ecosystem for the healthcare of the elderly. BRIEF DESCRIPTION OF THE FIGURE

[0020] These and other features, aspects, and advantages of the present invention will be better understood if the following detailed description is read with reference to the accompanying drawing, in which the same symbols consistently represent the same parts. The following applies: Fig. Figure 1 shows a block diagram of an adaptive AI-IoMT health ecosystem for proactive monitoring and support of elderly people.

[0021] Experts will also recognize that the elements in the drawing are shown for the sake of simplicity and are not necessarily to scale. For example, the flowcharts illustrate the process by highlighting the main steps to enhance understanding of the aspects of this disclosure. Furthermore, with regard to the design of the device, one or more components of the device may be represented in the drawing by conventional symbols, and the drawing may show only the specific details relevant to understanding the embodiments of this disclosure, so as not to clutter the drawing with details that are readily apparent to those skilled in the art after reading this description. Detailed description of the invention

[0022] For a better understanding of the inventive principles, reference is made below to the embodiment shown in the drawing, which is described in specific terminology. However, this does not limit the scope of the invention. Changes and further modifications of the illustrated system, as well as further applications of the inventive principles, are possible, as would normally occur to a person skilled in the art in the field of invention.

[0023] It is clear to the person skilled in the art that the preceding general description and the following detailed description are exemplary and explanatory of the invention and are not intended as a limitation of it.

[0024] References in this specification to “an aspect”, “another aspect”, or similar expressions mean that a particular feature, structure, or property described in connection with the embodiment is included in at least one embodiment of the present disclosure. Therefore, occurrences of the expressions “in one embodiment”, “in another embodiment”, and similar expressions in this specification may all refer to the same embodiment, but need not.

[0025] The terms "includes," "include," or other variations thereof are intended to cover non-exclusive inclusion, such that a process or method that includes a list of steps may not only contain those steps but may also include other steps not expressly listed or inherent in such process or method. Likewise, the statement "includes..." in the case of one or more devices, subsystems, elements, structures, or components does not, without further limitations, preclude the existence of other devices, subsystems, elements, structures, components, or additional devices, subsystems, elements, structures, or components.

[0026] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as understood by a person skilled in the art in the field of the invention. The systems, methods, and examples provided herein serve only for illustration and are not to be construed as limitations.

[0027] Embodiments of the present disclosure are described in detail below with reference to the attached drawing.

[0028] In Fig.Figure 1 shows a block diagram of an adaptive AI-IoMT healthcare ecosystem for proactively monitoring and supporting older adults in maintaining their independence. The system 100 comprises: several Internet of Medical Things (IoMT) sensors (102), including at least one wearable physiological sensor (102a), one or more environmental sensor units, a mobility tracking device, and a medication intake module (102b), with the IoMT sensors configured to continuously collect physiological, environmental, behavioral, and medication intake data; an edge processing device (104) with an embedded processor and a lightweight machine learning inference unit configured to detect anomalies in real time and generate local alerts in the absence of cloud connectivity;a cloud-based analytics engine (106) with one or more deep learning models configured to perform longitudinal health modeling, population-level clustering, and predictive risk forecasting using historical and real-time data; and a multi-channel communications unit (108) configured to securely transmit notifications, reports, and emergency triggers to caregivers, physicians, family members, and emergency services.

[0029] The hardware support for the health monitoring system, designed to provide autonomous assistance to the elderly, can be implemented through a network of medical sensor and computer components integrated into a secure, continuously operating structure. The wearable physiological sensor can include miniature electronic circuits for measuring pulse, body temperature, respiratory rate, and movement, all connected to a low-power control unit with a rechargeable energy source. The environmental sensors can comprise physical components for measuring air quality, room temperature, humidity, and ambient light, each housed in a sealed protective casing and connected to a data transmission circuit for continuous communication.The mobility tracking device can include a position receiver, motion sensors, and a local computer controller that continuously monitors the user's movement and orientation. The medication dispensing module can consist of a medication storage compartment equipped with weight or contact sensors to detect dispensing, as well as a control circuit to record medication consumption. All these sensor components are operationally connected to an edge processing unit containing a compact processor, internal memory, and a machine learning inference circuit. This unit processes sensor data locally, performs anomaly detection, and generates immediate alerts when no network connection to the cloud is available. The cloud-based analytics engine is implemented on high-performance servers with multiple processors, large amounts of RAM, and ample storage.It is configured to perform deep learning operations for modeling health trends, behavioral clustering, and predictive analytics using historical and real-time data. The multi-channel communication unit includes dedicated hardware for the secure transmission of data and alerts via SMS, voice call, email, and emergency call center. This ensures that caregivers, medical professionals, family members, and emergency responders receive reliable and timely information. Together, this hardware configuration enables continuous health monitoring, secure data transmission, intelligent decision-making, and autonomous operational support for older adults.

[0030] In one embodiment, the edge processing device (104) comprises a dedicated hardware accelerator selected from a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), or a neural processing unit (NPU). The accelerator is optimized to perform low-latency computations, including fall detection, arrhythmia detection, and respiratory abnormality monitoring, within a processing window of less than 200 milliseconds, ensuring that emergency alerts can be generated without delay even in the event of a network outage or disconnection from the cloud.

[0031] In one embodiment, the wearable physiological sensor (102a) is designed as a wrist-worn or patch-based device comprising a suite of sensor elements, including an electrocardiography (ECG) electrode, a photoplethysmography (PPG) sensor for SpO2 measurement, a three-axis accelerometer, a gyroscope, and a skin temperature probe. The device is also configured to integrate raw signal data streams through a multimodal fusion technique performed locally on the edge device to minimize noise, compensate for motion artifacts, and improve anomaly detection specificity across multiple simultaneous parameters.

[0032] In one embodiment, the cloud-based analytics engine (106) comprises a recurrent neural network (RNN) architecture with long-term short-term memory (LSTM) units configured to capture temporal dependencies in health data of older adults. The LSTM units are also integrated into a clustering mechanism, such as K-means or Gaussian mixture modeling, to group individuals into population-level cohorts for comparative risk profiling, and adaptive thresholds for anomaly detection are dynamically recalibrated based on cohort-level baseline values ​​and the individual user's historical health trajectory.

[0033] In one embodiment, the medication adherence module (102b) comprises an intelligent dispensing structure configured with load cells, RFID (Radio Frequency Identification) readers, and time-synchronized dispensing compartments. The module is also configured to verify adherence by correlating tablet removal with concurrent physiological changes such as heart rate variability or blood pressure stabilization. This enables closed-loop verification of medication adherence and eliminates reliance on self-reporting.

[0034] In one embodiment, the multi-channel communication unit (108) is configured to execute a hierarchical alerting protocol, whereby initial anomaly detection events trigger low-priority alerts to family members and caregivers via mobile notifications, while events classified as serious by the AI ​​engine, such as cardiac arrest, hypoxemia, or confirmed falls resulting in immobility, automatically trigger a high-priority emergency escalation, which simultaneously sends an SOS signal to the emergency medical services, activates voice-based auto-call functions, and generates a secure clinical report that is transmitted to a pre-registered physician dashboard.

[0035] In one embodiment, the system includes a data protection and security module with end-to-end encryption based on the Advanced Encryption Standard (AES) with 256-bit keys for all transmitted data, a blockchain-based, immutable audit log that records all data accesses and system decisions to ensure compliance with regulations, and an explainable AI layer configured to generate human-interpretable justifications for each anomaly classification or predictive risk output, ensuring that nurses and clinicians can understand and validate the decision-making process of the AI ​​analysis engine.

[0036] In one embodiment, the system architecture is modular and standards-based, allowing new IoMT sensors to be dynamically integrated via open application programming interfaces (APIs) compatible with the HL7 FHIR (Fast Healthcare Interoperability Resources) standards. This modularity enables seamless expansion of the ecosystem to include additional clinical workflows such as telerehabilitation, remote physiotherapy monitoring, or integration with hospital electronic health record systems, without requiring any changes to the underlying system infrastructure.

[0037] The system described in the claims above is realized through a multi-layered architecture that integrates distributed sensing, adaptive computing, and intelligent communication into a single ecosystem optimized for the healthcare of older adults. The system is based on a collection of IoMT sensors that continuously acquire diverse data streams from multiple modalities, including physiological signals, environmental conditions, mobility patterns, and medication adherence. These sensors provide raw and semi-processed data that represent the real-time condition of the older person and their environment. For example, the wearable device integrates an ECG electrode, SpO2 photoplethysmography, a three-axis accelerometer, a gyroscope, and a skin temperature sensor.The data from these sources are inherently heterogeneous, noisy, and prone to environmental artifacts such as motion disturbances or interference from ambient light. To address this, a multimodal fusion technique is performed locally on the edge processing device. This technique synchronizes timestamps across sensor inputs, filters noise through adaptive thresholding, and applies statistical smoothing to generate clean, contextually reliable feature sets for anomaly detection.

[0038] The edge device is the first intelligent computing layer and is designed to ensure low-latency operation during critical events. Integrated into the edge hub is a lightweight inference engine that uses a neural processing unit or an FPGA accelerator to detect anomalies within strict latency limits of less than 200 milliseconds. Anomaly detection is based on a hybrid approach that combines statistical thresholds with machine learning classification. For example, accelerometer and gyroscope data are processed using a threshold-based fall detection technique that calculates the resulting acceleration vectors and compares them to predefined fall velocity thresholds. To reduce false alarms, the technique incorporates contextual verification.It analyzes post-event immobility using accelerometer data, distinguishing genuine falls from harmless movements like sudden landings. Similarly, arrhythmia detection utilizes ECG signals processed through a short-term Fourier transform (STFT), combined with lightweight convolution filters trained to identify irregular RR intervals indicative of atrial fibrillation or tachycardia. Respiratory abnormalities are identified using frequency analysis of chest motion data in conjunction with PPG-based oxygen saturation measurements. These edge-level techniques are optimized for execution on limited hardware, ensuring that critical health abnormalities trigger immediate local alerts even without cloud connectivity.

[0039] While the edge device ensures real-time responsiveness, the cloud analytics engine enhances the system's intelligence by processing longitudinal data and performing predictive modeling. The cloud utilizes a recurrent neural network (RNN) architecture with long short-term memory (LSTM) units, specifically designed to capture temporal dependencies in health and behavioral data. The LSTM network is trained to detect deviations in long-term patterns, such as progressive deterioration of gait stability, a gradual decrease in resting heart rate variability, or consistent sleep cycle abnormalities. By storing historical data from individual users, the cloud engine can identify subtle health risks that, while not triggering edge alerts, may indicate the onset of chronic conditions.In addition to this individualized focus, the cloud system uses clustering techniques such as K-means or Gaussian mixture modeling, which group older users into cohorts based on similar health profiles. This enables adaptive calibration of thresholds, with the system adjusting individual parameters for anomaly detection not only by referencing the user's previous data but also by comparing them to population-level baseline values ​​derived from similar users.

[0040] The system also implements a feedback mechanism that refines the predictive models over time. When an anomaly is detected and a subsequent alert is confirmed or dismissed by nurses or physicians, the event is flagged and fed back into the edge and cloud learning pipelines. At the edge, reinforcement learning updates the threshold parameters to minimize the recurrence of false positives or false negatives. At the cloud, supervised learning updates the LSTM models to improve their ability to predict risk trajectories. This adaptive, self-learning process ensures that the system does not remain static but adapts to the user's lifestyle, health status, and environmental conditions.

[0041] A crucial component of the invention is the integration of medication adherence into physiological monitoring. The intelligent dosing module uses load cells and RFID-based tablet recognition to confirm tablet removal at the scheduled time. This adherence data is compared with concurrent physiological parameters, such as blood pressure reduction after taking antihypertensive medication or improved oxygenation after inhalation. By correlating pharmacological compliance with measurable physiological effects, the system enables closed-loop validation of medication adherence. The underlying technology uses correlation analyses and regression models to determine whether medication intake achieves the expected physiological outcomes.If medication adherence is documented but no physiological effect is observed, the system generates warning messages about potential problems such as improper intake, expired medication or ineffective therapy, thus enabling timely medical intervention.

[0042] The system's communication layer is based on a hierarchical alerting protocol controlled by the AI ​​engine. Detected anomalies are categorized according to their severity using a multi-level classification system. This classification is based on a weighted risk model that combines parameters such as the extent of the deviation from the norm, the duration of the anomaly, and contextual data from environmental sensors. For example, a sudden drop in oxygen saturation lasting longer than two minutes while the person is at rest is classified as a serious event, whereas a temporary fluctuation during physical exertion might be classified as minor.Severe anomalies trigger automated escalation protocols, including generating SOS alerts to emergency services, activating voice-based autocalls for pre-registered contacts, and delivering a detailed clinical summary to physicians. The clinical summary is generated using a structured report template that includes event-specific data streams, AI inference rationales, and recommended interventions. Lower-severity events are only communicated to caregivers and family members via mobile notifications, reducing alert fatigue while maintaining vigilance.

[0043] Data privacy and security are embedded at every level of the system's computing. Data is encrypted with AES-256 during transmission and secured at rest in the cloud infrastructure using multi-layered encryption keys. The system also utilizes a blockchain-based, immutable audit trail that logs every data access, anomaly detection, and system decision. Each block entry is cryptographically signed, providing nurses and physicians with verifiable evidence of data integrity and compliance with regulations such as HIPAA and GDPR. Furthermore, the AI ​​analytics engine includes an explainable AI module that generates human-interpretable rationales for its classifications and predictions. For example, in the case of an arrhythmia alert, the system provides annotated ECG tracings highlighting the irregular RR intervals and the decision thresholds used.This allows doctors to validate the AI ​​decision before starting treatment.

[0044] The system's modularity and scalability are ensured through the use of open APIs compliant with HL7 FHIR standards. This enables seamless integration with electronic health records, telemedicine platforms, and third-party medical devices. The architecture supports plug-and-play integration of new IoMT sensors without system redesign. For example, a future blood glucose sensor can be added to the existing ecosystem, and its data stream will be automatically integrated into the fusion and anomaly detection pipeline. Similarly, new clinical workflows, such as remote physiotherapy or postoperative rehabilitation, can be integrated via cloud-based extensions.

[0045] Overall, the invention combines continuous multimodal sensing, real-time edge intelligence, adaptive cloud-based analytics, and a secure, traceable decision-making system into a coherent system. The technical backbone ensures that the system evolves with the user, responds quickly in emergencies, predicts long-term risks, monitors medication adherence, and complies with strict data protection standards. By bridging the technical gaps of existing solutions, the invention offers a proactive, autonomous ecosystem for health monitoring that is robust, scalable, and tailored to the specific challenges of independent living for older adults.

[0046] The drawing and the preceding description show examples of embodiments. Those skilled in the art will recognize that one or more of the described elements can be combined to form a single functional element. Alternatively, certain elements can be divided into several functional elements. Elements of one embodiment can be added to another embodiment. For example, the sequence of the processes described here can be changed and is not limited to the manner described here. Furthermore, the actions of a flowchart need not be implemented in the sequence shown; nor does it necessarily have to be performed by all actions. Actions that are not dependent on other actions can also be performed in parallel with the other actions. The scope of the embodiments is in no way limited by these specific examples.Numerous variations are possible, whether explicitly stated in the specification or not, such as differences in structure, dimensions, and material use. The range of embodiments is at least as broad as specified in the following claims.

[0047] Advantages, further benefits, and problem solutions have been described above with regard to specific embodiments. However, the advantages, benefits, problem solutions, and all components that may lead to a particular advantage or solution occurring or becoming more apparent are not to be construed as critical, necessary, or essential features or components of any or all claims. REFERENCES 100 An Adaptive AI-IOMT Health Ecosystem For Proactive Monitoring and Support of Elderly People in Independent Living Arrangements. 102 Sensors for the Internet of Medical Things (Iomt) 102a Physiological Sensor 102b Medication intake module 104 Edge processing device 106 Cloud-based analytics engine 108 Multi-channel communication unit

Claims

[1] A health monitoring system for autonomous support of elderly people, the system comprising: a variety of Internet of Medical Things (IoMT) sensors, including at least one wearable physiological sensor, one or more environmental sensor units, a mobility tracking device and a medication intake module, wherein the IoMT sensors are configured to continuously collect physiological, environmental, behavioral and medication use data; an edge processing device that includes an embedded processor and a lightweight machine learning inference unit configured to perform real-time anomaly detection and generate local alerts in the absence of cloud connectivity; a cloud-based analytics engine comprising one or more deep learning models configured to perform longitudinal health modeling, population-level clustering, and predictive risk forecasting using historical and real-time data; and a multi-channel communication unit configured to securely transmit notifications, reports, and emergency triggers to nursing staff, clinical staff, family members, and emergency services, wherein the edge processing device includes a dedicated hardware accelerator selected from a field-programmable gate array (FPGA), the accelerator being optimized to perform low-latency computations, including fall detection, arrhythmia detection, and respiratory abnormality monitoring, within a processing window of less than 200 milliseconds;and wherein the wearable physiological sensor is designed as a wrist-worn or patch-based device comprising a set of sensor elements, including an electrocardiography (ECG) electrode, a photoplethysmography (PPG) sensor for SpO2 measurement, a three-axis accelerometer, a gyroscope, and a skin temperature probe, the device further being configured to integrate raw signal data streams through a multimodal fusion technique performed locally on the edge device to minimize noise, compensate for motion artifacts, and improve the specificity of anomaly detection across multiple concurrent parameters. [2] System according to claim 1, wherein the cloud-based analysis engine comprises a recurrent neural network (RNN) architecture with long-term short-term memory (LSTM) units configured to capture temporal dependencies in health data of older people, wherein the LSTM units are further integrated into a clustering mechanism such as K-means or Gaussian mixture modeling to classify individuals into population-level cohorts for comparative risk profiling, and wherein adaptive thresholds for anomaly detection are dynamically recalibrated based on cohort-level baseline values ​​and the individual user's historical health curve. [3] System according to claim 1, wherein the medication adherence module comprises an intelligent dispensing structure configured with load cells, RFID (Radio Frequency Identification) readers and time-synchronized dispensing compartments, wherein the module is further configured to verify adherence by correlating tablet removal with concurrent physiological changes such as heart rate variability or blood pressure stabilization, thereby enabling closed-loop verification of medication adherence and avoiding reliance on self-reporting. [4] System according to claim 1, wherein the multi-channel communication unit is configured to execute a hierarchical alerting protocol, wherein initial anomaly detection events trigger low-priority alerts to family members and caregivers via mobile notifications, while events classified as serious by the AI ​​engine, such as cardiac arrest, hypoxemia, or confirmed falls resulting in immobility, automatically trigger a high-priority emergency escalation, which simultaneously sends an SOS signal to the emergency medical services, activates voice-based autocall functions, and generates a secure clinical report that is transmitted to a pre-registered physician dashboard. [5] System according to claim 1, wherein the system comprises a data protection and security module that includes end-to-end encryption for all transmitted data, a blockchain-based immutable audit log that records all data accesses and system decisions for compliance with legal requirements, and an explainable AI layer configured to generate human-interpretable justifications for each anomaly classification or predictive risk output, thereby ensuring that nurses and clinicians can understand and validate the decision-making process of the AI ​​analysis engine.

Images