Home continuous fever early warning method and system based on wearable sign monitoring

Abstract

This invention discloses a method and system for home-based persistent fever early warning based on wearable vital sign monitoring. The method includes: S1, collecting multi-dimensional vital sign data of the user in real time through a wearable monitoring device, obtaining symptom data and vital sign data reported by the user, and preprocessing them; S2, constructing a fusion early warning model, inputting the preprocessed multi-dimensional vital sign data, symptom data, and vital sign data into the fusion early warning model, identifying whether the user is in a state of persistent fever, and classifying the risk; S3, sending early warning prompts and providing intervention suggestions to the user based on the risk classification results; S4, storing the multi-dimensional vital sign data, early warning records, and intervention feedback data, and periodically optimizing the parameters of the fusion early warning model. This invention achieves the determination and graded early warning of persistent fever at home by collecting multi-dimensional data through a wearable device and combining it with the calculation of a fusion early warning model.

Description

Technical Field

[0001] This invention relates to the field of medical monitoring and intelligent early warning technology, and more specifically to a method and system for home-based persistent fever early warning based on wearable vital sign monitoring. Background Technology

[0002] Fever is a common clinical symptom, while persistent fever (usually defined as a body temperature ≥38.0℃ lasting for more than 24 hours, or repeated fever for more than 3 days) often indicates potential infection, autoimmune disease, occult lesions, etc. If it is not detected and intervened in time at home, it may lead to the aggravation of the condition or even delay the best treatment time.

[0003] In existing technologies, home fever monitoring mostly relies on single body temperature measurement (such as electronic thermometers and forehead thermometers), which can only obtain body temperature data and cannot be combined with other vital signs, symptoms, and signs for comprehensive judgment, which has obvious limitations: on the one hand, single body temperature data is difficult to distinguish between physiological fever (such as after exercise or environmental factors) and pathological persistent fever, which is prone to false alarms or missed alarms; on the other hand, it cannot capture other abnormal signals that accompany fever (such as abnormal heart rate, shortness of breath, fatigue, rash, etc.), making it difficult to accurately classify the risk of fever and unable to provide targeted intervention suggestions for home users.

[0004] While existing technologies include fever warning and intelligent diagnosis, they are mostly focused on hospital scenarios, post-operative scenarios, or diagnosis of single causes. They do not offer wearable, multi-dimensional monitoring solutions for home scenarios, and in particular, they do not achieve integrated warnings based on "body temperature + multiple vital signs + symptoms + signs," thus failing to meet the needs of home users for accurate monitoring, risk warnings, and home intervention guidance for persistent fever.

[0005] Therefore, developing a method and device for continuous fever risk early warning that can combine multi-dimensional data, adapt to home scenarios, and be accurate and efficient has become an urgent technical problem to be solved. Summary of the Invention

[0006] The purpose of this invention is to provide a method and system for early warning of persistent fever at home based on wearable vital sign monitoring. By collecting multi-dimensional data through wearable devices and combining it with a fusion early warning model, the invention enables the determination and graded early warning of persistent fever at home.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A home-based persistent fever early warning method based on wearable vital sign monitoring includes the following steps: S1. Collect multi-dimensional vital sign data of users in real time through wearable monitoring devices, obtain symptom data and vital sign data reported by users, and perform preprocessing. S2. Construct a fusion early warning model. Input the pre-processed multi-dimensional vital sign data, symptom data, and vital sign data into the fusion early warning model to identify whether the user is in a state of persistent fever and to classify the risk. S3. Based on the risk classification results, send early warning prompts to users and provide intervention suggestions; S4. Store multi-dimensional vital sign data, early warning data, and intervention suggestion data, and optimize the parameters of the fusion early warning model.

[0008] Furthermore, in S1, the multi-dimensional vital sign data includes: body temperature, heart rate, respiratory rate, blood oxygen saturation, skin moisture, pulse rate variability, and blood pressure; The symptom data includes one or more of the following: fever-associated symptoms, systemic symptoms, local symptoms, and mental status; The vital signs data include one or more of the following: lymph node enlargement, pharyngeal congestion, presence or absence of conjunctival congestion, presence or absence of skin rash and its location, and presence or absence of abdominal tenderness.

[0009] Furthermore, in S1, the preprocessing process includes: outlier removal, missing data imputation, and data standardization; The process of removing abnormal data is as follows: using statistical outlier removal methods, marking logically contradictory data reported by users and reminding users to check and correct it; The missing data filling process specifically involves: using interpolation to fill in missing data; and using wearable monitoring devices to remind users to supplement symptom and sign data that have been missing for a long time or that users have not reported on time. The data standardization process specifically involves: unifying the magnitude of vital sign data and quantifying and encoding symptom and vital sign data.

[0010] Furthermore, in S2, the criterion for determining the continuous heating state is: Body temperature ≥38.0℃ and lasting ≥24 hours; body temperature repeatedly ≥38.0℃, with an interval of ≤8 hours between two fevers, and a cumulative number of ≥3 fevers, lasting ≥3 days; body temperature ≥37.5℃ and lasting ≥48 hours accompanied by abnormal vital signs or symptoms; The risk classification includes: low risk, medium risk, and high risk.

[0011] Furthermore, in S3, the channels for sending warning prompts to users include at least two of the following: prompts from wearable monitoring devices, prompts from terminal apps, and prompts from associated personnel; The intervention recommendations include providing guidance on home observation, physical cooling, or immediate medical attention based on the risk analysis.

[0012] The present invention also provides a system for implementing a home-based persistent fever early warning method based on wearable vital sign monitoring, comprising: a wearable monitoring module, a terminal interaction module, a data processing module, a fusion early warning module, an early warning prompt module, and a data storage module; The wearable monitoring module is used to collect multi-dimensional vital sign data of users in real time through wearable monitoring devices; The terminal interaction module is used to acquire symptom data and vital sign data reported by the user. The data processing module is used to preprocess multi-dimensional vital sign data, symptom data, and physical sign data; The fusion early warning module is used to construct a fusion early warning model and input preprocessed multi-dimensional vital sign data, symptom data and vital sign data into the fusion early warning model to identify whether the user is in a state of persistent fever and to classify the risk. The early warning module is used to send early warnings to users and provide intervention suggestions based on the risk classification results. The data storage module is used to store multi-dimensional vital sign data, early warning data, and intervention suggestion data, and to optimize the parameters of the fusion early warning model.

[0013] Furthermore, the wearable monitoring module adopts a lightweight and waterproof design; The wearable monitoring module includes: a body temperature sensor, a heart rate / pulse rate sensor, a respiratory rate sensor, and a blood oxygen saturation sensor.

[0014] Furthermore, the terminal interaction module includes: a mobile APP, a wearable device display screen, a tablet, and a smart speaker; The terminal interaction module is configured to provide a data reporting interface and a data display interface, and to implement an emergency assistance function; Furthermore, the data processing module and the fusion early warning module are deployed in at least one of the cloud server, the terminal interaction module, or the wearable monitoring module; The warning module sends warnings to users in the following ways: vibration, light, voice, pop-up window, SMS or telephone.

[0015] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects: This invention effectively overcomes the limitations of single objective data monitoring by collecting multi-dimensional vital sign data in real time through wearable devices and combining it with information reported by users, thus improving the comprehensiveness and continuity of data collection in home settings. By utilizing a fusion early warning model to comprehensively analyze and classify multi-source heterogeneous data, it can more accurately identify persistent fever, improving the accuracy and timeliness of early warnings, thereby guiding users to take timely and targeted intervention measures to reduce health risks. By establishing a closed-loop mechanism for data storage and regular optimization of model parameters, the system achieves adaptive updates, enhancing the model's adaptability to different individual differences and providing scientific and reliable auxiliary decision support for home-based persistent fever management. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0017] The following description, in conjunction with the accompanying drawings, further illustrates the home-based persistent fever early warning method and system based on wearable vital sign monitoring of the present invention. Figure 1 This is a flowchart of a home-based persistent fever early warning method based on wearable vital sign monitoring, according to Embodiment 1 of the present invention. Detailed Implementation

[0018] The specific embodiments of the present invention will be described in further detail below with reference to the accompanying drawings and examples. The following examples are for illustrative purposes only and are not intended to limit the scope of the invention.

[0019] To better understand the purpose, structure, and function of this invention, the invention will be described in further detail below with reference to the accompanying drawings.

[0020] like Figure 1 As shown, a method for home-based persistent fever early warning based on wearable vital sign monitoring includes the following steps: S1. Collect multi-dimensional vital sign data of users in real time through wearable monitoring devices, obtain symptom data and vital sign data reported by users, and perform preprocessing. In S1, the multi-dimensional vital signs data include: body temperature, heart rate, respiratory rate, blood oxygen saturation, skin moisture, pulse rate variability, and blood pressure. The symptom data includes one or more of the following: fever-associated symptoms, systemic symptoms, local symptoms, and mental status; The vital signs data include one or more of the following: lymph node enlargement, pharyngeal congestion, presence or absence of conjunctival congestion, presence or absence of skin rash and its location, and presence or absence of abdominal tenderness.

[0021] In this embodiment, the multi-dimensional vital sign data is collected in real time by multiple sensors of the wearable monitoring device, including at least body temperature, heart rate, respiratory rate, and blood oxygen saturation, and optionally one or more of skin humidity, pulse rate variability (HRV), and blood pressure; the body temperature acquisition accuracy is ≥0.1℃, the heart rate acquisition range is 40-180 beats / minute, the respiratory rate acquisition range is 8-40 beats / minute, and the blood oxygen saturation acquisition range is 70%-100%; the acquisition frequency is once every 5-10 minutes, and when any vital sign data exceeds the preset normal range (such as body temperature ≥37.5℃, heart rate >100 beats / minute or <60 beats / minute), the acquisition frequency is automatically increased to once every 1-2 minutes; the sensors include, but are not limited to, thermistor sensors, photoplethysmography (PPG) sensors, infrared light detection sensors, and humidity sensors.

[0022] Symptom data is reported voluntarily by users through terminals (including but not limited to mobile apps, wearable device displays, tablets, and smart speakers). It includes at least one or more of the following: fever-related symptoms and systemic symptoms. It may also include one or more of the following: local symptoms and mental state. The fever-related symptoms include at least one of chills, shivering, and night sweats. The systemic symptoms include at least one of fatigue, muscle aches, dizziness, and loss of appetite. The local symptoms include at least one of cough, sore throat, abdominal pain, joint pain, and rash. The mental state includes at least one of wakefulness, drowsiness, and irritability. The reporting frequency can be set to once every 4-8 hours. Users can supplement and report new symptoms in real time.

[0023] Vital signs data are reported voluntarily by the user or caregiver through the terminal, and include at least one or more of the following: presence or absence of swollen lymph nodes, presence or absence of pharyngeal congestion, and optionally one or more of the following: presence or absence of conjunctival congestion, presence or absence of rash and rash location, and presence or absence of abdominal tenderness. The reporting frequency is once every 8-12 hours, and supplementary reports are made in real time when vital signs change. Any method of obtaining user fever-related vital signs data through terminal interaction is acceptable.

[0024] In S1, the preprocessing process includes: outlier removal, missing data filling, and data standardization; The process of removing abnormal data is as follows: using statistical outlier removal methods, marking logically contradictory data reported by users and reminding users to check and correct it; The missing data filling process specifically involves: using interpolation to fill in missing symptom and vital sign data that have been missing for a long time or that users have not reported on time, and reminding users to supplement them through wearable monitoring devices; The data standardization process specifically involves: unifying the magnitude of vital sign data and quantifying and encoding symptom and vital sign data.

[0025] The preprocessing process in this embodiment specifically involves: preprocessing the collected multi-dimensional data to remove outlier data, fill in missing data, and standardize the data format to prepare for subsequent fusion analysis; specifically including: Abnormal data removal: Statistical outlier removal methods (including but not limited to the 3σ criterion and box plot method) are used to remove outliers from vital sign data. These outliers include, but are not limited to, body temperature >42℃ or <35℃, heart rate <40 beats / minute or >180 beats / minute, and blood oxygen saturation <70%. Logically contradictory data reported by users (such as simultaneously reporting "no rash" and "rash location is on the limbs") are marked and the user is reminded to check and correct them.

[0026] Missing data imputation: For missing vital sign data within a short period of time (≤30 minutes), interpolation methods (including but not limited to linear interpolation, polynomial interpolation, and mean interpolation) are used to imput the data; for missing data for a long period of time (>30 minutes) or for symptoms and vital sign data that the user has not reported on time, the data is marked as "not reported" and the user is reminded to supplement the data through the terminal.

[0027] Data standardization: Convert all vital sign data into a uniform scale (e.g., body temperature to °C, heart rate to beats / minute), and quantify and encode symptom and sign data (encoding rules include, but are not limited to, "none" as 0, "mild" as 1, "moderate" as 2, "severe" as 3, or other equivalent quantification methods) to facilitate subsequent model calculations; any method that can achieve multi-dimensional data standardization and quantification encoding.

[0028] S2. Construct a fusion early warning model. Input the pre-processed multi-dimensional vital sign data, symptom data, and vital sign data into the fusion early warning model for fusion calculation to determine whether the user is in a state of persistent fever and to classify the risk. In S2, the criteria for determining a persistent heating state are: Body temperature ≥38.0℃ and lasting ≥24 hours; body temperature repeatedly ≥38.0℃, with an interval of ≤8 hours between two fevers, and a cumulative number of ≥3 fevers, lasting ≥3 days; body temperature ≥37.5℃ and lasting ≥48 hours accompanied by abnormal vital signs or symptoms; The risk classification includes: low risk, medium risk, and high risk.

[0029] In this embodiment, the construction process of the fusion early warning model is as follows: 1. Prepare training data and annotations (a) Data source: Clinical home fever monitoring data (including vital signs such as body temperature and heart rate, symptoms / signs such as chills and pharyngeal congestion) + baseline vital signs data of healthy individuals were used. The data format was consistent with the output after S1 preprocessing (abnormality removal, missing data filling, standardization / quantization coding). (b) Data labeling: Clinical experts will assign two types of core labels to the data according to the invented criteria for determining persistent fever and risk grading rules: ① binary label (yes / no persistent fever); ② multi-category label (low / medium / high risk). (c) Data partitioning: The labeled data is simply divided into a training set (for model learning, accounting for 70%) and a validation test set (for model testing and optimization, accounting for 30%).

[0030] 2. Core Feature Selection and Organization Based on the multidimensional data preprocessed by S1, high clinical value features are retained, redundant data is removed, and the computational load of the model is reduced by performing only two core operations: (a) Feature classification: Identify 3 types of core input features—core vital signs (body temperature, heart rate, respiratory rate, blood oxygen saturation, with body temperature being the most core feature), quantitative symptoms (chills, fatigue, etc., 0-3 quantitative coding), and quantitative signs (pharyngeal congestion, rash, etc., 0-3 quantitative coding). (b) Feature screening: Based on clinical experience, low-value data that are not related to fever risk are directly removed, and only features that are clearly significant for fever determination and risk classification are retained (such as retaining pulse rate variability and blood pressure, and removing irrelevant secondary data).

[0031] 3. Build a simple model architecture (core) The model adopts a minimalist architecture of "first judgment, then fusion, then classification," containing only three functional modules. The core fusion algorithm uses the clinical weighted summation method (lightweight, easy to deploy, suitable for home scenarios, and can be replaced with a simple machine learning algorithm such as random forest in the cloud as needed), which perfectly meets the judgment and classification requirements of inventions. Module 1: Continuous Heating Determination Module The system incorporates three built-in judgment condition logics to automatically calculate the duration and frequency of abnormal body temperature. If any one condition is met, it is judged as a persistent fever; otherwise, it directly outputs "non-persistent fever, no risk" and terminates subsequent calculations. ① Body temperature ≥38.0℃ and lasting ≥24 hours; ② Body temperature repeatedly ≥38.0℃ (interval ≤8 hours), cumulatively ≥3 times and lasting ≥3 days; ③ Body temperature ≥37.5℃ and lasting ≥48 hours, accompanied by abnormal vital signs / symptoms.

[0032] Module 2: Multi-source data fusion module Based on the premise of "persistent fever", a comprehensive risk score (0-10 points) is calculated by weighted summation of the screened features to provide a basis for risk classification: (1) Set clinical weights: The weights of each feature are determined by experts. Body temperature has the highest weight (e.g., 0.3), followed by heart rate / respiratory rate / blood oxygen saturation (e.g., 0.1 each). Symptoms / signs are allocated according to clinical value (e.g., chills 0.05, confusion 0.08). All weights are summed to 1. (2) Calculate single feature scores: vital signs are scored according to the degree of abnormality (e.g., 38.0-38.9℃ gets 2 points, heart rate > 120 beats / min gets 3 points), and symptoms / signs are scored directly according to the quantitative code (0=0 points, 1=1 point, 2=2 points, 3=3 points). (3) Calculate the comprehensive score: Comprehensive risk score = single feature score × corresponding weight, and sum the scores of all features.

[0033] Module 3: Risk Classification Output Module The comprehensive risk score is directly matched with the invention's classification criteria to output the final risk level, while retaining details of anomalous features: Low risk: 0-3 points (only persistent fever, no other abnormalities); Medium risk: 4-7 points (persistent fever + 1-2 minor abnormalities); High risk: ≥8 points (persistent fever + multiple serious abnormalities).

[0034] 4. Simple Model Training and Parameter Initialization (1) The core of training the model with training set data is to optimize the feature weights so that the error between the model output score / grading result and the result labeled by clinical experts is minimized. (2) For different groups (children / adults / elderly) and different underlying diseases (hypertension / diabetes), separate exclusive feature weights are set (such as increasing the weight of body temperature for children and increasing the weight of blood oxygen saturation for the elderly), and a simple weight library is established. The model can automatically match according to user information.

[0035] 5. Model validation, tuning, and lightweight adaptation (1) Validation and optimization: Use the validation test set to test the model, focusing on the false negative rate of persistent fever and the consistency of risk classification. If the judgment is inaccurate, only the feature weights or scoring thresholds need to be fine-tuned (e.g., low risk can be fine-tuned to 0-4 points), without complicated operations. (2) Lightweight adaptation: The model is simplified according to the deployment scenario to meet the requirements of multi-terminal deployment of inventions: Cloud-based: The weighted summation method can be retained or replaced with a simple machine learning algorithm, supporting multi-user concurrency; Terminal APP: Simplify model calculation steps while retaining complete judgment and hierarchical logic; Wearable devices: Only retain the "core judgment logic + simplified weighted summation", and only output low / high risk coarse classification to adapt to low computing power of devices.

[0036] 6. Reserved parameter iterative optimization interface To meet the model optimization requirements of invention S4, a simplified optimization interface is reserved in the model, eliminating the need for complex algorithms: (1) The model automatically connects to the data storage module to obtain the user's monitoring data, early warning records, and intervention feedback data; (2) Set a fixed optimization period (e.g., monthly / quarterly) and fine-tune the feature weights and scoring thresholds based on the accumulated home fever data; (3) The optimized parameters are synchronized to the model of the terminal / wearable device through the cloud to achieve adaptive updating of the model.

[0037] In this embodiment, a fusion early warning model is established. Preprocessed vital sign data, symptom data, and vital sign data are input into the model for multi-dimensional fusion analysis to determine whether the user is in a state of persistent fever and to classify the fever risk. Specifically, this includes: Determination of persistent fever: Based on body temperature data and the collection time window, determine whether the conditions for persistent fever are met—if any of the following conditions are met, it is determined to be a suspected case of persistent fever: ① Body temperature ≥38.0℃ and duration ≥24 hours; ② Body temperature repeatedly ≥38.0℃ (interval between two fevers ≤8 hours) and the cumulative number of times ≥3 times and the duration ≥3 days; ③ Body temperature ≥37.5℃ and duration ≥48 hours, accompanied by at least one abnormal vital sign or symptom.

[0038] Fusion analysis: Body temperature data is used as the core indicator, combined with vital signs data such as heart rate, respiratory rate, and blood oxygen saturation (e.g., heart rate >100 beats / minute indicates infection risk, respiratory rate >22 breaths / minute indicates worsening condition, and blood oxygen saturation <93% indicates hypoxia risk), as well as symptom and sign data such as chills, shivering, rash, and swollen lymph nodes. A comprehensive risk score is calculated using fusion algorithms (including but not limited to weighted summation, machine learning algorithms, deep learning algorithms, and fuzzy comprehensive evaluation methods). The core of the fusion algorithm is the multi-dimensional data fusion of "body temperature + multiple vital signs + symptoms + signs".

[0039] Risk Classification: Based on the comprehensive risk score, the risk of persistent fever is divided into at least 3 levels, with specific classification criteria as follows (which can be slightly adjusted according to actual needs): Low risk (score 0-3): Only meets the criteria for persistent fever, with no other abnormal vital signs, symptoms and signs, suggesting physiological fever or mild infection, and can be observed at home; Medium risk (score 4-7): If persistent fever is present, accompanied by 1-2 abnormal vital signs (such as slightly elevated heart rate, slightly decreased blood oxygen saturation) or 1-2 mild symptoms / signs (such as mild fatigue, mild pharyngeal congestion), it suggests possible infection and requires enhanced monitoring and home intervention recommendations. High risk (score ≥ 8 points): Meeting the criteria for persistent fever, accompanied by multiple abnormal vital signs (such as heart rate > 120 beats / minute, respiration > 28 breaths / minute, blood oxygen saturation < 90%) or severe symptoms / signs (such as high fever and chills, confusion, widespread rash), indicating a serious condition, requiring immediate medical attention.

[0040] The weighting coefficients of the fusion early warning model are obtained through training with multi-center clinical data. The weighting coefficients can be dynamically adjusted for users of different age groups (children, adults, and the elderly) and different underlying diseases (diabetes, hypertension, and immunodeficiency) to improve the accuracy of early warning. Any fusion early warning model that is trained with multi-center clinical data and whose parameters can be dynamically adjusted.

[0041] S3. Based on the risk classification results, send early warning prompts to users and provide intervention suggestions; In S3, the warning channels for sending warning prompts to users include at least two of the following: wearable monitoring device prompts, terminal APP prompts, and prompts from associated personnel; The intervention recommendations include providing guidance on home observation, physical cooling, or immediate medical attention based on the risk level.

[0042] This embodiment specifically involves: based on the fusion analysis results, sending early warning alerts to users and related personnel (such as family members and community doctors) through at least two or more early warning channels, and providing targeted home intervention guidance based on risk levels; specifically including: Low-risk warning: Send a "Persistent fever, low risk" notification to remind users to monitor their body temperature regularly, drink plenty of water, rest more, and avoid fatigue. If the fever lasts for more than 48 hours, report additional symptoms and signs.

[0043] Medium-risk warning: Send a "medium-risk persistent fever" alert, listing the abnormal data and possible causes in detail, and providing home intervention suggestions (including but not limited to physical cooling methods, dietary suggestions, and symptomatic medication suggestions, please indicate "as directed by your doctor"), reminding you to monitor vital signs every 1-2 hours, and to seek medical attention promptly if symptoms worsen.

[0044] High-risk warning: Immediately send an emergency alert of "persistent high risk of fever" to relevant personnel, clearly stating "immediate medical attention required", and provide at least one emergency assistance function (including but not limited to navigation to nearby medical institutions, emergency contact dialing, and synchronization of monitoring data to medical institutions); any emergency warning and emergency assistance function for high-risk status.

[0045] S4. Store multi-dimensional vital sign data, early warning records, and intervention feedback data, and regularly optimize the parameters of the fusion early warning model.

[0046] This embodiment specifically involves storing users' multi-dimensional monitoring data, fusion analysis results, early warning records, and intervention feedback data on storage media (including but not limited to cloud servers, local caches, and mobile storage devices) to form users' personal health records; periodically iterating and optimizing the parameters of the fusion early warning model (including but not limited to weighting coefficients, judgment thresholds, and grading standards) based on the stored multi-user data to improve the accuracy and adaptability of the early warning model; simultaneously, supporting users and doctors to query historical monitoring data to provide a basis for subsequent diagnosis and treatment; and implementing any method that enables data storage, model iteration optimization, and historical data query.

[0047] Example 2 This embodiment provides a system for implementing a home-based persistent fever early warning method based on wearable vital sign monitoring, including: a wearable monitoring module, a terminal interaction module, a data processing module, a fusion early warning module, an early warning prompt module, and a data storage module; The wearable monitoring module is used to collect multi-dimensional vital sign data of users in real time through wearable monitoring devices; The wearable monitoring module in this embodiment, serving as the core of data acquisition, adopts a lightweight and waterproof design and can be worn on any part of the human body (including but not limited to the wrist, chest, armpit, and ankle), adapting to daily home activities. It includes multiple sensors and data acquisition units. Body temperature sensor: Employs a high-precision sensor (including but not limited to thermistor sensors, thermocouple sensors, and infrared sensors), with a measurement range of 35.0-42.0℃ and an accuracy of ≤0.1℃. It collects the user's surface temperature or core temperature in real time and can be automatically calibrated.

[0048] Heart rate / pulse rate sensor: Employs photoplethysmography (PPG) sensor or other equivalent sensor to collect heart rate and pulse rate variability (HRV) in real time, with anti-motion interference capability, and a measurement range of 40-180 beats / minute.

[0049] Respiratory rate sensor: It acquires data through chest and abdominal motion sensing, indirect acquisition of PPG signals, or other equivalent methods. The measurement range is 8-40 breaths / minute, with an accuracy of ±1 breath / minute.

[0050] Blood oxygen saturation sensor: Employs infrared light detection technology or other equivalent technologies, with a measurement range of 70%-100% and an accuracy of ±2%.

[0051] Optional sensors include one or more of the following: skin humidity sensor and blood pressure sensor. The skin humidity sensor collects the humidity of the user's skin surface to help determine whether there are symptoms such as night sweats; the blood pressure sensor collects the user's blood pressure data to help assess cardiovascular abnormalities associated with fever.

[0052] Data acquisition unit: controls the acquisition frequency of each sensor, performs preliminary filtering on the acquired vital sign data, and then transmits it to the data processing module.

[0053] The terminal interaction module is used to acquire symptom data and vital sign data reported by the user. The terminal interaction module in this embodiment includes at least one terminal device (including but not limited to a mobile APP, a small display screen built into a wearable monitoring device, a tablet, and a smart speaker), serving as a user interaction entry point and realizing the following functions: Data reporting: Provides a symptom and sign reporting interface with standardized options (such as symptom selection and sign grading selection) and supports real-time supplementary reporting by users.

[0054] Data display: Real-time display of user vital signs data, fusion analysis results, risk classification, early warning prompts and intervention suggestions; Settings: Allows users to adjust monitoring frequency, alarm volume, associated personnel information (contact information of family members and doctors), and basic disease information (to facilitate model optimization); Emergency assistance: Set up an emergency dial button or other emergency assistance methods to quickly contact relevant personnel or emergency centers when a high-risk warning is issued; The data processing module is used to preprocess multi-dimensional vital sign data, symptom data, and physical sign data. In this embodiment, the data processing module is located at least one of the wearable monitoring module, the terminal device, and the cloud server, and works together to complete data preprocessing. Local preprocessing: The data processing unit in the wearable monitoring module or terminal device performs preliminary anomaly removal on the collected vital sign data to reduce the amount of data transmission; Cloud-based preprocessing: The data processing unit in the cloud server performs complete anomaly removal, missing data filling, and data standardization on all received data (vital signs, symptoms, and physical signs), generates a standardized data matrix, and transmits it to the fusion early warning module.

[0055] The fusion early warning module is used to construct a fusion early warning model. It inputs preprocessed multi-dimensional vital sign data, symptom data, and vital sign data into the fusion early warning model for fusion calculation to determine whether the user is in a state of persistent fever and to classify the risk. In this embodiment, the fusion early warning module constructs a fusion early warning model based on an algorithm model (including but not limited to machine learning algorithms, deep learning algorithms, and weighted summation algorithms), and embeds it in at least one of the cloud server and terminal devices. Its core functions include: Data fusion: Receives standardized multi-dimensional data and calculates a comprehensive risk score through a fusion algorithm; any algorithm that can achieve multi-dimensional data fusion of "body temperature + multiple vital signs + symptoms + signs" falls within the protection scope of this module; Persistent fever determination: Based on body temperature data and time window, determine whether it is a suspected persistent fever state; any persistent fever determination function based on the determination logic of this invention falls within the protection scope of this module; Risk classification: The fever risk level is determined based on the comprehensive risk score.

[0056] Model iteration: Based on stored historical data, the model parameters are optimized periodically using optimization algorithms (including but not limited to gradient descent, genetic algorithm, and random forest algorithm) to adapt to the early warning needs of different groups.

[0057] The early warning module is used to send early warnings to users and provide intervention suggestions based on the risk classification results. The early warning module in this embodiment includes a multi-channel early warning unit to ensure that early warning information is delivered in a timely manner, and includes at least two or more early warning channels: Wearable device alerts: Warning alerts are sent through one or more of the following methods: vibration, light, and voice (small speaker). Different alert methods are used for low risk, medium risk, and high risk (e.g., green light + short vibration for low risk, yellow light + medium vibration for medium risk, and red light + long vibration + voice alert for high risk). Terminal App Prompt: Send warning prompts and intervention suggestions through one or more of the following methods: pop-up window, push message, and ringtone; Relevant personnel reminder: When a high-risk warning is issued, the warning information will be sent to relevant personnel simultaneously via one or more of the following methods: SMS, APP push, and telephone, informing users of the current status and the measures to be taken; The data storage module is used to store multi-dimensional vital sign data, early warning records, and intervention feedback data, and to periodically optimize the parameters of the fusion early warning model.

[0058] In this embodiment, the data storage module employs at least one storage method (including but not limited to cloud server storage, local cache storage, and mobile storage device storage), preferably a dual storage method of cloud server + local cache: Local caching: The wearable monitoring module or terminal device has a built-in storage unit that caches vital sign data from the most recent 24-72 hours to prevent data loss during network interruptions; Cloud storage: The cloud server stores all user monitoring data, integrated analysis results, early warning records, and intervention feedback data, forming a traceable personal health record. It supports data encryption protection to ensure user privacy and security, while also providing data support for model iteration.

[0059] The wearable monitoring module adopts a lightweight and waterproof design; The wearable monitoring module includes: a body temperature sensor, a heart rate / pulse rate sensor, a respiratory rate sensor, and a blood oxygen saturation sensor.

[0060] The terminal interaction module includes: a mobile APP, a wearable device display screen, a tablet, and a smart speaker; The terminal interaction module is configured to provide a data reporting interface and a data display interface, and to implement an emergency assistance function; The data processing module and the fusion early warning module are deployed in at least one of the cloud server, terminal interaction module, or wearable monitoring module; The warning module sends warnings to users in the following ways: vibration, light, voice, pop-up window, SMS or telephone.

[0061] Compared with the prior art, this embodiment also has the following significant advantages: 1. Overcoming the limitations of single body temperature monitoring and achieving multi-dimensional integrated early warning: Unlike existing single body temperature monitoring or single cause early warning technologies, this invention combines multiple vital signs such as body temperature, heart rate, respiratory rate, and blood oxygen saturation, as well as the symptoms and signs accompanying fever, for comprehensive analysis. This can effectively distinguish between physiological fever and pathological persistent fever, reduce the probability of false and missed early warnings, and improve the accuracy of early warning. 2. Adapted to home scenarios, enhancing convenience and practicality: The wearable device is lightweight and waterproof, suitable for daily home activities, eliminating the need for frequent manual measurements; the terminal interaction module is easy to operate, supports self-reporting of symptoms and signs, and is suitable for various groups (children, adults, and the elderly); multi-channel early warning prompts ensure that users and related personnel receive early warning information in a timely manner. 3. Precise risk grading and targeted intervention guidance: Based on multi-dimensional data fusion to calculate a comprehensive risk score, fever risk is divided into three levels: low, medium and high, corresponding to different warning prompts and home intervention suggestions. This avoids excessive tension in low-risk situations and prevents delays in medical treatment in high-risk situations, while providing users with scientific home management guidance. 4. The model can be iteratively optimized to adapt to different groups of people: Through multi-user data accumulated in the cloud, the weighting coefficients of the fusion early warning model are regularly optimized and dynamically adjusted for users of different age groups and different underlying diseases to improve the model's adaptability and accuracy; at the same time, the stored historical data can provide a reference for subsequent diagnosis and treatment, realizing closed-loop management of "monitoring-early warning-intervention-traceability".

[0062] Example 2 This invention provides a specific implementation of the home-based persistent fever risk early warning method based on wearable vital sign monitoring in Example 1: This embodiment uses an adult home user (without underlying diseases) as an example to illustrate the specific implementation process of the early warning method: S1. Multi-dimensional data collection: Users wear a wearable wrist device that collects body temperature (every 10 minutes), heart rate, respiratory rate, blood oxygen saturation, HRV, and skin moisture in real time; users report symptoms (such as "chills, mild fatigue") and signs (such as "mild pharyngeal congestion") every 6 hours through a mobile APP. S2. Data Preprocessing: The body temperature data collected by the device is 38.2℃ (normal range), heart rate is 95 beats / minute (normal range), respiratory rate is 20 breaths / minute (normal range), and blood oxygen saturation is 96% (normal range), with no abnormal data; the symptoms and signs data reported by the user have no logical contradictions, and after standardization, they are coded as "chills=1, fatigue=1, pharyngeal congestion=1"; S3. Multi-dimensional data fusion analysis: After 24 hours of continuous monitoring, the user's body temperature remained between 38.1-38.3℃, meeting the criteria for persistent fever. After fusion analysis, the comprehensive risk score was 4 points (2 points for core body temperature score + 2 points for symptom and sign score), and the user was classified as medium risk. S4. Early Warning and Intervention Guidance: The wearable device emits a yellow light and vibrates moderately, and the mobile APP pushes a "medium risk of persistent fever" prompt, suggesting "physical cooling (wiping the forehead and neck with warm water), drinking more water, eating a light diet, monitoring body temperature every 2 hours, and seeking medical attention promptly if high fever and chills occur"; no prompts were sent to related personnel. S5. Data storage and iterative optimization: The cloud stores users' 24-hour monitoring data, risk scores, and early warning records; subsequently, by combining multi-user data, the model weighting coefficients are optimized to improve the accuracy of medium-risk assessment.

[0063] It should be noted that this embodiment is only one specific implementation method. Any implementation method based on the technical solution of this invention, such as replacing the sensor type (e.g., replacing the thermistor sensor with an infrared sensor), adjusting the acquisition frequency (e.g., adjusting the conventional acquisition frequency to once every 8 minutes), modifying the fusion algorithm (e.g., replacing the weighted summation method with the random forest algorithm), or fine-tuning the risk classification scoring range (e.g., adjusting low risk to 0-4 points), falls within the protection scope of this invention.

[0064] The early warning device used in this embodiment includes a wearable wrist device, a mobile APP, and a cloud server, and its specific structure is as follows: Wearable wrist device: Weight ≤20g, waterproof rating IP67, built-in body temperature sensor (accuracy 0.1℃), PPG heart rate / HRV sensor, respiratory rate sensor, blood oxygen saturation sensor, skin humidity sensor, built-in small display screen (can display real-time body temperature and heart rate), supports Bluetooth 5.0 communication, battery life ≥7 days, with vibration and light warning functions.

[0065] Mobile App: Supports iOS and Android systems. The interface includes four modules: "Real-time Monitoring", "Symptom Reporting", "Early Warning Center" and "Personal Center". It allows users to set associated personnel (family members' mobile phone numbers), basic medical information, and supports emergency dialing function.

[0066] Cloud server: It adopts Alibaba Cloud server, which has the functions of encrypted data storage, data preprocessing, fusion early warning model operation, and data iterative optimization. It can support 100,000+ concurrent users for monitoring at the same time, with a response time of ≤1 second.

[0067] When the device is running, the wearable wrist device collects vital sign data in real time and transmits it to a mobile APP via Bluetooth. Users can report symptoms and vital sign data through the APP. The APP uploads all data to the cloud server. After data preprocessing and fusion analysis, risk classification and early warning prompts are generated and pushed to the wearable device and mobile APP simultaneously, so as to realize accurate early warning of persistent fever at home.

[0068] It should be noted that the device in this embodiment is only one specific structure. Any device structure based on the technical solution of this invention that adjusts the wearing position of the wearable device (such as changing it to chest wearing), replaces the communication method (such as replacing Bluetooth with NB-IoT), changes the cloud server type (such as replacing Alibaba Cloud with Tencent Cloud), or adds or replaces sensors (such as adding a blood pressure sensor) falls within the protection scope of this invention.

[0069] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for early warning of persistent fever at home based on wearable vital sign monitoring, characterized in that, Includes the following steps: S1. Collect multi-dimensional vital sign data of users in real time through wearable monitoring devices, obtain symptom data and vital sign data reported by users, and perform preprocessing. S2. Construct a fusion early warning model. Input the pre-processed multi-dimensional vital sign data, symptom data, and vital sign data into the fusion early warning model to identify whether the user is in a state of persistent fever and to classify the risk. S3. Based on the risk classification results, send early warning prompts to users and provide intervention suggestions; S4. Store multi-dimensional vital sign data, early warning data, and intervention suggestion data, and use optimization algorithms to optimize the parameters of the fusion early warning model.

2. The home-based persistent fever early warning method based on wearable vital sign monitoring according to claim 1, characterized in that, In S1, the multi-dimensional vital signs data include: body temperature, heart rate, respiratory rate, blood oxygen saturation, skin moisture, pulse rate variability, and blood pressure. The symptom data includes one or more of the following: fever-associated symptoms, systemic symptoms, local symptoms, and mental status; The vital signs data include one or more of the following: lymph node enlargement, pharyngeal congestion, presence or absence of conjunctival congestion, presence or absence of skin rash and its location, and presence or absence of abdominal tenderness.

3. The home-based persistent fever early warning method based on wearable vital sign monitoring according to claim 1, characterized in that, In S1, the preprocessing process includes: outlier removal, missing data filling, and data standardization; The process of removing abnormal data is as follows: using statistical outlier removal methods, marking logically contradictory data reported by users and reminding users to check and correct it; The missing data filling process specifically involves: using interpolation to fill in missing data; and using wearable monitoring devices to remind users to supplement symptom and sign data that have been missing for a long time or that users have not reported on time. The data standardization process specifically involves: unifying the magnitude of vital sign data and quantifying and encoding symptom and vital sign data.

4. The home-based persistent fever early warning method based on wearable vital sign monitoring according to claim 1, characterized in that, In S2, the criteria for determining a persistent heating state are: Body temperature ≥38.0℃ and lasting ≥24 hours; body temperature repeatedly ≥38.0℃, with an interval of ≤8 hours between two fevers, and a cumulative number of ≥3 fevers, lasting ≥3 days; Body temperature ≥37.5℃ for ≥48 hours and accompanied by abnormal vital signs or symptoms; The risk classification includes: low risk, medium risk, and high risk.

5. The home-based persistent fever early warning method based on wearable vital sign monitoring according to claim 1, characterized in that, In S3, the channels for sending early warning prompts to users include at least two of the following: prompts from wearable monitoring devices, prompts from terminal apps, and prompts from associated personnel. The intervention recommendations include providing guidance on home observation, physical cooling, or immediate medical attention based on the risk analysis.

6. A home-based persistent fever early warning system based on wearable vital sign monitoring, used to implement the home-based persistent fever early warning method based on wearable vital sign monitoring as described in any one of claims 1 to 5, characterized in that, include: Wearable monitoring module, terminal interaction module, data processing module, fusion early warning module, early warning prompt module, and data storage module; The wearable monitoring module is used to collect multi-dimensional vital sign data of users in real time through wearable monitoring devices; The terminal interaction module is used to acquire symptom data and vital sign data reported by the user. The data processing module is used to preprocess multi-dimensional vital sign data, symptom data, and physical sign data; The fusion early warning module is used to construct a fusion early warning model and input preprocessed multi-dimensional vital sign data, symptom data and vital sign data into the fusion early warning model to identify whether the user is in a state of persistent fever and to classify the risk. The early warning module is used to send early warnings to users and provide intervention suggestions based on the risk classification results. The data storage module is used to store multi-dimensional vital sign data, early warning data, and intervention suggestion data, and to optimize the parameters of the fusion early warning model.

7. The home-based persistent fever early warning system based on wearable vital sign monitoring according to claim 6, characterized in that, The wearable monitoring module adopts a lightweight and waterproof design; The wearable monitoring module includes: a body temperature sensor, a heart rate / pulse rate sensor, a respiratory rate sensor, and a blood oxygen saturation sensor.

8. The home-based persistent fever early warning system based on wearable vital sign monitoring according to claim 6, characterized in that, The terminal interaction module includes: a mobile APP, a wearable device display screen, a tablet, and a smart speaker; The terminal interaction module is configured to provide a data reporting interface and a data display interface, and to implement an emergency assistance function.

9. The home-based persistent fever early warning system based on wearable vital sign monitoring according to claim 6, characterized in that, The data processing module and the fusion early warning module are deployed in at least one of the cloud server, terminal interaction module, or wearable monitoring module; The warning module sends warnings to users in the following ways: vibration, light, voice, pop-up window, SMS or telephone.

Images