Self-powered multi-parameter wearable health monitoring system and working method thereof

Abstract

The invention belongs to the technical field of wearable equipment, and discloses a self-powered multi-parameter wearable health monitoring system and a working method thereof, and the system comprises a dual-mode self-powered module which is used for generating and outputting electric energy to a DC-DC voltage stabilization chip; the multi-modal sensor array is used for collecting predetermined various vital sign index information and environmental data; the edge AI processing unit is used for performing AI reasoning according to the various vital sign index information and the environmental data, and generating and outputting a health risk assessment result; the dynamic power consumption management module is used for collecting the input power of the DC-DC voltage stabilization chip and the current power consumption of the system, and generating and sending a power consumption scheduling instruction to the multi-mode sensor array and the edge AI processing unit according to the input power of the DC-DC voltage stabilization chip and the current power consumption of the system; according to the invention, the power consumption management and endurance strategy precision is improved, the phenomenon of insufficient energy supply or waste is avoided, and all-weather, continuous and high-precision health monitoring can be realized.

Description

Technical Field

[0001] This invention belongs to the field of wearable device technology, and specifically relates to a self-powered multi-parameter wearable health monitoring system and its working method. Background Technology

[0002] With the rapid development of mobile internet, wearable sensors, and artificial intelligence (AI) technologies, wearable health monitoring devices have been widely used in the field of personal health management. They have gradually evolved from the initial single heart rate monitors and pedometers into comprehensive platforms that can simultaneously record multiple parameters such as blood oxygen, blood pressure, body temperature, and Galvanic Skin Response (GSR). Currently, mainstream products on the market, such as Apple Watch, Fitbit, and Huawei Health Band, play an important role in scenarios such as chronic disease management for the elderly, athlete status monitoring, and telemedicine due to their convenience and multifunctionality. They help users understand their own health status in real time and provide data support for medical personnel.

[0003] Although wearable health monitoring devices are widely used, their lack of dynamic monitoring of energy collection in actual use leads to low accuracy in power management and battery life strategies, making them prone to insufficient or wasted energy. Specifically, in terms of energy supply, existing devices rely on a single energy source, mainly pre-charged lithium batteries. Battery capacity is limited by size, making it difficult to balance higher sampling rates and multimodal sensing with extended battery life. Some manufacturers extend usage time by using low-power chips and optimizing software algorithms, while some research attempts to integrate flexible photovoltaic or thermoelectric materials into watch faces as auxiliary energy sources. However, flexible photovoltaics have low power density under indoor lighting conditions, and thermoelectric power generation is even lower due to temperature differences, meaning that the energy supply of existing devices still cannot support the requirements of high-frequency data collection around the clock. Regarding power management and battery life strategies, existing products mainly rely on static thresholds, lacking dynamic awareness of energy collection. The power output of photovoltaics and thermoelectric power generation varies drastically with the environment, easily leading to insufficient or wasted energy. This limits the achievement of the goal of all-weather, continuous, and high-precision health monitoring for wearable health monitoring devices, hindering their deeper application in more scenarios. Summary of the Invention

[0004] In view of the technical problems existing in the prior art, the present invention provides a self-powered multi-parameter wearable health monitoring system and its working method, so as to solve the technical problem that the existing health monitoring wearable devices lack dynamic energy collection, resulting in low accuracy of power consumption management and battery life strategy, and are prone to insufficient energy supply or waste.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows: This invention provides a self-powered multi-parameter wearable health monitoring system, including a dual-mode self-powered module, a DC-DC voltage regulator chip, a multi-modal sensor array, an edge AI processing unit, and a dynamic power consumption management module; The dual-mode self-powered module is used to generate and output electrical energy to the DC-DC voltage regulator chip; wherein, the dual-mode self-powered module adopts a stacked structure including a flexible organic photovoltaic thin film and a flexible thermoelectric material layer; The DC-DC voltage regulator chip is used to provide power to the multimodal sensor array, the edge AI processing unit, and the dynamic power management module; The multimodal sensor array is used to collect a variety of predetermined vital signs and environmental data. The edge AI processing unit is used to perform AI reasoning based on the collected information of various vital signs and environmental data, and to generate and output health risk assessment results. The dynamic power management module is used to collect the input power of the DC-DC regulator chip and the current power consumption of the system, and generate and send power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the collected input power of the DC-DC regulator chip and the current power consumption of the system; wherein, the power scheduling instructions are used to control the power switch of the multimodal sensor array to turn on and off, and are also used to adjust the data sampling rate and AI inference trigger interval of the edge AI processing unit.

[0006] Furthermore, the flexible organic photovoltaic film and the flexible thermoelectric material layer are bonded together with flexible conductive adhesive and connected in parallel to the input terminal of the DC-DC voltage regulator chip; wherein, the flexible organic photovoltaic film is made of P3HT:PCBM system or non-perfluoropolymer material, and the flexible thermoelectric material layer is made of PEDOT:PSS and Bi2Te3 nanocomposite material.

[0007] Furthermore, the DC-DC voltage regulator chip incorporates a maximum power point tracking algorithm; wherein, the maximum power point tracking algorithm is used to monitor the power output curves of the flexible organic photovoltaic film and the flexible thermoelectric material layer in real time, and to regulate the working state of the flexible organic photovoltaic film and the flexible thermoelectric material layer based on the power output curves of the flexible organic photovoltaic film and the flexible thermoelectric material layer.

[0008] Furthermore, the multimodal sensor array includes a flexible substrate and a plurality of vital sign monitoring sensors integrated on the flexible substrate; wherein the plurality of vital sign monitoring sensors are disposed on the flexible substrate by flexible printing technology.

[0009] Furthermore, several vital sign monitoring sensors include optical heart rate / blood oxygen sensors, skin resistance sensors, body temperature sensors, ambient temperature and humidity sensors, behavior monitoring sensors, and non-invasive blood glucose sensing modules; among them, the behavior monitoring sensors include a three-axis accelerometer and a three-axis gyroscope.

[0010] Furthermore, the edge AI processing unit adopts an ARM Cortex-M55-based MCU, which supports preset health risk assessment models; The preset health risk assessment model includes a multimodal feature fusion layer and a temporal anomaly detection layer. The multimodal feature fusion layer is used to perform cross-modal attention fusion on the collected various vital sign indicators and environmental data, and output an embedded sequence. The temporal anomaly detection layer is used to model the embedded sequence using a two-layer LSTM and output the health risk assessment result.

[0011] Furthermore, the health risk assessment results include health risk scores and categories of health risk abnormalities.

[0012] Furthermore, the process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the collected input power of the DC-DC regulator chip and the current power consumption of the system is as follows: The energy margin ratio is calculated based on the input power of the DC-DC regulator chip and the current power consumption of the system. The energy margin ratio is compared with a preset margin ratio threshold, and a power scheduling instruction is generated and sent to the multimodal sensor array and the edge AI processing unit.

[0013] Furthermore, the dynamic power management module can also generate and send power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the health risk assessment results; The process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the health risk assessment results is as follows: Based on the health risk assessment results, read the preset key vital signs information; The preset key vital signs information is compared with preset indicator thresholds, and power consumption scheduling instructions are generated and sent to the multimodal sensor array and the edge AI processing unit.

[0014] The present invention also provides a method for operating a self-powered multi-parameter wearable health monitoring system, comprising: The dual-mode self-powered module is used to supply power to the DC-DC voltage regulator chip; The dynamic power management module collects the input power of the DC-DC regulator chip and the current power consumption of the system, and generates and sends power scheduling instructions based on the collected input power of the DC-DC regulator chip and the current power consumption of the system. Using a multimodal sensor array, and according to power consumption scheduling instructions, a variety of predetermined vital signs and environmental data are collected; Using an edge AI processing unit, AI reasoning is performed based on collected information from various vital signs and environmental data to generate and output health risk assessment results.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: The self-powered multi-parameter wearable health monitoring system provided by this invention adopts a dual-mode self-powered module with a stacked structure including flexible organic photovoltaic thin film and flexible thermoelectric material layer. This improves the energy source, breaks through the bottleneck of relying solely on lithium battery power supply and its limited capacity, and effectively enhances the energy supply capacity, providing an energy foundation for all-weather high-frequency data acquisition. The system utilizes a dynamic power management module to collect the input power of the DC-DC regulator chip and the current power consumption of the system, generating power scheduling commands to control the on / off state of the power switch of the multi-modal sensor array and regulate the data sampling rate and AI inference trigger interval of the edge AI processing unit. This overcomes the shortcomings of existing products that rely on static thresholds and lack dynamic perception of energy acquisition, improving the accuracy of power management and battery life strategies, avoiding insufficient or wasted energy supply, and facilitating all-weather, continuous, and high-precision health monitoring, thus promoting its deep application in more scenarios.

[0016] The working method of the self-powered multi-parameter wearable health monitoring system provided by the present invention possesses all the advantages of the aforementioned self-powered multi-parameter wearable health monitoring system. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 A schematic diagram of the structure of the self-powered multi-parameter health monitoring system provided in the embodiment; Figure 2 The flowchart illustrates the working method of the self-powered multi-parameter health monitoring system provided in this embodiment. Detailed Implementation

[0019] To make the technical problems solved by the present invention, the technical solutions, and the beneficial effects clearer, the following specific embodiments provide a further detailed description of the present invention. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit the scope of the invention.

[0020] The self-powered multi-parameter wearable health monitoring system provided by the present invention includes a dual-mode self-powered module, a DC-DC voltage regulator chip, a multi-modal sensor array, an edge AI processing unit, and a dynamic power consumption management module.

[0021] The dual-mode self-powered module generates and outputs electrical energy to the DC-DC voltage regulator chip. The dual-mode self-powered module employs a stacked structure comprising a flexible organic photovoltaic thin film and a flexible thermoelectric material layer. The DC-DC voltage regulator chip provides electrical energy to the multimodal sensor array, the edge AI processing unit, and the dynamic power management module. The multimodal sensor array collects predetermined information on various vital signs and environmental data. The edge AI processing unit performs AI inference based on the collected information on various vital signs and environmental data, generating and outputting health risk assessment results. The dynamic power management module collects the input power of the DC-DC voltage regulator chip and the current system power consumption, and generates and sends power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on these data. The power scheduling instructions control the power switch of the multimodal sensor array and regulate the data sampling rate and AI inference trigger interval of the edge AI processing unit.

[0022] In the above embodiments, the dual-mode self-powered module adopts a stacked structure comprising a flexible organic photovoltaic thin film and a flexible thermoelectric material layer, enabling simultaneous utilization of both photovoltaic and thermoelectric energy harvesting methods. This enriches the energy sources and effectively alleviates the endurance problem caused by a single energy source, providing a more reliable energy guarantee for long-term equipment operation. The dynamic power management module collects the input power of the DC-DC voltage regulator chip and the current system power consumption, and generates and sends power scheduling commands in real time. This enables dynamic perception of energy harvesting and dynamic management of power consumption, greatly improving the accuracy of power management and endurance strategies, and effectively avoiding insufficient or wasted energy supply. Furthermore, the power scheduling commands generated by the dynamic power management module can not only control the power switch of the multimodal sensor array but also adjust the data sampling rate and AI inference trigger interval of the edge AI processing unit. Based on the actual energy supply and the current system needs, it can rationally allocate energy, increasing the data sampling rate and AI inference when energy is sufficient. The invention improves the frequency of sampling and inference to obtain more detailed and accurate health data; it reduces sampling rate and inference frequency during energy shortages to extend device battery life, thus better balancing the accuracy of health monitoring and the device's endurance, which is conducive to achieving the goal of all-weather, continuous, and high-precision health monitoring. This invention effectively solves the problems of energy supply and power consumption management in existing health monitoring wearable devices. Through self-powering and dynamic power consumption management, the device can operate more stably and continuously, reducing problems such as frequent charging or data acquisition interruptions due to insufficient energy. This provides users with a more convenient and reliable user experience, helping to improve user acceptance and satisfaction with health monitoring wearable devices, enabling the device to better achieve all-weather, continuous, and high-precision health monitoring. It not only enhances the device's effectiveness in existing application scenarios such as chronic disease management for the elderly, athlete status monitoring, and telemedicine, but also makes it possible for deeper applications in more scenarios with high requirements for health monitoring accuracy and continuity, expanding the device's application scope and market prospects.

[0023] The following specific embodiments further explain the self-powered multi-parameter wearable health monitoring system provided by the present invention: Example 1 As attached Figure 1As shown in the figure, this embodiment provides a self-powered multi-parameter wearable health monitoring system, including a dual-mode self-powered module, a DC-DC voltage regulator chip, a multi-modal sensor array, an edge AI processing unit, and a dynamic power management module. The output terminal of the dual-mode self-powered module is connected to the input terminal of the DC-DC voltage regulator chip. The output terminal of the DC-DC voltage regulator chip is connected to the power supply terminals of the multi-modal sensor array, the edge AI processing unit, and the dynamic power management module. The dynamic power management module is electrically connected to the dual-mode self-powered module, the DC-DC voltage regulator chip, the multi-modal sensor array, the edge AI processing unit, and the dynamic power management module. The multi-modal sensor array is also electrically connected to the edge AI processing unit.

[0024] The dual-mode self-powered module is used to generate and output electrical energy to the DC-DC voltage regulator chip based on the photovoltaic and thermoelectric effects. The DC-DC voltage regulator chip is used to perform voltage conversion processing on the electrical energy generated by the dual-mode self-powered module and provide electrical energy to the multimodal sensor array, the edge AI processing unit, and the dynamic power consumption management module. The multimodal sensor array is used to collect pre-determined information on various vital signs and environmental data. The edge AI processing unit is used to perform AI inference based on the collected information on various vital signs, generate and output health risk assessment results. The dynamic power consumption management module is used to... The system collects the input power of the DC-DC regulator chip and the current power consumption of the system, and generates and sends power scheduling instructions to the multi-mode sensor array and the edge AI processing unit based on the collected input power of the DC-DC regulator chip and the current power consumption of the system. Optionally, the dynamic power management module can also generate and send power scheduling instructions to the multi-mode sensor array and the edge AI processing unit based on the health risk assessment results. The power scheduling instructions are used to control the power switch of the multi-mode sensor array and to adjust the data sampling rate and AI inference trigger interval of the edge AI processing unit.

[0025] In this embodiment, the dual-mode self-powered module adopts a stacked structure comprising a flexible organic photovoltaic film and a flexible thermoelectric material layer. The flexible organic photovoltaic film and the flexible thermoelectric material layer are bonded together with flexible conductive adhesive and connected in parallel to the input terminal of the DC-DC voltage regulator chip. The stacked structure comprising the flexible organic photovoltaic film and the flexible thermoelectric material layer is located on the upper surface of the multi-parameter wearable health monitoring system. Preferably, the flexible organic photovoltaic film is made of P3HT:PCBM system or non-perfluoropolymer material with a thickness of less than 30µm. The flexible thermoelectric material layer is made of PEDOT:PSS and Bi2Te3 nanocomposite material with a thickness of 50µm and is disposed in contact with the user's skin.

[0026] It should be noted that under normal indoor illuminance (200–500 lux), the peak power density of the flexible organic photovoltaic film is 6-12 mW; under outdoor sunlight (≥30 klux), the peak power density of the flexible organic photovoltaic film can reach 30 mW; when the temperature difference between human skin surface and environment is 2-5℃, the peak power density of the flexible thermoelectric material layer is 5-10 mW; when the flexible organic photovoltaic film and the flexible thermoelectric material layer are connected in parallel to the DC-DC voltage regulator chip, the dual-mode self-powered module can continuously supply 20-30 mW per day, which is sufficient to support the low-power operation mode of the system and achieve "zero charging" or very little charging requirement.

[0027] In this embodiment, the DC-DC voltage regulator chip has a built-in maximum power point tracking (MPPT) algorithm to monitor the power output curves of the photovoltaic and thermoelectric effects of the dual-mode self-powered module in real time, so as to automatically switch the working mode of the dual-mode self-powered module.

[0028] Specifically, the maximum power point tracking algorithm is used to monitor the power output curve of the flexible organic photovoltaic film and the power output curve of the flexible thermoelectric material layer in real time, and to regulate the working state of the flexible organic photovoltaic film and the flexible thermoelectric material layer according to the power output curve of the flexible organic photovoltaic film and the flexible thermoelectric material layer, so that the flexible organic photovoltaic film or the flexible thermoelectric material layer can work independently, or the flexible organic photovoltaic film and the flexible thermoelectric material layer can work in parallel and synchronously.

[0029] The regulated output of the DC-DC voltage regulator chip is provided to the multimodal sensor array, the edge AI processing unit, and the dynamic power management module through a low dropout regulator (LDO) to provide power to the multimodal sensor array, the edge AI processing unit, and the dynamic power management module, ensuring that the output voltage fluctuation of the dual-mode self-powered module does not affect the stability of the system.

[0030] In this embodiment, the multimodal sensor array includes a flexible substrate and several vital sign monitoring sensors. The flexible substrate is a PCB substrate with a thickness of 0.1 mm, and the several vital sign monitoring sensors are integrated on the flexible substrate using flexible printing technology. The voltage of all vital sign monitoring sensors is supplied by a 3.3V regulated voltage output from the DC-DC voltage regulator chip. The wiring between the several vital sign monitoring sensors adopts differential routing to reduce cross-interference between the vital sign monitoring sensors.

[0031] Several vital sign monitoring sensors include an optical heart rate / blood oxygen sensor, a skin resistance sensor, a body temperature sensor, an ambient temperature and humidity sensor, a behavior monitoring sensor, and a non-invasive blood glucose sensing module. The optical heart rate / blood oxygen sensor includes a dual-wavelength LED and a photodiode, with a 2mm gap between the LED and the photodiode, and the LED has wavelengths of 660nm and 940nm. Preferably, the optical heart rate / blood oxygen sensor uses reflective measurement technology. The skin resistance sensor uses Ag / AgCl flexible electrodes with a 5nm gap for mood and stress monitoring. The body temperature sensor uses a flexible thermistor, fitted to the skin, with a monitoring error of ±0.1℃. The ambient temperature and humidity sensor uses a MEMS-type temperature and humidity chip encapsulated in a pre-designed waterproof layer. The behavior monitoring sensor is used for user posture and fall detection, and includes a three-axis accelerometer and a three-axis gyroscope. The non-invasive blood glucose sensing module is used for blood glucose monitoring using a non-invasive estimation method based on skin microcurrent and spectral characteristics.

[0032] In this embodiment, the edge AI processing unit is based on an ARM Cortex-M55 MCU (Microcontroller Unit), which integrates a 2-TOPS NPU (supporting 8-bit quantization models) and has 256KB SRAM and 1MB Flash. The ARM Cortex-M55 MCU supports a preset health risk assessment model. The preset health risk assessment model uses publicly available multimodal health datasets (such as PhysioNet and MIMIC IV) for transfer learning and is continuously fine-tuned on the user device through federated learning to adapt to individual differences.

[0033] Specifically, the pre-defined health risk assessment model includes a multimodal feature fusion layer and a temporal anomaly detection layer. The multimodal feature fusion layer performs cross-modal attention fusion on various collected vital sign indicators and environmental data, outputting an embedded sequence. The temporal anomaly detection layer uses a two-layer LSTM to model the embedded sequence and outputs the health risk assessment result. Preferably, the multimodal feature fusion layer uses a lightweight Transformer Encoder with 4 layers and a head=4, which performs cross-modal attention fusion on various collected vital sign indicators and environmental data such as heart rate, blood oxygen, skin resistance, body temperature, IMU, and blood glucose, outputting a unified 64-dimensional embedded sequence. The two-layer LSTM has 32 hidden units. The health risk assessment result includes a health risk score and a health risk anomaly category. The health risk score is a score of 0-1, and the health risk anomaly category includes, for example, arrhythmia, hypoglycemia, and fall warning.

[0034] It should be noted that the edge AI processing unit has a default sampling rate of 1Hz for various predetermined vital signs and environmental data, which is increased to 10Hz when a key event is triggered. The collected vital signs and environmental data are amplified by a low-noise preamplifier and sent to the ADC channel of the MCU. The MCU performs denoising, calibration, and segmentation preprocessing operations to obtain preprocessed feature vectors. The preprocessed feature vectors are directly written to shared memory via DMA and provided to the NPU for use. In the denoising process, denoising is performed based on moving average and wavelet threshold. In the calibration process, linear correction is performed using the sensor's factory calibration coefficients. In the segmentation process, continuous data is divided into 5-second time windows to generate preprocessed feature vectors containing time domain, frequency domain, and statistical features.

[0035] It should also be noted that when performing AI inference based on the collected various vital signs and environmental data, the NPU first reads the preprocessed feature vector stored in the MCU's shared memory via DMA; then, it uses a preset health risk assessment model to perform inference analysis and obtain health risk assessment results including health risk scores and abnormal health risk categories; noise reduction is performed by pre-quantizing the preset health risk assessment model into an 8-bit model to reduce computing power requirements; finally, the health risk assessment results are written back to the MCU to trigger corresponding power scheduling or communication commands.

[0036] In this embodiment, the process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit in the dynamic power management module based on the collected input power of the DC-DC regulator chip and the current power consumption of the system includes the following steps: (1.1) Based on the input power of the collected DC-DC regulator chip and the current power consumption of the system, the energy margin ratio is calculated; the calculation process of the energy margin ratio is as follows:

[0037] in, This refers to the energy surplus ratio; This refers to the input power of the DC-DC regulator chip. This represents the current power consumption of the system.

[0038] (1.2) Compare the energy margin ratio with a preset margin ratio threshold, generate and send power scheduling instructions to the multimodal sensor array and the edge AI processing unit; wherein, the preset margin ratio threshold includes a threshold and threshold The power consumption scheduling instructions include high-energy mode instructions, balanced mode instructions, and low-energy mode instructions; specifically, if the following conditions are met... Then a high-energy mode instruction is generated; if the conditions are met... Then generate a balanced mode instruction; if the conditions are met... Then a low-energy mode instruction is generated; preferably, the threshold is... The value is 1.2, and the threshold. The value is 0.8.

[0039] It is worth noting that the high-energy mode command specifically includes triggering all sensors in the multimodal sensor array to be fully turned on, adjusting the sampling rate of predetermined vital signs and environmental data to 1Hz, and setting the AI ​​inference frequency of the edge AI processing unit to once every 1 second; the balanced mode command specifically includes turning off non-critical sensors in the multimodal sensor array, adjusting the sampling rate of predetermined vital signs and environmental data to 0.5Hz, and setting the AI ​​inference frequency of the edge AI processing unit to once every 5 seconds; the low-energy mode command specifically includes keeping only the critical sensors in the multimodal sensor array on, adjusting the sampling rate of predetermined vital signs and environmental data to 0.2Hz, and setting the AI ​​inference frequency of the edge AI processing unit to once every 30 seconds; among them, non-critical sensors in the multimodal sensor array include environmental temperature and humidity sensors, and critical sensors in the multimodal sensor array include optical heart rate / blood oxygen sensors.

[0040] In the dynamic power management module, the process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the health risk assessment results includes: reading preset key vital sign information based on the health risk assessment results; wherein, the preset key vital sign information includes, for example, heart rate and blood oxygen; comparing the preset key vital sign information with preset indicator thresholds, generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit; for example, when abnormal preset key vital sign information is detected, such as heart rate > 120 bpm and blood oxygen < 90%, a high-energy mode instruction is generated, which will briefly increase the sampling rate to obtain more refined data even in a low-energy state.

[0041] Specifically, the power scheduling instruction directly controls the sensor power switch in the multimodal sensor array, the sampling frequency of the ADC in the edge AI processing unit, and the NPU inference trigger interval in the edge AI processing unit through the MCU register mapping to achieve millisecond-level response; wherein, the power scheduling instruction is evaluated and generated every 10 seconds to ensure that the system remains stable when energy fluctuates.

[0042] In this embodiment, the self-powered multi-parameter wearable health monitoring system further includes a communication module, a user terminal, and a cloud-based collaborative subsystem. The communication module facilitates communication between the edge AI processing unit, the cloud-based collaborative subsystem, and the user terminal. Employing BLE 5.0 long-distance mode (PHY 2M) and a preset channel selection algorithm, the communication module encrypts and pushes a health summary to the user terminal every 5 minutes. The health summary includes the health risk score and abnormal health risk category for the most recent hour. The user terminal is a mobile phone or tablet. Preferably, for critical warnings with a health risk score greater than 0.8, the communication module immediately notifies and triggers vibration or voice alerts on the user's mobile phone or tablet. The cloud-based collaborative subsystem receives the health summary and performs long-term trend analysis based on it, such as blood glucose fluctuation curves and sleep quality assessments. It also sends model fine-tuning parameters back to the device to achieve a federated learning closed loop. Furthermore, users can view historical reports, risk trend charts, and doctor recommendations on the user terminal. Notably, all local data is encrypted using AES 256 GCM and the transport layer uses TLS. 1.3; The device generates a unique Device ID and key pair during the initial pairing. Subsequent OTA (Over-the-Air) updates are verified through signature to prevent malicious tampering.

[0043] In this embodiment, the system as a whole adopts a fully flexible PCB (polyimide substrate) and a medical-grade silicone shell for integrated packaging; the thickness is less than or equal to 1.2mm, the weight is about 12g, and it conforms to the ergonomic curve, so it can fit the wrist, forearm or chest; the shell adopts a double-layer silicone structure, with the inner layer being waterproof and sweatproof, and the outer layer providing anti-slip texture; the waterproof rating is IP68, which can be immersed in 1m deep water for 30 minutes; the bending life is greater than or equal to 10,000 times / 180° bending cycle; the dual-mode self-powered module adopts a stress dispersion structure to ensure that the power attenuation is less than 5%; the system surface is fluorinated to prevent corrosion from sweat and salt spray.

[0044] Working methods and principles: As attached Figure 2 As shown in the figure, the working method of a self-powered multi-parameter wearable health monitoring system provided in this embodiment includes the following steps: Step 1: Use the dual-mode self-powered module to supply power to the DC-DC regulator chip.

[0045] Step 2: Using the dynamic power management module, collect the input power of the DC-DC regulator chip and the current power consumption of the system, and generate and send power scheduling instructions based on the collected input power of the DC-DC regulator chip and the current power consumption of the system.

[0046] Step 3: Using a multimodal sensor array, collect predetermined vital signs and environmental data according to power consumption scheduling instructions.

[0047] Step 4: Using the edge AI processing unit, perform AI reasoning based on the collected various vital sign indicators and environmental data to generate and output health risk assessment results.

[0048] It should be noted that in the self-powered multi-parameter wearable health monitoring system described in Embodiment 1, after the device is powered on, the dual-mode self-powered module immediately supplies power to the DC-DC voltage regulator chip; at this time, the dynamic power management module begins to evaluate the current power supply status of the system; then, under the control of power scheduling instructions, the multi-modal sensor array collects predetermined information on various vital signs and environmental data; subsequently, the edge AI processing unit performs denoising, calibration, and windowing on the collected information on various vital signs and environmental data to obtain preprocessed feature vectors, and writes the preprocessed feature vectors into shared memory; then, the NPU reads the preprocessed feature vectors from the shared memory... The system performs multimodal fusion operations and predictive analysis based on a time-series anomaly detection model, outputting health risk assessment results including a health risk score and a category of health risk anomalies. If the risk score exceeds a threshold or energy reserves decrease, the dynamic power management module instantly generates and sends power scheduling instructions to adjust the operating mode, ensuring continuous critical monitoring. When a critical warning occurs, it is pushed to the user terminal in real time via the communication module, and health summaries are periodically uploaded to the cloud for long-term analysis and model fine-tuning at preset intervals. Furthermore, users can view real-time sensor data, historical reports, and health recommendations on the user terminal, and can manually switch monitoring modes or activate exercise modes.

[0049] The self-powered multi-parameter wearable health monitoring system described in this embodiment employs a dual-mode self-powered module with a stacked structure of flexible organic photovoltaic thin film and flexible thermoelectric material layer. This overcomes the limitations of existing devices with a single energy source and low auxiliary energy power density, enabling more efficient power generation and providing a more abundant and stable energy source for the system, supporting all-weather high-frequency data acquisition. Simultaneously, the dynamic power management module can collect input power and the current power consumption of the system, and generate power scheduling instructions accordingly to precisely control the multi-modal sensor array and edge AI processing unit. This solves the problem of low accuracy in power management and battery life strategies, and the tendency for insufficient or wasted energy supply, caused by the lack of dynamic perception of energy acquisition in existing products. This helps to achieve all-weather, continuous, and high-precision health monitoring, promoting the in-depth application of this device in more scenarios.

[0050] In this embodiment, the dual-mode self-powered module adopts a parallel connection of flexible organic photovoltaic film and flexible thermoelectric material layer. Utilizing the photovoltaic effect of the flexible organic photovoltaic film and the thermoelectric effect of the flexible thermoelectric material layer, it achieves all-weather, near-zero charging energy supply, breaking through the traditional battery life bottleneck. The edge AI processing unit can complete real-time fusion and anomaly prediction of multi-source data such as heart rate, blood oxygen, GSR, body temperature, and posture on resource-constrained wearable devices, significantly reducing latency and improving privacy and security. The dynamic power management module, based on real-time energy sensing and adaptive power management, ensures that the system always maintains optimal working state under different lighting and temperature differences. The entire system adopts a fully flexible PCB and medical-grade silicone encapsulation to achieve a thin, comfortable, and durable wearing experience, while also considering large-area energy collection and sensing layout. It introduces BLE+TLS dual-layer encryption, AES local storage, and OTA signature verification to ensure the confidentiality and integrity of user health data. Through cloud subsystems and collaborative fine-tuning of models between the cloud and the device, it continuously improves the accuracy of individualized predictions, making it suitable for various scenarios such as chronic disease management for the elderly and athlete status monitoring.

[0051] The self-powered multi-parameter wearable health monitoring system of this invention adopts a dual-mode self-powered module with a stacked structure including a flexible organic photovoltaic thin film and a flexible thermoelectric material layer, achieving all-weather, all-day energy self-sufficiency. It employs an edge AI processing unit, enabling real-time health risk assessment on resource-constrained wearable devices, reducing reliance on the cloud. A dynamic power management module collects the input power of the DC-DC regulator chip and the current system power consumption, generating power scheduling commands to control the power switch of the multi-modal sensor array and regulate the data sampling rate and AI inference trigger interval of the edge AI processing unit. Based on real-time energy acquisition, the sensor sampling rate is dynamically adjusted to maximize battery life, improving the accuracy of power management and battery life strategies, avoiding insufficient or wasted energy supply, and facilitating all-weather, continuous, and high-precision health monitoring, thus promoting its deeper application in more scenarios.

[0052] The above embodiments are merely one of the implementation methods for achieving the technical solution of the present invention. The scope of protection claimed by the present invention is not limited to this embodiment, but also includes any variations, substitutions and other implementation methods that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention.

Claims

1. A self-powered, multi-parameter wearable health monitoring system, characterized in that, It includes a dual-mode self-powered module, a DC-DC voltage regulator chip, a multi-modal sensor array, an edge AI processing unit, and a dynamic power management module; The dual-mode self-powered module is used to generate and output electrical energy to the DC-DC voltage regulator chip based on the photovoltaic effect and thermoelectric effect; wherein, the dual-mode self-powered module adopts a stacked structure including a flexible organic photovoltaic thin film and a flexible thermoelectric material layer; The DC-DC voltage regulator chip is used to provide power to the multimodal sensor array, the edge AI processing unit, and the dynamic power management module; The multimodal sensor array is used to collect a variety of predetermined vital signs and environmental data. The edge AI processing unit is used to perform AI reasoning based on the collected information of various vital signs and environmental data, and to generate and output health risk assessment results. The dynamic power management module is used to collect the input power of the DC-DC regulator chip and the current power consumption of the system, and generate and send power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the collected input power of the DC-DC regulator chip and the current power consumption of the system; wherein, the power scheduling instructions are used to control the power switch of the multimodal sensor array to turn on and off, and are also used to adjust the data sampling rate and AI inference trigger interval of the edge AI processing unit.

2. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The flexible organic photovoltaic film and the flexible thermoelectric material layer are bonded together with flexible conductive adhesive and connected in parallel to the input terminal of the DC-DC voltage regulator chip; wherein, the flexible organic photovoltaic film is made of P3HT:PCBM system or non-perfluoropolymer material, and the flexible thermoelectric material layer is made of PEDOT:PSS and Bi2Te3 nanocomposite material.

3. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The DC-DC voltage regulator chip has a built-in maximum power point tracking algorithm; wherein, the maximum power point tracking algorithm is used to monitor the power output curve of the flexible organic photovoltaic film and the power output curve of the flexible thermoelectric material layer in real time, and to regulate the working state of the flexible organic photovoltaic film and the flexible thermoelectric material layer according to the power output curve of the flexible organic photovoltaic film and the power output curve of the flexible thermoelectric material layer.

4. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The multimodal sensor array includes a flexible substrate and several vital sign monitoring sensors integrated on the flexible substrate; wherein the several vital sign monitoring sensors are disposed on the flexible substrate using flexible printing technology.

5. The self-powered multi-parameter wearable health monitoring system according to claim 4, characterized in that, Several vital sign monitoring sensors include an optical heart rate / blood oxygen sensor, a skin resistance sensor, a body temperature sensor, an ambient temperature and humidity sensor, a behavior monitoring sensor, and a non-invasive blood glucose sensing module; among them, the behavior monitoring sensor includes a three-axis accelerometer and a three-axis gyroscope.

6. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The edge AI processing unit uses an ARM Cortex-M55-based MCU, which supports preset health risk assessment models. The preset health risk assessment model includes a multimodal feature fusion layer and a temporal anomaly detection layer. The multimodal feature fusion layer is used to perform cross-modal attention fusion on the collected various vital sign indicators and environmental data, and output an embedded sequence. The temporal anomaly detection layer is used to model the embedded sequence using a two-layer LSTM and output the health risk assessment result.

7. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The health risk assessment results include a health risk score and a category of health risk abnormalities.

8. The self-powered multi-parameter wearable health monitoring system according to claim 1, characterized in that, The process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the input power of the collected DC-DC regulator chip and the current power consumption of the system is as follows: The energy margin ratio is calculated based on the input power of the DC-DC regulator chip and the current power consumption of the system. The energy margin ratio is compared with a preset margin ratio threshold, and a power scheduling instruction is generated and sent to the multimodal sensor array and the edge AI processing unit.

9. A self-powered, multi-parameter wearable health monitoring system according to claim 8, characterized in that, The dynamic power management module can also generate and send power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the health risk assessment results. The process of generating and sending power scheduling instructions to the multimodal sensor array and the edge AI processing unit based on the health risk assessment results is as follows: Based on the health risk assessment results, read the preset key vital signs information; The preset key vital signs information is compared with preset indicator thresholds, and power consumption scheduling instructions are generated and sent to the multimodal sensor array and the edge AI processing unit.

10. The operating method of a self-powered multi-parameter wearable health monitoring system as described in any one of claims 1-9, characterized in that, include: The dual-mode self-powered module is used to supply power to the DC-DC voltage regulator chip; The dynamic power management module collects the input power of the DC-DC regulator chip and the current power consumption of the system, and generates and sends power scheduling instructions based on the collected input power of the DC-DC regulator chip and the current power consumption of the system. Using a multimodal sensor array, and according to power consumption scheduling instructions, a variety of predetermined vital signs and environmental data are collected; Using an edge AI processing unit, AI reasoning is performed based on collected information from various vital signs and environmental data to generate and output health risk assessment results.

Images