An interactive, multimodal, intelligent monitoring system and method for physiological parameters

Abstract

This application relates to the field of health monitoring technology, specifically to an interactive multimodal physiological parameter intelligent monitoring system and method. The system includes a wearable device and a remote service platform. The wearable device includes: a multimodal signal acquisition module for simultaneously acquiring the user's heart sound signals and PPG signals, and can switch to acquiring voice signals; a physiological parameter processing module for generating at least two physiological parameters based on the heart sound signals and PPG signals; an interaction triggering module for responding to user triggers to establish a communication link and switching the multimodal signal acquisition module to acquire voice signals; a data uploading module for synchronously uploading the voice signals and physiological parameters to the remote service platform; and an audio feedback module for receiving and playing audio feedback from the remote service platform. This system achieves high-precision continuous monitoring of physiological parameters while also supporting one-click triggering of remote medical interaction, improving the convenience and immediacy of health management.

Description

Technical Field

[0001] This application relates to the field of wearable health monitoring technology, specifically to an interactive multimodal physiological parameter intelligent monitoring system and method. Background Technology

[0002] With cardiovascular disease becoming a major global health threat, accurate, continuous, and convenient non-invasive blood pressure monitoring has become a key requirement in clinical medicine and health management. Traditional cuff blood pressure monitors (based on Korotkoff sounds or oscillometric principles), due to their inherent intermittent measurement pattern, cannot capture instantaneous fluctuations in blood pressure, diurnal rhythms, and rapid changes caused by daily activities and emotional stress. This greatly limits their application value in cardiovascular event early warning, evaluation of antihypertensive drug efficacy, and personalized treatment plan development. Furthermore, the repeated inflation and deflation of the cuff not only causes discomfort, but the measurement results are also easily affected by factors such as measurement position, posture, and cuff size, introducing errors.

[0003] To overcome the aforementioned limitations, non-invasive continuous blood pressure monitoring technologies based on pulse transit time (PTT) or pulse arrival time (PAT) have emerged. This technology traditionally estimates blood pressure by simultaneously measuring electrocardiogram (ECG) and photoplethysmography (PPG) signals. However, this classic approach faces several challenges: First, obtaining high-quality ECG signals typically requires fixing electrodes to the chest or limbs, increasing device complexity and significantly impacting comfort, aesthetics, and long-term adherence. Second, ECG signals are highly susceptible to interference during daily activities, leading to inaccurate R-wave detection and directly affecting the reliability of PAT calculations. Third, using the ECG R-wave as a time reference inherently introduces electromechanical delay, resulting in variability and weakening the stability and individual universality of the blood pressure estimation model.

[0004] Furthermore, existing technologies also have shortcomings in other important areas involving continuous physiological parameter assessment, such as the monitoring of baroreflex sensitivity (BRS). The traditional "gold standard" method for BRS assessment is invasive, complex, and not continuous, making it difficult to promote. Meanwhile, non-invasive assessment schemes based on continuous non-invasive blood pressure and heart rate signals are highly dependent on the measurement quality of the baseline signals for accuracy. Currently, there is a lack of convenient devices that can provide high-precision, continuously synchronized blood pressure and heart rate interval signals suitable for long-term daily monitoring.

[0005] On the other hand, monitoring heart sound signals—a direct source of information reflecting cardiac mechanical activity—is considered a promising technological direction, but its application in wearable devices still faces significant challenges. First, the heart sound vibration signals collected by wearable devices are extremely weak and easily masked by environmental noise and motion artifacts. Second, heart sound signals vary greatly among different individuals and under different conditions, making stable detection of key features such as the first and second heart sounds difficult. Third, achieving accurate physiological parameter calculations requires extremely high synchronization precision between heart sounds and PPG signals, placing stringent demands on the hardware system. Fourth, integrating high-performance heart sound sensors and PPG sensors within limited spaces such as rings presents engineering bottlenecks.

[0006] In addition to the limitations of the aforementioned monitoring technologies, existing wearable health monitoring devices also have significant shortcomings in user interaction and service integration. Current devices largely remain at the level of "passive data collection," requiring users to perform subsequent operations through smartphone applications. In emergency situations where a user suddenly feels unwell, this lengthy interaction path may delay timely assistance. Furthermore, the lack of an immediate and seamless linkage mechanism between monitoring data and professional medical services makes it difficult for users to obtain timely and targeted guidance. There are also concerns about privacy breaches when conducting voice consultations in public places. Summary of the Invention

[0007] To overcome the shortcomings of the prior art, this application provides an interactive multimodal physiological parameter intelligent monitoring system and method, which integrates the synchronous acquisition of multimodal signals such as heart sounds, photoplethysmography pulse waves, etc., and can achieve high-precision continuous estimation of physiological parameters such as blood pressure and baroreflex sensitivity, and can directly initiate remote medical consultations through touch interaction.

[0008] The first aspect of this application provides an interactive multimodal physiological parameter intelligent monitoring system, including a wearable device and a remote service platform, wherein the wearable device includes a wearable body worn on the user's finger or wrist and a component located on the wearable body:

[0009] The multimodal signal acquisition module is used to simultaneously acquire the user's heart sound signal and PPG signal, and switch to acquire the user's voice signal. A physiological parameter processing module is used to generate at least two physiological parameters based on at least the heart sound signal and the PPG signal; An interactive trigger module is used to establish a communication link with the remote service platform when triggered by a user, and to switch the multimodal signal acquisition module to acquire the voice signal; The data upload module is used to synchronously upload the voice signal and the physiological parameters to the remote service platform; An audio feedback module is used to receive and play audio feedback from the remote service platform.

[0010] In an optional embodiment, the wearable body includes a smart ring, and the multimodal signal acquisition module includes: At least one heart sound / voice sensor is disposed on the side wall of the smart ring for contact with the palmar skin of the proximal phalanx of the finger and for collecting the user's heart sound signal and / or the user's voice signal; At least one PPG sensor is disposed on the side of the smart ring near the heart sound / voice sensor for collecting the PPG signal; A multi-channel synchronous analog front-end circuit is electrically connected to the heart sound / speech sensor and the PPG sensor, and is configured to synchronously acquire the heart sound signal and the PPG signal; The heart sound / voice sensor includes a piezoelectric vibration sensor, which is a bone conduction sensor; or the heart sound / voice sensor includes a microelectromechanical system (MEMS) microphone.

[0011] In an optional embodiment, the multimodal signal acquisition module further includes a pair of electrocardiogram electrodes for acquiring electrocardiogram signals; The multi-channel synchronous analog front-end circuit is electrically connected to the ECG electrodes, the heart sound / speech sensor, and the PPG sensor, and is used to synchronously acquire the heart sound signal, the PPG signal, and the ECG signal, and the inter-channel time synchronization deviation of the multi-channel synchronous analog front-end circuit is less than 100 microseconds.

[0012] In an optional embodiment, the physiological parameters include blood pressure parameters, and the physiological parameter processing module is configured to generate the blood pressure parameters through the following steps: Extract the first heart sound feature point from the heart sound signal and extract the pulse wave origin from the PPG signal; Using the first heart sound feature point as a reference for the onset time of cardiac contraction, the time interval between the first heart sound feature point and the pulse wave initiation point is calculated. The time interval is input into a pre-trained personalized calibration model, which outputs estimates of systolic and diastolic blood pressure. The personalized calibration model is configured to be built based on initial calibration data and updated according to the real-time acquired speech signals and physiological parameters.

[0013] In an optional embodiment, the physiological parameter processing module also includes a built-in blood pressure classification unit, which is used to classify the blood pressure parameters according to pre-stored classification criteria and to take different early warning strategies based on the classification results.

[0014] In an optional embodiment, the physiological parameters include baroreflex sensitivity parameters, and the physiological parameter processing module is configured to generate baroreflex sensitivity parameters through the following steps: Extract a continuous first heart sound time series from the continuous heart sound signal, and calculate the heartbeat interval sequence based on the first heart sound time series; The systolic blood pressure sequence is obtained from the continuously generated blood pressure parameters; The correlation between the systolic blood pressure sequence and the intercardiac interval sequence was analyzed using either a sequence method or a spectral method, and the baroreflex sensitivity parameter was calculated.

[0015] In an optional embodiment, the physiological parameter processing module includes a multimodal AI fusion estimation model, which is configured to predict based at least on the heart sound signal and the PPG signal, and output at least two physiological parameters among blood pressure parameter, baroreflex sensitivity parameter, heart rate variability and vascular elasticity index. And / or, the multimodal AI fusion estimation model employs a neural network architecture based on an attention mechanism, used to dynamically adjust the contribution weights corresponding to the heart sound signal and the PPG signal based on the features extracted from the heart sound signal and the PPG signal.

[0016] In an optional embodiment, when the wearable body is a smart ring, the interaction triggering module includes a touch switch, which is configured to recognize at least two predefined gestures among single click, double click, long press, or swipe.

[0017] In an optional embodiment, the remote service platform includes a mobile terminal and a cloud service platform. The interaction triggering module further includes a wireless communication module, which is configured to: when establishing a communication link with the remote service platform, preferentially connect directly to the cloud service platform via a cellular mobile communication network; when the direct connection fails, switch to connect to the mobile terminal via a short-range wireless communication protocol, and the mobile terminal establishes a relay connection with the cloud service platform.

[0018] A second aspect of this application provides an interactive multimodal physiological parameter intelligent monitoring method, characterized in that it is applied to the aforementioned interactive multimodal physiological parameter intelligent monitoring system, the method comprising: Simultaneously collect the user's heart sound signal and PPG signal; Based on the heart sound signal and the PPG signal, at least two physiological parameters are generated in real time. In response to a trigger signal from the user, a communication link is established with the remote service platform, and the multimodal signal acquisition module is switched to acquire the user's voice signal; The voice signal and the physiological parameters are simultaneously uploaded to the remote service platform; Receive and play audio feedback from the remote service platform and implement early warning functions.

[0019] This application has at least the following beneficial effects: This application utilizes a multimodal signal acquisition module and a physiological parameter processing module to replace or supplement electrocardiogram (ECG) signals with heart sound signals and fuse them with PPG signals, achieving more physiologically sound and stable high-precision continuous blood pressure monitoring. Furthermore, the physiological parameter processing module enables comprehensive multi-parameter evaluation, monitoring not only blood pressure but also various cardiovascular health indicators such as baroreflex sensitivity, heart rate variability, and vascular elasticity. Further, this application incorporates an interactive trigger module, a data upload module, and an audio feedback module on the wearable device. Users can directly trigger remote medical services through simple touch gestures, establishing two-way voice communication with the remote medical service platform, greatly simplifying the path to seeking help in emergencies or routine consultations, and improving response speed and user experience. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying 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.

[0021] Figure 1 This is a schematic diagram of the overall architecture of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 2 This is a cross-sectional view of the internal structure of the smart ring of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 3 The timing diagram for synchronous acquisition of multimodal signals from the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 4 The flowchart of the blood pressure estimation algorithm of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 5 This is a schematic diagram illustrating the interaction between the wearable device and the remote service platform of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1. Figure 6 This is a diagram illustrating the multimodal AI fusion model architecture of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1. Figure 7 A schematic diagram of the mobile terminal application interface of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 8 This is a schematic diagram of the modules of the interactive multimodal physiological parameter intelligent monitoring system provided in Example 1; Figure 9 This is a flowchart illustrating the interactive multimodal physiological parameter intelligent monitoring method provided in Example 1.

[0022] Figure label: 100 - Wearable devices; 200 - Mobile terminals; 300 - Cloud service platforms; 10 - Multimodal signal acquisition module; 20 - Physiological parameter processing module; 30 - Interactive triggering module; 40 - Data upload module; 50 - Audio feedback module; 60 - Wearable device; 1-Heart sound / voice sensor; 2-PPG sensor; 3-ECG electrode; 4-Temperature sensor; 5-Six-axis inertial measurement unit; 6-Finger switch; 7-Bone conduction oscillator; 8-Battery; 9-Main circuit board; 10-Wireless charging coil. Detailed Implementation

[0023] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0024] like Figure 1-8 As shown, an interactive multimodal physiological parameter intelligent monitoring system includes a wearable device 100 and a remote service platform. The wearable device 100 includes a wearable body 60 worn on the user's finger or wrist and a device located on the wearable body 60: The multimodal signal acquisition module 10 is used to simultaneously acquire the user's heart sound signal and PPG signal, and switch to acquire the user's voice signal; Physiological parameter processing module 20 is used to generate at least two physiological parameters based on at least heart sound signals and PPG signals; The interactive trigger module 30 is used to establish a communication link with the remote service platform when triggered by the user, and to switch the multimodal signal acquisition module 10 to acquire voice signals. Data upload module 40 is used to synchronously upload voice signals and physiological parameters to the remote service platform; The audio feedback module 50 is used to receive and play audio feedback from the remote service platform and to implement an early warning function. The sensor used to collect heart sound signals is also part of the audio feedback module's voice acquisition component.

[0025] This application, by simultaneously acquiring heart sound signals and PPG signals, avoids the reliance on electrocardiogram (ECG) electrodes found in traditional methods, significantly improving wearing comfort and aesthetics. Utilizing heart sounds, a signal directly reflecting cardiac mechanical activity, as a time reference provides a more stable and reliable data foundation for continuous blood pressure estimation based on the pulse wave conduction time principle. Secondly, the system integrates an interactive trigger module 30, a data upload module 40, and an audio feedback module 50, allowing users to initiate communication with a remote service platform with a single click on the wearable device itself, without the need for external devices such as mobile phones. This greatly shortens the path to assistance and improves the timeliness of health intervention. Thirdly, this system integrates multimodal signal acquisition, local processing, wireless communication, and audio interaction functions into a single finger or wrist wearable device 100. Users can seamlessly switch from daily continuous monitoring to proactive remote medical consultations without operating multiple devices or complex processes, making health monitoring and management more convenient and intelligent.

[0026] The remote service platform includes a mobile terminal 200 (such as a mobile phone) and a cloud service platform 300 (such as a cloud medical service platform). The mobile terminal 200 acts as an intermediary for interaction between the edge computing node and the user; the cloud service platform 300 provides advanced computing, data storage and professional medical services.

[0027] like Figure 2 As shown, in a preferred embodiment, the wearable body 60 includes a smart ring, and the multimodal signal acquisition module 10 includes: At least one heart sound / voice sensor 1 is disposed on the side wall of the smart ring for contact with the palmar skin of the proximal phalanx of the finger and for collecting the user's heart sound signal and / or collecting the user's voice signal; At least one PPG sensor 2 is located on the side of the smart ring near the heart sound / voice sensor 1 for collecting PPG signals; A multi-channel synchronous analog front-end circuit is electrically connected to heart sound / voice sensor 1 and PPG sensor 2, and is configured to synchronously acquire heart sound signals and PPG signals.

[0028] This application places the heart sound / voice sensor 1 on the side wall of the ring and in contact with the palmar skin of the proximal phalanx of the finger. This position not only effectively picks up the heart sound vibration signal transmitted through bone conduction, but also naturally reuses it as a high-sensitivity voice acquisition microphone, greatly saving internal space and hardware costs. The PPG sensor 2 is arranged adjacent to the heart sound / voice sensor 1, making the two signal sources highly close in space, reducing measurement errors caused by differences in physiological position, and facilitating subsequent fusion analysis. Furthermore, the configuration of the multi-channel synchronous analog front-end circuit ensures that the heart sound and PPG signals achieve high-precision time synchronization at the acquisition source, effectively overcoming the delay and jitter problems that may be caused by software synchronization.

[0029] Among them, the heart sound / voice sensor 1 includes a piezoelectric vibration sensor, which is a bone conduction sensor; or the heart sound / voice sensor 1 includes a microelectromechanical system microphone.

[0030] Specifically, the heart sound / voice sensor 1 is a high-sensitivity MEMS microphone or piezoelectric vibration sensor with a frequency response covering 20-1000Hz and an equivalent noise level of less than 30dBA. Both of these devices have the characteristics of low power consumption, small size and high sensitivity, and are especially suitable for long-term stable operation under the limited space and power supply constraints of the ring.

[0031] Accordingly, the PPG sensor 2 consists of LEDs and photodetectors that support multi-wavelength configurations, and integrates an ambient light suppression circuit. The multi-channel synchronous analog front-end circuit not only supports hardware-level synchronous sampling of sensing channels such as heart sounds (1-2kHz) and PPG (100-500Hz), with a time deviation between channels of less than 100 microseconds, but also each channel is equipped with an independent programmable gain amplifier and filter.

[0032] Furthermore, the multimodal signal acquisition module 10 also includes a pair of ECG electrodes 3 for acquiring ECG signals; the multi-channel synchronous analog front-end circuit is electrically connected to the ECG electrodes 3, the heart sound / speech sensor 1 and the PPG sensor 2 for synchronously acquiring heart sound signals, PPG signals and ECG signals, and the inter-channel time synchronization deviation of the multi-channel synchronous analog front-end circuit is less than 100 microseconds.

[0033] This application achieves high-precision hardware-level synchronous acquisition of ECG signals, heart sound signals, and PPG signals by further integrating a pair of ECG electrodes 3 into a smart ring and utilizing a multi-channel synchronous analog front-end circuit with a time synchronization deviation of less than 100 microseconds between channels. The extremely high synchronization accuracy (<100 microseconds) fundamentally ensures the accuracy of subsequent calculations of key timing parameters such as pulse wave conduction time, providing a more reliable and information-rich raw data foundation for the establishment of models such as blood pressure and baroreflex sensitivity. Simultaneously, the introduction of ECG signals not only serves as a redundant reference to improve system robustness but also directly expands the functional boundaries of the device, enabling it to synchronously provide classic electrophysiological parameters such as heart rate variability. This achieves integrated, high-precision panoramic monitoring of the cardiovascular system's "electrical activity-mechanical activity-peripheral vascular activity" on a single micro wearable device 100, significantly enhancing the system's clinical value and the comprehensiveness of health assessment.

[0034] Furthermore, such as Figure 3 , 4 As shown, the physiological parameters include blood pressure parameters. The physiological parameter processing module 20 is configured to generate blood pressure parameters through the following steps: extracting the first heart sound feature point from the heart sound signal and extracting the pulse wave initiation point from the PPG signal. Using the first heart sound feature point as a reference for the onset of cardiac contraction, the time interval between the first heart sound feature point and the pulse wave initiation point is calculated. The time interval is input into a pre-trained personalized calibration model, which outputs estimates of systolic and diastolic blood pressure.

[0035] The personalized calibration model is configured to be built based on the initial calibration data and updated according to the real-time acquired speech signals and physiological parameters.

[0036] At the core algorithm level, the system ensures accurate time-stamp alignment of multimodal signals through hardware synchronization and software timestamps. Specifically, the heart sound signal processing employs a dedicated workflow: preprocessing using bandpass filtering and adaptive noise cancellation; extracting the envelope using Hilbert transform or the Teager energy operator and combining it with PPG information for heart sound localization and segmentation; comprehensively utilizing heart sound and PPG information, or employing a U-Net-based deep learning model, to achieve accurate detection of feature points for the first heart sound S1 / second heart sound S2; and extracting multi-dimensional features in the time domain, frequency domain, time-frequency domain, and morphology.

[0037] Specifically, the blood pressure estimation algorithm mainly includes two paths: one is a personalized calibration model based on heart sound-PPG feature time (PTT_audio) (e.g., BP=α×ln(PTT_audio)+β×PTT_audio+γ×HR+δ); the other is a machine learning model with multiple features as input, such as PTT_audio, heart rate, and heart rate variability (e.g., gradient boosting decision tree). The system completes personalized calibration through initial paired measurements, dynamic data collection, and long-term adaptive updates.

[0038] This application replaces the traditional approach of relying on the ECG R wave by extracting the first heart sound (S1) feature point from the heart sound signal and using it as a reference for the onset time of cardiac mechanical contraction. This fundamentally eliminates the physiological variations and errors introduced by the "electromechanical delay," making the pulse wave conduction time interval, which is the core of the calculation, more accurate and stable. Secondly, it introduces an updatable personalized calibration model. The system can not only perform individualized modeling based on initial data to overcome the limitations of universal population models, but also dynamically adjust and optimize the model based on long-term real-time collected voice signals (which can reflect state and symptoms) and physiological parameters. This allows the blood pressure estimate to continuously adapt to the user's physiological changes, achieving truly long-term, adaptive, and accurate monitoring.

[0039] Furthermore, the physiological parameter processing module 20 has a built-in blood pressure classification unit, which is used to classify blood pressure parameters according to pre-stored classification criteria and take different early warning strategies based on the classification results.

[0040] For example, the blood pressure classification criteria are based on international hypertension management guidelines and can automatically classify estimated blood pressure values ​​into normal, elevated, stage 1 hypertension, stage 2 hypertension, or stage 3 hypertension. Specifically, the classification thresholds can be set as follows: systolic blood pressure <120 mmHg and diastolic blood pressure <80 mmHg is "normal"; systolic blood pressure 120-129 mmHg and diastolic blood pressure <80 mmHg is "elevated"; systolic blood pressure 130-139 mmHg or diastolic blood pressure 80-89 mmHg is "stage 1 hypertension"; systolic blood pressure 140-179 mmHg or diastolic blood pressure 90-119 mmHg is "stage 2 hypertension"; and systolic blood pressure ≥180 mmHg or diastolic blood pressure ≥120 mmHg is "stage 3 hypertension".

[0041] This blood pressure classification and early warning system integrates multiple intelligent early warning mechanisms, including instant threshold warning, continuous trend warning, day-night mode warning, and multi-parameter comprehensive warning, forming a rule engine that automatically matches different levels of response strategies based on classification results. The instant threshold warning is triggered when a single blood pressure measurement reaches "Grade 2 hypertension" or higher; the continuous trend warning is triggered when three consecutive measurements within a preset time window (e.g., 10 minutes) are all at "Grade 1 hypertension" or higher, or when the blood pressure shows a continuous upward trend; the day-night mode warning is triggered by analyzing 24-hour blood pressure data and identifying abnormal blood pressure rhythm patterns such as "non-dipper" or "reverse dipper"; and the multi-parameter comprehensive warning is triggered when, along with elevated blood pressure, abnormal heart rate, heart sound characteristics, or baroreflex sensitivity (BRS) values ​​are detected. The corresponding response strategies include: for "elevation" or "Grade 1 hypertension", the system will send a single short vibration of the ring and / or push gentle health advice through the application; for "Grade 2 hypertension", the system will trigger intermittent strong vibration of the ring and a prominent pop-up alarm in the mobile application, prompting the user to rest or retest; for "Grade 3 hypertension" or any level of critical warning (such as a rapid rise in blood pressure), the system will immediately trigger continuous strong vibration of the ring and an audible alarm, and will automatically upload an emergency alarm package containing real-time data, location and warning level to the remote service platform through the data upload module 40. It can also automatically dial a preset emergency contact number through the associated mobile terminal 200 when the user confirms or the system determines that there is no response.

[0042] This application integrates a blood pressure classification unit into the physiological parameter processing module 20, enabling real-time, intelligent status assessment and risk grading intervention of continuously monitored blood pressure parameters. This allows the system to go beyond simple data display and possess preliminary medical decision support capabilities: for minor abnormalities, it can provide mild reminders or lifestyle suggestions; for significant or critical increases in blood pressure, it can activate strong reminders, local alarms, and even send emergency alerts to a remote service platform in conjunction with the data upload module 40, thereby constructing a proactive health management chain from daily monitoring to tiered early warning. This significantly lowers the barrier for users to understand professional data and improves the timeliness of health risk perception; on the other hand, through differentiated early warning strategies, it avoids early warning fatigue, ensures response efficiency and reliability in truly high-risk situations, and enhances the system's practicality and security.

[0043] Furthermore, the physiological parameters include baroreflex sensitivity parameters, and the physiological parameter processing module 20 is configured to generate baroreflex sensitivity parameters through the following steps: Extract the continuous first heart sound time series from the continuous heart sound signal, and calculate the heartbeat interval sequence based on the first heart sound time series; Obtain the systolic blood pressure sequence from continuously generated blood pressure parameters; The correlation between systolic blood pressure sequences and intercardiac interval sequences was analyzed using sequential or spectral methods, and baroreflex sensitivity parameters were calculated.

[0044] Specifically, the pressure reflection sensitivity estimation algorithm uses a sequence method (time domain) to find positively correlated sequences between systolic blood pressure sequences and intercardiac interval sequences and calculates the regression slope; or it uses a spectrum method (frequency domain) to perform synchronous spectrum analysis on the two sequences and calculate the transfer function gain in the low-frequency band as the BRS estimate.

[0045] This application calculates the heart rate interval by directly extracting the first heart sound time series from continuous heart sound signals, replacing the traditional reliance on electrocardiograms and ensuring the same high temporal accuracy and signal source consistency as blood pressure monitoring systems. Simultaneously, it directly obtains synchronized systolic blood pressure sequences using continuously generated blood pressure parameters, thus constructing high-quality, homologous blood pressure-heart rate interval data pairs. Based on this, correlation analysis using sequence methods or spectral methods can accurately calculate baroreflex sensitivity parameters. This technical approach successfully transfers BRS assessment, previously only achievable under laboratory conditions with complex equipment, to wearable smart rings, enabling long-term, dynamic monitoring of autonomic nervous system function. It provides an unprecedentedly convenient tool for cardiovascular risk assessment, disease prognosis, and efficacy observation, greatly enhancing the system's clinical depth and health management value.

[0046] Furthermore, the physiological parameter processing module 20 includes a multimodal AI fusion estimation model, which is configured to predict based at least on the heart sound signal and the PPG signal, and output at least two physiological parameters among blood pressure parameter, baroreflex sensitivity parameter, heart rate variability and vascular elasticity index.

[0047] Specifically, such as Figure 6 As shown, the multimodal AI fusion estimation model includes an input layer, a feature extraction branch, a fusion layer, and an output layer. The input layer takes into account heart sound signals (i.e., multi-channel time series), PPG signals, and motion signals from different locations. The feature extraction branch includes a heart sound processing network, a PPG processing network, and a motion signal processing network. The heart sound processing network extracts features from the heart sound signals. The PPG processing network extracts features from the PPG signals, and the motion signal processing network extracts features from the motion signals. The fusion layer fuses the features extracted from the heart sound signals, PPG signals, and motion signals. The output layer outputs blood pressure, BRS, HRV, etc.

[0048] This application uses fused features extracted from heart sounds and PPG signals as input, and employs an end-to-end deep learning framework to estimate multiple key parameters, including blood pressure, baroreflex sensitivity, heart rate variability, and vascular elasticity index, in a parallel and collaborative manner. This design overcomes the limitations of traditional methods that calculate parameters sequentially or independently. On one hand, the model can fully mine and utilize deep, complementary physiological correlations between different signal modes, enabling mutual verification and enhancement of the parameter estimation process, thus achieving better overall estimation accuracy and robustness than a single model. On the other hand, integrated estimation significantly reduces repetitive signal preprocessing and feature extraction steps, lowering processing latency and power consumption, making it more suitable for resource-constrained wearable devices. Furthermore, it simultaneously outputs multiple pathophysiologically closely related parameters, providing users and medical professionals with a more comprehensive and integrated panoramic view of the cardiovascular system, greatly enhancing the comprehensiveness and reliability of health risk assessment and early warning.

[0049] Furthermore, the multimodal AI fusion estimation model employs an attention-based neural network architecture to dynamically adjust the contribution weights of the heart sound signal and the PPG signal based on the features extracted from the heart sound signal and the PPG signal.

[0050] This application introduces an attention-based neural network architecture into the multimodal AI fusion estimation model, which enables dynamic and intelligent allocation of the contribution weights of heart sounds and PPG signals, thereby significantly improving the estimation performance and robustness of the model in complex real-world scenarios.

[0051] In terms of interaction, when the wearable body 60 is a smart ring, the interaction trigger module 30 includes a touch switch 6, which is configured to recognize at least two predefined gestures among single click, double click, long press or swipe.

[0052] For example, the predefined gestures correspond to functions: a single click is used to query device status, a double click is used to initiate an emergency remote consultation, a long press is used for device pairing, and a swipe is used to switch between monitoring and communication modes. Specifically, in an interactive remote medical communication system, when a user triggers emergency communication by double-clicking, the system executes an automated process: after ring vibration confirmation, it initiates an audio link establishment protocol, switches the heart sound / voice sensor 1 to bone conduction microphone mode, and packages and sends an "electronic medical record summary package" containing recent physiological trends and health summaries to the cloud. After the doctor receives the consultation, a two-way bone conduction audio stream is established to achieve low-latency, high-privacy hands-free calling.

[0053] This application fully utilizes the inherent dexterity of the fingers, allowing users to directly switch modes, trigger functions, or request emergency assistance on the same device using pre-defined, clearly defined gestures such as single clicks, double clicks, long presses, or swipes. This eliminates the need to operate the phone or search for specific buttons, achieving a seamless "hand-to-hand" experience and significantly improving accessibility and speed in emergencies such as sudden discomfort. Furthermore, mapping complex functions to simple gestures maintains the ring's clean and compact appearance while providing rich control capabilities, effectively solving the problem of limited physical interfaces on micro-devices. In addition, direct touch operation avoids privacy issues that might arise from voice activation or screen-on operation in public places, enhancing the privacy of the interaction. This innovative interaction method not only significantly improves the user experience but also transforms health monitoring devices from passive data loggers into intelligent interactive terminals that users can actively and seamlessly engage with.

[0054] like Figure 2 As shown, in a preferred embodiment, the audio feedback module 50 employs a bone conduction microphone and a bone conduction vibrator 7, which are mounted on the wearable body 60. The audio feedback module 50 utilizes active noise reduction technology and echo cancellation algorithms to specifically optimize the quality of voice communication via bone conduction or skin contact on devices such as smart rings, ensuring clear and intelligible voice signals even in noisy environments and guaranteeing effective communication during remote consultations.

[0055] Furthermore, the remote service platform includes a mobile terminal 200 and a cloud service platform 300. The interaction trigger module 30 also includes a wireless communication module, which is configured to: when establishing a communication link with the remote service platform, prioritize direct connection with the cloud service platform 300 via a cellular mobile communication network; when the direct connection fails, switch to connecting to the mobile terminal 200 via a short-range wireless communication protocol, and establish a relay connection between the mobile terminal 200 and the cloud service platform 300.

[0056] The system architecture of this invention is designed in three layers: the smart ring itself, serving as a data acquisition and primary processing terminal; the mobile terminal 200 application, acting as an edge computing node and intermediary for user interaction; and the cloud service platform 300, providing advanced computing, data storage, and professional medical services. The mobile terminal 200 application handles data aggregation, lightweight processing, and the interactive interface, while the cloud platform provides secure data storage, complex AI analysis, an online consultation platform, an intelligent early warning system, and health management services.

[0057] This application significantly improves the reliability, adaptability, and user experience of establishing communication links between the system and the remote service platform by configuring a wireless communication module with an intelligent dual-mode connection strategy. This strategy prioritizes establishing a direct connection between the wearable device 100 and the cloud service platform 300 via a cellular mobile communication network. This ensures that the device maintains independent communication capabilities even when the mobile terminal 200 is not nearby or unable to assist (e.g., when the user is exercising outdoors), enabling data upload and service access anytime, anywhere, thus enhancing the system's autonomy and availability. Furthermore, the direct connection reduces intermediate data transmission links, helping to reduce transmission latency and improve privacy and security. When the direct connection fails due to poor network signal or other reasons, the system can automatically and seamlessly switch to a backup path, connecting to the user's mobile terminal 200 via a short-range wireless communication protocol and using the latter as a relay gateway to establish a connection with the cloud. This intelligent redundancy design ensures that the communication link is eventually established in most scenarios, greatly improving the system's connection success rate and robustness in complex real-world environments, and avoiding the risk of critical health data failing to be uploaded or remote services failing to be triggered due to network problems.

[0058] Furthermore, the mobile terminal 200 is configured to communicate with the wearable device 100 and the cloud service platform 300, including: The data receiving module is used to receive heart sound signals and photoplethysmography (PPG) signals synchronously collected by the wearable device 100 via a short-range wireless communication protocol, or to receive at least two physiological parameters generated by the wearable device 100 based on the heart sound signals and PPG signals. An edge processing module is used to run a lightweight local processing model to generate the at least two physiological parameters, based at least on the heart sound signal and PPG signal, when the raw signal is received. The user interface is used to visually display the physiological parameters, historical data trends, and historical voice data generated based on audio feedback to the user. The relay communication module is used as a relay gateway when the communication connection between the wearable device 100 and the cloud service platform 300 fails to be established. It receives data uploaded by the wearable device 100 and forwards it to the cloud service platform 300, and forwards feedback information from the cloud service platform 300 to the wearable device 100.

[0059] The user interface is also configured to: receive and display warning information from the wearable device 100 or the cloud service platform 300; provide interactive controls for manually triggering remote medical consultation services; and display text or graphic guidance information from certified medical professionals that is synchronously pushed by the cloud service platform 300 during the remote consultation process.

[0060] like Figure 7 The diagram shows the user interface of the mobile terminal 200. The main interface visually displays the physiological parameters to the user, including a real-time vital signs dashboard. The historical data interface displays historical data trends, including trend graphs, and supports viewing by day / week / month. The consultation record interface displays historical voice data, including a list of historical consultations, and allows playback of recordings.

[0061] Furthermore, the cloud service platform 300 is configured to communicate with at least one wearable device 100 and at least one mobile terminal 200, and the cloud service platform 300 includes: The data aggregation and storage service is used to receive and securely store user multimodal physiological data uploaded from the wearable device 100 or mobile terminal 200, the data including at least heart sound signals, photoplethysmography (PPG) signals, at least two physiological parameters generated based thereon, and user voice signals. Advanced analytics and computing services are used to run sophisticated AI analytics models to perform in-depth analysis of users’ long-term physiological data, predict trends, assess health risks, and generate personalized health summary reports. The telemedicine interface service provides an online consultation platform that connects certified medical professionals. When a user initiates a consultation, the user's real-time physiological parameters, health summary report, and voice communication link are simultaneously pushed to the terminal of the receiving medical professional. The intelligent early warning and notification service is used to analyze real-time uploaded physiological parameters based on preset clinical rules and machine learning models. When abnormal patterns or risks are identified, it generates graded early warning information and pushes it to the user's mobile terminal 200 and / or associated medical professional terminals.

[0062] In remote service platforms, particularly on the terminal interfaces used by doctors or health managers, when a user initiates a remote consultation, the system simultaneously displays the user's real-time physiological parameter data, historical data trend curves, and a health summary report automatically generated by the system. This provides professionals with a comprehensive and intuitive view for decision support.

[0063] like Figure 9 As shown, this application also proposes an interactive multimodal physiological parameter intelligent monitoring method, applied to an interactive multimodal physiological parameter intelligent monitoring system. The method includes: Simultaneously collect the user's heart sound signals and PPG signals; At least two physiological parameters are generated in real time based on heart sound signals and PPG signals; In response to a trigger signal from the user, a communication link is established with the remote service platform, and the multimodal signal acquisition module 10 is switched to acquire the user's voice signal; The voice signal and physiological parameters are uploaded to the remote service platform simultaneously. Receive and play audio feedback from the remote service platform and implement early warning functions.

[0064] The interactive multimodal physiological parameter intelligent monitoring method proposed in this application overcomes the electromechanical delay problem of traditional methods by simultaneously acquiring heart sounds and PPG signals, using heart sounds directly instead of ECG as the time reference. This provides a better foundation for the continuous and accurate estimation of parameters such as blood pressure and baroreflex sensitivity. The multimodal AI fusion model and high-precision synchronous acquisition hardware further ensure data reliability. Users can trigger the process with a single finger gesture, automatically switching the device to communication mode and simultaneously uploading real-time physiological parameters and voice descriptions to a remote platform. Users can also directly obtain audio feedback through the device, constructing an instant service loop of "monitoring-diagnosis-feedback," shortening the help path and protecting privacy.

[0065] To make the technical solution of this invention clearer and more complete, the invention will be described in detail below with reference to specific embodiments. These embodiments are intended to demonstrate typical application scenarios and workflows of this invention, and are not intended to limit the invention.

[0066] Example 1: Routine Health Monitoring This embodiment describes a scenario in which a 45-year-old hypertensive patient wears the smart ring described in this invention for long-term health management in daily life.

[0067] like Figure 1-8 As shown, the smart ring worn by the patient integrates a heart sound / voice sensor 1, a PPG sensor 2, an electrocardiogram (ECG) electrode, a temperature sensor 4, a six-axis inertial measurement unit (IMU) 5, a battery 8, a main circuit board 9, and a wireless charging coil 10. The battery 8 and the wireless charging coil 10 power the smart ring; the main circuit board 9 is connected to the heart sound / voice sensor 1, the PPG sensor 2, the ECG electrode, the temperature sensor 4, and the six-axis inertial measurement unit (IMU).

[0068] The ring connects to a smartphone via Bluetooth 5.2. It has a built-in 80mAh lithium polymer battery, providing up to 72 hours of battery life in default continuous monitoring mode. Upon first use, users pair the ring with the mobile app and perform three synchronized measurements using a certified upper arm blood pressure monitor to complete a personalized initial calibration of the blood pressure estimation model.

[0069] When the user naturally wakes up at 6:30 AM, the ring system automatically records and analyzes the "morning peak blood pressure." The analysis results show a systolic blood pressure of 145 mmHg and a diastolic blood pressure of 92 mmHg. Based on the built-in 2023 AHA / ACC hypertension guideline classification, the system classifies it as "Stage 2 hypertension." The ring then sends a slight vibration to the user, and simultaneously, the mobile application pushes a health recommendation containing the specific blood pressure readings and the advice, "Morning blood pressure is high; it is recommended to get up slowly and avoid strenuous activity."

[0070] After breakfast at 8:00 AM, the user initiated a precise measurement by simply tapping the touch area on the ring's surface. The ring entered active measurement mode, simultaneously acquiring high-quality heart sounds and PPG signals within 30 seconds. The processed results showed a current blood pressure of 138 / 85 mmHg and a heart rate of 68 beats per minute. This measurement data was automatically recorded and synchronized to the phone's health application and cloud-based health records.

[0071] At 12:30 PM, the user engaged in light activity after lunch. Continuous monitoring revealed a physiological decrease in blood pressure, with a systolic pressure of 126 mmHg and a diastolic pressure of 81 mmHg. The system marked this "postprandial hypotension" phenomenon in the log, information that will help in a more comprehensive assessment of the user's dynamic blood pressure pattern.

[0072] At 22:00, the system automatically switches to sleep monitoring mode based on the user's historical sleep patterns or the resting state detected by the IMU. In this mode, to balance power consumption and data integrity, the system adjusts to synchronous sampling of heart sounds and PPG every 30 minutes. Through analysis of the nighttime data, the system found that the user's nighttime blood pressure drop was less than 10% of the average daytime blood pressure, exhibiting a "non-dipper blood pressure" pattern. This finding was recorded as an important cardiovascular risk warning.

[0073] At 6:00 AM the following morning, the system automatically generated a comprehensive health report for the past 24 hours. The report showed that the user's 24-hour average blood pressure was 136 / 84 mmHg, with a high blood pressure variability (BPV) index, while the calculated baroreflex sensitivity (BRS) value was 4.2 ms / mmHg, which was at a low level. Based on these analysis results, the system provided personalized management suggestions through the application, such as: "Consider discussing with your doctor adjusting the timing of your antihypertensive medication to the evening to improve nighttime blood pressure patterns; and it is recommended to increase moderate aerobic activity during the day, which may help improve autonomic nervous system regulation." Example 2: Remote consultation in an emergency This embodiment demonstrates how users can use the system of this invention to seek immediate help and obtain professional guidance during a sudden health crisis.

[0074] A 58-year-old male user with a 10-year history of hypertension suddenly experienced severe dizziness accompanied by blurred vision. When the discomfort occurred, the user did not need to look for his phone; he simply double-tap the touch area of ​​the smart ring he was wearing with a finger of his other hand.

[0075] Upon receiving the double-tap command, the ring immediately responded with two short vibrations as confirmation. The system then activated the highest-priority emergency protocol. Equipped with a 5G RedCap module, the ring prioritized and successfully established a direct data connection with the cloud-based medical service platform, bypassing the intermediary smartphone.

[0076] Upon establishing a connection, the system automatically transmits a structured "emergency data packet" to the cloud. This data packet contains the following key information: the current real-time estimated blood pressure value (192 / 118 mmHg, automatically labeled as critical stage 3 hypertension by the system), a trend chart showing the continuous and rapid rise in blood pressure over the past 10 minutes, the current irregular high heart rate (112 bpm), and a summary of the user's pre-filled personal health record (including age, gender, history of hypertension, etc.).

[0077] Upon receiving the emergency request, the cloud platform prioritized its allocation to an online emergency room doctor using an algorithm. The doctor completed the consultation within approximately 3 seconds of the request being sent. The doctor's consultation interface prominently displayed the warning "[Emergency] Stage 3 hypertension, accompanied by arrhythmia," along with the user's real-time vital signs dashboard and historical trend curves.

[0078] Real-time interaction and medical intervention: A two-way bone conduction audio link was immediately established. The doctor issued a reassuring instruction to the user via computer: "This is the remote emergency center. Please remain calm. I am monitoring your blood pressure and heart rate data in real time. The situation is indeed quite urgent." The user's response was clearly collected and transmitted through the heart sound / voice sensor 1 on the ring (now switched to bone conduction microphone mode): "I feel very dizzy and have chest tightness..." Combining the real-time data and the user's complaints, the doctor immediately provided initial guidance: "Please sit down immediately, lean against a support, and try to take deep breaths. Do you have any spare fast-acting antihypertensive medication, such as nifedipine, with you?" The entire instructional call lasted approximately 5 minutes.

[0079] To further ensure user safety, in cases where the doctor authorizes or the system determines the risk to be extremely high, the integrated emergency service module can automatically access the phone's permissions to dial local emergency numbers (such as 120), and send the user's real-time location, the generated emergency data package, and a summary of the consultation record to the emergency center that is about to respond. The complete timeline of this emergency event, all interaction data, and medical instructions are automatically archived by the system into a complete electronic emergency record, which can be pushed to the information system of the hospital that subsequently treats the user, upon user authorization.

[0080] Example 3: Postoperative Rehabilitation Monitoring This embodiment is applied to the home rehabilitation stage of patients after coronary stent surgery, demonstrating the system's continuous and comprehensive value in chronic disease management.

[0081] For these patients, the core monitoring objectives focus on three points: first, to strictly control blood pressure within the target range (usually <130 / 80 mmHg); second, to objectively quantify the recovery of autonomic nerve function by continuously assessing baroreflex sensitivity (BRS); and third, to monitor the physiological response after taking various medications (especially antihypertensive drugs and anticoagulants).

[0082] To achieve refined management, the system has implemented several advanced functions. First, the intelligent medication reminder function goes beyond simple timed reminders; it integrates with the user's individual blood pressure circadian rhythm, issuing a vibration reminder at an appropriate time before their blood pressure typically begins to rise, thus optimizing medication efficacy. Second, based on the patient's postoperative recovery stage, the system provides personalized rehabilitation exercise plans through an application, using IMU data to monitor activity level and intensity to ensure safe and effective exercise. Third, with user authorization, the system can be set to automatically share daily or weekly monitoring report summaries with the attending physician, facilitating remote monitoring of rehabilitation progress. Fourth, the system has set continuous monitoring thresholds for BRS (Brain Regulatory System). If the BRS value remains below the preset safety threshold for several consecutive days, the system generates an abnormal warning indicating "potentially delayed recovery of autonomic nervous function," prompting the user to seek timely follow-up examination.

[0083] Example 4: Sports and Health Management This embodiment is intended for fitness enthusiasts or people who exercise regularly, demonstrating the application of the system in exercise physiological monitoring.

[0084] Users wear the ring continuously during exercise. The system focuses on monitoring the following parameters: blood pressure response during exercise, observing whether the rise is within reasonable expectations; cardiac load, indirectly assessing myocardial contractility by analyzing the amplitude and frequency changes of the heart sound signal S1 during exercise; and the rate of blood pressure decrease and the recovery speed of BRS value during exercise recovery. These two indicators are important bases for assessing an individual's cardiovascular system recovery ability and exercise tolerance.

[0085] Throughout the exercise, users do not need to frequently check their phones. The system interacts in two ways: first, when the heart rate or blood pressure exceeds the user's personal safety range, it provides a real-time alert via the vibration of the ring, suggesting that the user reduce the intensity of the exercise; second, at the user's preset rest intervals, it provides immediate feedback by briefly announcing key indicators (such as current heart rate and duration of the exercise interval) via the bone conduction oscillator 7. After the exercise, the system automatically analyzes the data from the entire exercise period and recovery period, generating a comprehensive exercise report that includes exercise load assessment, cardiac response analysis, and recovery ability score, helping users to scientifically adjust their training plans.

[0086] Based on the above embodiments, the present invention achieves multiple integrated innovations through systematic design, which are summarized as follows: 1. Signal Modality Innovation: In a wearable ring form factor, heart sound signals are creatively used to replace traditional electrocardiogram (ECG) signals as the proximal reference for calculating pulse wave conduction time, eliminating electromechanical delay errors and providing a more solid theoretical foundation. Hardware synchronization and deep fusion analysis of multimodal physiological information such as heart sounds, PPG, ECG, and motion signals are achieved. A sensor multiplexing mechanism is designed, enabling the heart sound / voice sensor 1 to perform both physiological monitoring and bone conduction communication functions.

[0087] 2. Algorithm Innovation: A heart sound feature extraction algorithm optimized for ring-based data acquisition scenarios was developed, enabling robust localization of S1 / S2 even under conditions of weak signal and high interference. A multi-task learning AI fusion model was adopted, capable of simultaneously and accurately estimating multiple physiological parameters such as blood pressure, BRS, and HRV from multimodal inputs. A continuous personalized adaptive calibration process was established, allowing the algorithm model to continuously optimize as the user's usage time increases, improving individual applicability.

[0088] 3. Interactive Innovation: It achieves screenless interaction based on touch gestures and haptic feedback, allowing users to control complex devices without visual attention. A one-click emergency medical consultation path is designed, simplifying the process of seeking professional help to the extreme. The innovative application of bone conduction technology in two-way privacy communication effectively protects user privacy while ensuring call clarity.

[0089] 4. System Innovation: A complete closed-loop health management service has been constructed, encompassing continuous monitoring, intelligent analysis, risk warning, and real-time remote medical intervention. International clinical guidelines (such as hypertension classification standards) are directly embedded into the core logic of the device, enabling automatic integration of monitoring results with clinical decision support. A distributed intelligent computing architecture with "end (ring)-edge (phone)-cloud (platform)" collaboration is adopted, ensuring powerful functionality while also considering the device's power consumption and response speed.

[0090] Furthermore, to ensure that the product of this invention can meet the expected performance, safety, and compatibility standards, the following describes the key non-functional requirements: Manufacturing requirements: The inner lining of the ring that comes into contact with the skin must be made of medical-grade silicone or thermoplastic polyurethane (TPU) material that has passed the ISO10993 biocompatibility test to ensure long-term wear safety. The overall waterproof rating must reach IP68 or higher to withstand daily hand washing, sweating, and rain. To accommodate different users, multiple size options with ring inner diameters ranging from 16mm to 24mm in 1mm increments must be provided. The device's operating temperature range should cover 0°C to 45°C, and its storage temperature range should cover -20°C to 60°C to adapt to various living environments.

[0091] Clinical validation requirements: Blood pressure measurement accuracy must be strictly validated according to international standards such as ISO 81060-2 (applicable to relevant parts of wearable non-invasive blood pressure monitoring devices), and an authoritative test report must be issued. Repeatability tests must be conducted under different wearing tightness and limb activity states to demonstrate measurement stability. The product must undergo long-term stability testing for no less than 6 months to ensure that performance does not significantly degrade over time. Clinical trials must cover multiple typical populations, including hypertensive patients, normotensive individuals, the elderly, and young adults, to demonstrate its broad applicability.

[0092] Software Ecosystem: The system should provide open application programming interfaces (APIs) that allow third-party health or fitness applications to securely access relevant data with user authorization. Health data transmission and storage formats should support internationally recognized medical data exchange standards such as HL7 and FHIR. The cloud platform must have the capability to interface with mainstream hospital electronic health record systems or regional medical information platforms using standard interfaces. The user interface and voice interaction should support at least Chinese and English to meet the needs of users in different regions.

[0093] Business Model: This invention can realize value through various business models: directly selling smart ring hardware devices to consumers; providing users with value-added services such as advanced data analysis, personalized health reports, and remote doctor consultations through a subscription model; providing customized group health management solutions for hospitals, medical examination centers, insurance companies, or large enterprises; and using aggregated anonymized group health data for public health research or medical service optimization, provided that the data is strictly desensitized and complies with ethical regulations.

[0094] Those skilled in the art will understand that the modules or steps described above in this application can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. Optionally, they can be implemented using computer-executable program code, thereby allowing them to be stored in a storage device for execution by a computing device, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, this application is not limited to any particular combination of hardware and software.

[0095] Note that the above description is merely a preferred embodiment and the technical principles employed in this application. Those skilled in the art will understand that this application is not limited to the specific embodiments described herein, and various obvious changes, readjustments, and substitutions can be made without departing from the scope of protection of this application. Therefore, although this application has been described in detail through the above embodiments, this application is not limited to the above embodiments. Many other equivalent embodiments may be included without departing from the concept of this application, and the scope of this application is determined by the scope of the appended claims.

[0096] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. An interactive multimodal physiological parameter intelligent monitoring system, characterized in that, This includes wearable devices and remote service platforms, wherein the wearable device comprises a wearable body worn on the user's fingers or wrist and a component located on the wearable body: The multimodal signal acquisition module is used to simultaneously acquire the user's heart sound signal and PPG signal, and switch to acquire the user's voice signal. A physiological parameter processing module is used to generate at least two physiological parameters based on at least the heart sound signal and the PPG signal; An interactive trigger module is used to establish a communication link with the remote service platform when triggered by a user, and to switch the multimodal signal acquisition module to acquire the voice signal; The data upload module is used to synchronously upload the voice signal and the physiological parameters to the remote service platform; The audio feedback module is used to receive and play audio feedback from the remote service platform and to implement the early warning function.

2. The interactive multimodal physiological parameter intelligent monitoring system according to claim 1, characterized in that, The wearable device includes a smart ring, and the multimodal signal acquisition module includes: At least one heart sound / voice sensor is disposed on the side wall of the smart ring for contact with the palmar skin of the proximal phalanx of the finger and for collecting the user's heart sound signal and / or the user's voice signal; At least one PPG sensor is disposed on the side of the smart ring near the heart sound / voice sensor for collecting the PPG signal; A multi-channel synchronous analog front-end circuit is electrically connected to the heart sound / speech sensor and the PPG sensor, and is configured to synchronously acquire the heart sound signal and the PPG signal; The heart sound / voice sensor includes a piezoelectric vibration sensor, which is a bone conduction sensor; or the heart sound / voice sensor includes a microelectromechanical system (MEMS) microphone.

3. The interactive multimodal physiological parameter intelligent monitoring system according to claim 2, characterized in that, The multimodal signal acquisition module also includes a pair of electrocardiogram electrodes for acquiring electrocardiogram signals; The multi-channel synchronous analog front-end circuit is electrically connected to the ECG electrodes, the heart sound / speech sensor, and the PPG sensor, and is used to synchronously acquire the heart sound signal, the PPG signal, and the ECG signal, and the inter-channel time synchronization deviation of the multi-channel synchronous analog front-end circuit is less than 100 microseconds.

4. The interactive multimodal physiological parameter intelligent monitoring system according to claim 1, characterized in that, The physiological parameters include blood pressure parameters, and the physiological parameter processing module is configured to generate the blood pressure parameters through the following steps: Extract the first heart sound feature point from the heart sound signal and extract the pulse wave origin from the PPG signal; Using the first heart sound feature point as a reference for the onset time of cardiac contraction, the time interval between the first heart sound feature point and the pulse wave initiation point is calculated. The time interval is input into a pre-trained personalized calibration model, which outputs estimates of systolic and diastolic blood pressure. The personalized calibration model is configured to be built based on initial calibration data and updated according to the real-time acquired speech signals and physiological parameters.

5. The interactive multimodal physiological parameter intelligent monitoring system according to claim 4, characterized in that, The physiological parameter processing module also has a built-in blood pressure classification unit, which is used to classify the blood pressure parameters according to pre-stored classification criteria and take different early warning strategies based on the classification results.

6. The interactive multimodal physiological parameter intelligent monitoring system according to claim 4 or 5, characterized in that, The physiological parameters include baroreflex sensitivity parameters, and the physiological parameter processing module is configured to generate baroreflex sensitivity parameters through the following steps: Extract a continuous first heart sound time series from the continuous heart sound signal, and calculate the heartbeat interval sequence based on the first heart sound time series; The systolic blood pressure sequence is obtained from the continuously generated blood pressure parameters; The correlation between the systolic blood pressure sequence and the intercardiac interval sequence was analyzed using either a sequence method or a spectral method, and the baroreflex sensitivity parameter was calculated.

7. The interactive multimodal physiological parameter intelligent monitoring system according to claim 1, characterized in that, The physiological parameter processing module includes a multimodal AI fusion estimation model, which is configured to predict based at least on the heart sound signal and the PPG signal, and output at least two physiological parameters among blood pressure parameter, baroreflex sensitivity parameter, heart rate variability and vascular elasticity index. And / or, the multimodal AI fusion estimation model employs a neural network architecture based on an attention mechanism, used to dynamically adjust the contribution weights corresponding to the heart sound signal and the PPG signal based on the features extracted from the heart sound signal and the PPG signal.

8. The interactive multimodal physiological parameter intelligent monitoring system according to claim 1, characterized in that, When the wearable device is a smart ring, the interaction trigger module includes a touch switch, which is configured to recognize at least two predefined gestures among single click, double click, long press, or swipe.

9. The interactive multimodal physiological parameter intelligent monitoring system according to claim 8, characterized in that, The remote service platform includes a mobile terminal and a cloud service platform. The interaction triggering module also includes a wireless communication module. The wireless communication module is configured to: when establishing a communication link with the remote service platform, prioritize direct connection with the cloud service platform via a cellular mobile communication network; when the direct connection fails, switch to connecting to the mobile terminal via a short-range wireless communication protocol, and establish a relay connection between the mobile terminal and the cloud service platform.

10. An interactive, multimodal, intelligent monitoring method for physiological parameters, characterized in that: The method, applied to the interactive multimodal physiological parameter intelligent monitoring system as described in any one of claims 1-9, comprises: Simultaneously collect the user's heart sound signal and PPG signal; Based on the heart sound signal and the PPG signal, at least two physiological parameters are generated in real time. In response to a trigger signal from the user, a communication link is established with the remote service platform, and the multimodal signal acquisition module is switched to acquire the user's voice signal; The voice signal and the physiological parameters are simultaneously uploaded to the remote service platform; Receive and play audio feedback from the remote service platform and implement early warning functions.

Images