Multi-state recognition-based cross-state dynamic threshold active review health early warning method
By using multi-state recognition and dynamic threshold adjustment of smart devices, the problems of false alarms and missed alarms of smart wearable devices under different physiological states have been solved, and precise health management and resource optimization have been achieved.
Patent Information
- Application Number
- CN202511778115.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-11-28
- Publication Date
- 2026-03-03
AI Technical Summary
Existing smart wearable devices use static warning thresholds under different physiological conditions, leading to false alarms and missed alarms. They cannot adapt to the differences in physiological indicators between day and night, resulting in poor user experience and waste of medical resources.
A multi-state recognition and cross-state dynamic threshold active review health warning method is adopted. Multimodal physiological data is collected in real time through intelligent devices to identify the user's physiological state and dynamically adjust the warning threshold according to the state. The method is combined with a hierarchical response mechanism for review and confirmation.
It improved the accuracy of health monitoring, reduced the false alarm rate and the risk of missed reports, and achieved precise health management and resource optimization.
Smart Images
Figure CN121601266A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of health monitoring technology, and in particular to a method for proactive review and early warning of health based on multi-state identification and cross-state dynamic threshold. Background Technology
[0002] In existing technologies, health monitoring functions of smart wearable devices (such as smartwatches and wristbands) generally use static warning thresholds. For example, a fixed high threshold (e.g., 120 beats / minute) is set for heart rate, and the same threshold is used for warnings regardless of whether the user is active during the day or asleep. This approach has significant drawbacks: during daytime activities, data noise is introduced due to factors such as movement and loosening of the device, making it easy to trigger false alarms using static thresholds, causing users to ignore the alerts; during sleep at night, physiological indicators (such as heart rate, blood oxygen, blood pressure, and respiratory rate) decrease significantly, and using daytime thresholds cannot identify hidden risks (such as sleep tachycardia, sleep bradycardia, sleep hypoxemia, and sleep hypertension or hypoxemia), resulting in missed warnings. Furthermore, existing technologies lack intelligent hierarchical confirmation mechanisms. When an anomaly is detected, they either do not trigger an alarm or directly trigger the highest level alarm. Currently, none of the mainstream smartwatches have been able to set different thresholds for different physiological states (especially night and day). The human body's physiological baseline at night is significantly different from that during the day, requiring a separate threshold. The daytime threshold cannot be used, resulting in a poor user experience, a high risk of false alarms, and a waste of medical resources.
[0003] Therefore, it is necessary to provide a method for proactive review and early warning of health conditions based on multi-state identification and cross-state dynamic thresholds to overcome the above-mentioned shortcomings. Summary of the Invention
[0004] The purpose of this invention is to provide a health early warning method based on multi-state identification and cross-state dynamic threshold active review, which aims to solve the problems of how to improve the accuracy of health monitoring, reduce false alarm rate and missed alarm risk, and provide intelligent hierarchical response capability to achieve accurate and reliable health management.
[0005] To achieve the above objectives, the present invention provides a method for proactive review and early warning of health conditions based on multi-state identification and cross-state dynamic thresholds, comprising: Step S10: Collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data; Step S20: Based on the multimodal physiological data, use the pre-trained classification model built into the smart device to identify the user's current physiological state, which includes at least daytime activity state and sleep state; Step S30: Based on the identified current physiological state, retrieve the set of warning thresholds corresponding to the current physiological state from the preset dynamic threshold library; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states; Step S40: When the real-time monitored user physiological parameters exceed the current warning threshold, a tiered response process is initiated; wherein, the tiered response process includes: Initial warning stage: A gentle reminder is issued and the user is prompted to rest. At the same time, a rest mode of a preset duration is initiated to conduct key measurements to obtain and verify physiological data. During the sleep state, the sampling time is extended and silent verification is performed. Verification and judgment stage: Based on the verified physiological data, determine whether an abnormality is confirmed; Tiered response phase: If the review and judgment confirm an anomaly, the corresponding response action is triggered according to the anomaly level, which includes low risk, medium risk and high risk.
[0006] In a preferred embodiment, the multimodal physiological data further includes time information, which is used to identify the current physiological state in conjunction with a preset user habit schedule; wherein the user habit schedule includes typical sleep periods learned through historical data.
[0007] In a preferred embodiment, the dynamic threshold library is dynamically adjusted based on the user's personal baseline data through a self-learning mechanism, specifically including: For each physiological indicator, the long-term rolling mean μ and standard deviation σ are calculated under each physiological state to obtain the individual baseline; Based on the personal baseline, a dynamic threshold is set for each physiological state using a threshold adjustment formula. The target threshold is equal to the rolling average μ of the corresponding physiological indicator plus the product of the adjustment coefficient k and the standard deviation σ of the corresponding physiological indicator. The adjustment coefficient k for each physiological indicator is different in the daytime activity state and the sleep state. The individual baseline and thresholds are recalculated periodically or irregularly to accommodate changes in the user's physiological state.
[0008] In a preferred embodiment, the initial warning stage includes emitting equipment vibration and displaying prompt information; in the resting mode, the duration of the key measurement is 30s-60s, and during this period, a higher sampling frequency and stricter signal quality judgment criteria are used compared to conventional acquisition.
[0009] In a preferred embodiment, the graded response phase includes: Low-risk abnormalities: Record events and generate health reports, and regularly remind users to pay attention; Medium-risk anomaly: Triggers a local alarm on the device and displays brief instructions; notifies family and friends when necessary conditions are met. High-risk anomaly: Automatically sends a help text message / call containing location information to preset emergency contacts, and / or simultaneously triggers a linkage with nearby emergency centers.
[0010] In a preferred embodiment, it further includes: Step 50: Based on physiological data collected over a long period of time under different physiological states, calculate cross-state trend characteristics, and use machine learning models to generate early profiles of user health risks, and issue early warnings when the risk index exceeds a preset risk threshold.
[0011] In a preferred embodiment, the calculation steps of the cross-state trend characteristics include: Collect users' historical physiological data under different physiological states, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data; Calculate cross-state indicators and their rate of change of trends, and use a machine learning model to generate a risk index based on the calculation results; When the risk index exceeds the preset risk threshold, an early risk report is generated and the user is notified, and an early profile of the user's health risk and early warning suggestions are output.
[0012] In a preferred embodiment, during the review and judgment stage, the preset monitoring time is automatically extended, and confirmation is made in conjunction with the signal quality index. An anomaly is only confirmed when the calculated signal quality index exceeds a preset threshold.
[0013] The present invention also provides a health early warning device based on multi-state recognition and cross-state dynamic threshold active review, comprising: The physiological data acquisition module is used to collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data. The physiological state recognition module is used to identify the user's current physiological state based on the multimodal physiological data and using a pre-trained classification model built into the smart device. The physiological state includes at least daytime activity state and sleep state. The warning threshold retrieval module is used to retrieve a set of warning thresholds corresponding to the current physiological state from a preset dynamic threshold library based on the identified current physiological state; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states; The tiered response initiation module is used to initiate a tiered response process when the user's physiological parameters monitored in real time exceed the currently invoked warning threshold; the tiered response process includes: Initial warning stage: A gentle reminder is issued and the user is prompted to rest. At the same time, a rest mode of a preset duration is initiated to conduct key measurements to obtain and verify physiological data. During the sleep state, the sampling time is extended and silent verification is performed. Verification and judgment stage: Based on the verified physiological data, determine whether an abnormality is confirmed; Tiered response phase: If the review and judgment confirm an anomaly, the corresponding response action is triggered according to the anomaly level, which includes low risk, medium risk and high risk.
[0014] The health early warning method based on multi-state recognition and cross-state dynamic threshold active review provided by this invention adaptively adjusts the monitoring threshold to match the physiological characteristics of different states through multimodal data fusion and state recognition. The physiological state-dependent dynamic threshold fundamentally reduces false alarms and missed alarms, making the monitoring data more clinically valuable. The multi-level response mechanism performs secondary confirmation, filters false alarms and ensures timely response to real risks. The multi-level response ensures rapid response under real danger, while active review reduces the negative impact of false alarms. Attached Figure Description
[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of the present invention and should not be regarded as a limitation on the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0016] Figure 1 The flowchart of the active review and health warning method based on multi-state identification and cross-state dynamic threshold provided by the present invention; Figure 2 This is a schematic diagram of a dynamic threshold library using heart rate as an example in an exemplary embodiment. Figure 3 A schematic diagram illustrating the early warning monitoring of heart rate by intelligent devices under different physiological states; Figure 4 A schematic diagram illustrating the early warning monitoring of blood pressure by a smart device under different physiological states; Figure 5 This is a schematic diagram illustrating the dynamic threshold of blood oxygen during daytime activity levels as determined by a smart device. Figure 6 This is a schematic diagram illustrating the dynamic threshold of blood oxygen during sleep using a smart device. Figure 7This is a framework diagram of the health early warning device based on multi-state recognition and cross-state dynamic threshold active review provided by the present invention. Detailed Implementation
[0017] To make the objectives, technical solutions, and beneficial effects of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described in this specification are merely for explaining the invention and are not intended to limit the invention.
[0018] It should also be understood that the terminology used in this specification is for the purpose of describing particular embodiments only and is not intended to limit the invention. As used in this specification and the appended claims, the singular forms “a,” “an,” and “the” are intended to include the plural forms unless the context clearly indicates otherwise.
[0019] It should also be further understood that the term "and / or" as used in this specification and the appended claims refers to any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.
[0020] Example 1 In an embodiment of the present invention, a method for proactive review of health warning based on multi-state identification and cross-state dynamic threshold is provided. By fusion of multimodal data and state identification, the monitoring threshold is adaptively adjusted to match the physiological characteristics of different states, so as to more accurately identify the user's health risks, reduce false alarms and false negatives, and make the monitoring data more clinically valuable.
[0021] like Figure 1 As shown, the method for proactive review of health warning based on multi-state identification and cross-state dynamic threshold includes steps S10-S50.
[0022] Step S10: Collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data.
[0023] The smart device incorporates a sensor array (including an accelerometer, gyroscope, ambient light sensor, heart rate sensor, blood oxygen sensor, and temperature sensor) and a state recognition algorithm module (i.e., a classification model). The user continuously wears the smart device with its health monitoring function enabled. Furthermore, the multimodal physiological data also includes time information. This time information is used in conjunction with a pre-set user schedule to identify the current physiological state.
[0024] In this embodiment, multimodal data is acquired in real time or periodically, including: Motion data: The body motion frequency (unit: Hz) and amplitude (unit: m / s²) are calculated using accelerometers and gyroscopes, and then motion features such as the standard deviation of body motion energy and peak frequency are extracted; Ambient light data: Illumination intensity (unit: lux) is acquired through an ambient light sensor, and then light features, such as the moving average of ambient light intensity, are extracted. Physiological data: Heart rate (unit: beats / minute) and respiratory rate (unit: breaths / minute) are acquired through heart rate and respiratory rate sensors, and then physiological characteristics such as heart rate variability (HRV), respiratory rate stability, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data are extracted; Time information: Combine the device system time with the user's habitual schedule (such as typical sleep periods learned through historical data), and then extract time features, such as the degree of matching between the current time and the user's habitual schedule.
[0025] Step S20: Based on multimodal physiological data, use the pre-trained classification model built into the smart device to identify the user's current physiological state, which includes at least daytime activity state and sleep state.
[0026] State recognition is performed using a pre-trained classification model (such as a decision tree, support vector machine, or lightweight neural network). The model input is the aforementioned feature vector, and the output is a physiological state label, including at least "daytime activity state" and "sleep state," which can be further subdivided into sub-states such as "resting," "mild activity," "moderate activity," "vigorous activity," "light sleep," and "deep sleep." It should be noted that, in addition to parsing sensor data, mobile phone usage status (such as screen unlock frequency) or connected smart home devices (such as lighting status) can be incorporated as auxiliary judgment criteria. Furthermore, the classification model can be fine-tuned online based on user feedback or historical data to improve recognition accuracy. Moreover, the recognition principles and specific implementation steps for user physiological states (mainly distinguishing between activity and sleep states; the human body may sleep during the day and be active at night) can refer to existing technologies, and this invention does not limit them.
[0027] Step S30: Based on the identified current physiological state, retrieve the set of warning thresholds corresponding to the current physiological state from the preset dynamic threshold library; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states.
[0028] Specifically, monitoring parameters are adjusted based on the identified physiological state: in the "sleep state", a higher sampling frequency (e.g., once per second) and a more refined signal filtering algorithm (e.g., low-pass filtering to remove motion noise) are used; in the "daytime activity state", an anti-motion interference algorithm (e.g., adaptive filtering or waveform morphology analysis) is used.
[0029] This step involves retrieving the warning threshold bound to the current state from the dynamic threshold library. Figures 3-6 As shown, the dynamic threshold library sets independent thresholds for the same physiological indicator under different states. For example, the high-risk heart rate threshold during sleep is 90 beats / minute; the high-risk heart rate threshold during daytime activity is 150 beats / minute. The initial values of this dynamic threshold library can be based on publicly available clinical guidelines, or personalized initial value suggestions can be provided to users of different ages, genders, and underlying diseases through cloud-based big data models. It is understandable that, combined with... Figure 2 As shown in the figure, the "high-risk" threshold (red) for the "heart rate" indicator dynamically changes under different states. For example, in "sleep state," a heart rate exceeding 90 beats / minute may trigger a high-risk warning; while in "vigorous activity" state, the threshold relaxes to 190 beats / minute; the "medium-risk" (yellow) and "low-risk" (green) thresholds also follow the same dynamic principle, fully embodying the core improvement idea of "state adaptation" in this invention. Therefore, it fundamentally reduces false alarms and false negatives, making the monitoring data more clinically valuable.
[0030] Furthermore, the dynamic threshold library is dynamically adjusted based on the user's personal baseline data through a self-learning mechanism. Specifically, this includes: for each physiological indicator (such as heart rate), calculating the long-term rolling average (μ) and standard deviation (σ) under each physiological state (such as daytime activity and nighttime sleep) to obtain the personal baseline; and setting dynamic thresholds for each physiological state based on the personal baseline using a threshold adjustment formula; for example, high-risk threshold = μ + k σ, where k is an adjustment coefficient, dynamically set according to the user's state (e.g., k=2 for sleep state, k=3 for daytime activity state); the individual baseline and threshold are recalculated periodically or irregularly (e.g., every 7 days) to adapt to changes in the user's physiological state. Therefore, by dynamically adjusting the threshold through a self-learning mechanism, personalized monitoring can be achieved to adapt to different users' physiological changes.
[0031] Step S40: When the real-time monitored user physiological parameters exceed the current warning threshold, the graded response process is initiated; wherein, the graded response process includes: the initial warning stage, the review and judgment stage, and the graded response stage.
[0032] Initial warning phase: A gentle reminder is issued, prompting the user to rest. Simultaneously, a pre-set rest mode is activated for focused measurements to obtain and verify physiological data. Specifically, the initial warning phase includes device vibration and displaying a prompt message (e.g., "Possible abnormality detected; please sit quietly for a moment"). In rest mode, the focused measurement duration is 30-60 seconds, during which a higher sampling frequency (e.g., twice per second) and stricter signal quality standards (e.g., signal-to-noise ratio >10dB) are used compared to regular data collection. During sleep, the sampling time is extended and silent verification is performed to avoid disrupting normal sleep and preventing false alarms.
[0033] Verification and Judgment Phase: Based on the verified physiological data, a determination is made as to whether an abnormality is confirmed. For example, if the verified data returns to the normal range (e.g., heart rate below the threshold), it is recorded as a "suspected false alarm" and no external alarm is triggered; if the verified data confirms an abnormality, a graded response is triggered according to the abnormality level (low, medium, high). Abnormality levels include low risk, medium risk, and high risk.
[0034] In an exemplary embodiment, the exception level classification rule can be as follows: Low-risk abnormality: Reviewed physiological data exceed the threshold but the magnitude is <10%, and the duration is <1 minute; Intermediate-risk abnormality: physiological data exceeding the threshold by 10%-30%, or lasting for 1-5 minutes; High-risk abnormality: physiological data exceeding the threshold by more than 30%, or lasting for more than 5 minutes, or no response from the user.
[0035] It should be noted that the above data range is for illustrative purposes only and does not represent actual application.
[0036] Furthermore, during the review and judgment phase, the preset monitoring time is automatically extended (to 60 seconds), and a silent review is performed using the Signal Quality Index (SQI) (especially in sleep mode). The SQI calculation formula is: SQI = (Number of valid signal segments / Total number of signal segments) × 100%. An anomaly is only confirmed when the calculated SQI exceeds a preset threshold (SQI > 90%). Therefore, detailed alerts and a silent mode can ensure the reliability of the review data and reduce false alarms.
[0037] Tiered response phase: If the review and judgment confirm an anomaly, the corresponding response action is triggered according to the anomaly level.
[0038] Specifically, the tiered response phase includes: Low-risk abnormalities: Record events and generate health reports, and regularly remind users to pay attention; Medium-risk anomaly: Triggers local device alarms (such as continuous vibration and sound) and displays brief instructions (such as "Please keep quiet"), and notifies family and friends when necessary conditions are met; High-risk anomaly: Automatically sends a help text message / call containing location information to preset emergency contacts, and / or simultaneously triggers a linkage with nearby emergency centers.
[0039] The response process terminates upon completion of the action or manual cancellation by the user. Therefore, this tiered response approach enables differentiated risk management, optimizing resource allocation and user experience.
[0040] Furthermore, in some embodiments, the method further includes step 50.
[0041] Step 50: Based on long-term collected physiological data under different physiological states, calculate cross-state trend characteristics and use a machine learning model to generate an early profile of user health risks. When the risk index exceeds a preset risk threshold, an early warning is issued. Therefore, cross-state trend analysis can provide early risk insights and enable preventative health management.
[0042] Specifically, the calculation steps for cross-state trend characteristics include steps S51-S53.
[0043] Step S51: Collect historical physiological data of the user under different physiological states, including daytime resting heart rate, nighttime average heart rate, sleep heart rate variability, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data. Step S52: Calculate cross-state indicators and their trend change rates, and generate a risk index based on the calculation results using a machine learning model.
[0044] For example, cross-state indicators such as diurnal heart rate variation (ΔHR) = daytime resting heart rate - nighttime average heart rate; then calculate the rate of change of trend, such as the sliding window average rate of change of ΔHR.
[0045] In addition, machine learning models (such as linear regression or time series models) are used to generate risk indices corresponding to various physiological indicators. For example, the heart failure risk index = α. ΔHRt+β HRVt+γ, where ΔHRt is the negative trend slope of ΔHR (unit: beats / minute / month), HRVt is the negative trend slope of heart rate variability (unit: ms / month), and α, β, and γ are model coefficients that can be obtained by fitting training data.
[0046] Step S53: When the risk index exceeds the preset risk threshold, generate an early risk report and notify the user, and output the user's early health risk profile and warning suggestions.
[0047] Therefore, this step, through the signal quality index, further ensures the validity of the verification data and enhances the robustness of the system.
[0048] In summary, the health early warning method based on multi-state recognition and cross-state dynamic threshold proactive review provided by this invention adaptively adjusts the monitoring threshold to match the physiological characteristics of different states through multimodal data fusion and state recognition. Moreover, the physiological state-dependent dynamic threshold fundamentally reduces false alarms and missed alarms, making the monitoring data more clinically valuable. Secondary confirmation is performed through a multi-level response mechanism to filter false alarms and ensure timely response to real risks. Multi-level response ensures rapid response under real danger, while proactive review reduces the negative impact of false alarms.
[0049] Example 2 This invention also provides a health early warning device 100 based on multi-state recognition and cross-state dynamic threshold active review. This device uses multi-modal data fusion and state recognition to adaptively adjust monitoring thresholds to match the physiological characteristics of different states, thereby more accurately identifying user health risks, reducing false alarms and missed alarms, and making monitoring data more clinically valuable. It should be noted that the implementation principle and specific implementation method of the health early warning device 100 based on multi-state recognition and cross-state dynamic threshold active review can be found in existing technologies, and will not be elaborated upon here.
[0050] like Figure 7 As shown, the health early warning device 100 based on multi-state recognition, cross-state dynamic threshold active review includes: The physiological data acquisition module 10 is used to collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data. The physiological state recognition module 20 is used to identify the user's current physiological state based on multimodal physiological data and using a pre-trained classification model built into the smart device. The physiological state includes at least daytime activity state and sleep state. The warning threshold calling module 30 is used to call the set of warning thresholds corresponding to the current physiological state from the preset dynamic threshold library according to the identified current physiological state; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states. The tiered response initiation module 40 is used to initiate a tiered response process when the real-time monitored user physiological parameters exceed the currently invoked warning threshold; wherein the tiered response process includes: Primary warning stage: A gentle reminder is issued and the user is prompted to rest. At the same time, a rest mode of a preset duration is started to conduct key measurements to obtain and verify physiological data. During the sleep state, the sampling time is extended and a silent verification is performed. Review and judgment stage: Based on the reviewed physiological data, determine whether an abnormality has been confirmed; Tiered response phase: If the review confirms an anomaly, the corresponding response action is triggered according to the anomaly level, which includes low risk, medium risk and high risk.
[0051] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is merely an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiments can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit. Furthermore, the specific names of the functional units and modules are only for easy differentiation and are not intended to limit the scope of protection of this application. The specific working process of the units and modules in the above system can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0052] In the above embodiments, the descriptions of each embodiment have different focuses. For parts that are not described in detail or recorded in a certain embodiment, please refer to the relevant descriptions of other embodiments.
[0053] Those skilled in the art will recognize that the units and method steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.
[0054] In the embodiments provided by this invention, it should be understood that the disclosed systems, devices / terminal equipment, and methods can be implemented in other ways. For example, the system or device / terminal equipment embodiments described above are merely illustrative. For instance, the division of modules or units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the mutual coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection of systems or units may be electrical, mechanical, or other forms.
[0055] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0056] The present invention is not limited to the description in the specification and embodiments, and thus other advantages and modifications can be readily realized by those skilled in the art. Therefore, the present invention is not limited to the specific details, representative devices and illustrated examples shown and described herein without departing from the spirit and scope of the general concept as defined by the claims and their equivalents.
Claims
1. A method for proactive review and early warning of health conditions based on multi-state identification and cross-state dynamic threshold, characterized in that, include: Step S10: Collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data; Step S20: Based on the multimodal physiological data, use the pre-trained classification model built into the smart device to identify the user's current physiological state, which includes at least daytime activity state and sleep state; Step S30: Based on the identified current physiological state, retrieve the set of warning thresholds corresponding to the current physiological state from the preset dynamic threshold library; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states; Step S40: When the real-time monitored user physiological parameters exceed the current warning threshold, a tiered response process is initiated; wherein, the tiered response process includes: Initial warning stage: A gentle reminder is issued and the user is prompted to rest. At the same time, a rest mode of a preset duration is initiated to conduct key measurements to obtain and verify physiological data. During the sleep state, the sampling time is extended and silent verification is performed. Verification and judgment stage: Based on the verified physiological data, determine whether an abnormality is confirmed; Tiered response phase: If the review and judgment confirm an anomaly, the corresponding response action is triggered according to the anomaly level, which includes low risk, medium risk and high risk.
2. The method for proactive health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, The multimodal physiological data also includes time information, which is used to identify the current physiological state by combining it with a preset user habit schedule; wherein, the user habit schedule includes typical sleep periods learned through historical data.
3. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, The dynamic threshold library is dynamically adjusted based on the user's personal baseline data through a self-learning mechanism, specifically including: For each physiological indicator, the long-term rolling mean μ and standard deviation σ are calculated under each physiological state to obtain the individual baseline; Based on the personal baseline, a dynamic threshold is set for each physiological state using a threshold adjustment formula. The target threshold is equal to the rolling average μ of the corresponding physiological indicator plus the product of the adjustment coefficient k and the standard deviation σ of the corresponding physiological indicator. The adjustment coefficient k for each physiological indicator is different in the daytime activity state and the sleep state. The individual baseline and thresholds are recalculated periodically or irregularly to accommodate changes in the user's physiological state.
4. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, The initial warning stage includes emitting equipment vibration and displaying prompt information; in the resting mode, the duration of the key measurement is 30s-60s, and during this period, a higher sampling frequency and stricter signal quality judgment standards are used compared to the conventional acquisition.
5. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, The graded response phase includes: Low-risk abnormalities: Record events and generate health reports, and regularly remind users to pay attention; Medium-risk anomaly: Triggers a local alarm on the device and displays brief instructions; High-risk anomaly: Automatically sends a help text message / call containing location information to preset emergency contacts, and / or simultaneously triggers a linkage with nearby emergency centers.
6. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, Also includes: Step 50: Based on physiological data collected over a long period of time under different physiological states, calculate cross-state trend characteristics, and use machine learning models to generate early profiles of user health risks, and issue early warnings when the risk index exceeds a preset risk threshold.
7. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 6, characterized in that, The calculation steps for the cross-state trend characteristics include: Collect users' historical physiological data under different physiological states, including daytime resting heart rate, nighttime average heart rate, and sleep heart rate variability; Calculate cross-state indicators and their rate of change of trends, and use a machine learning model to generate a risk index based on the calculation results; When the risk index exceeds the preset risk threshold, an early risk report is generated and the user is notified, and an early profile of the user's health risk and early warning suggestions are output.
8. The method for proactive review and health early warning based on multi-state identification and cross-state dynamic threshold as described in claim 1, characterized in that, During the verification and judgment phase, the preset monitoring time is automatically extended, and the signal quality index is used for confirmation. An anomaly is only confirmed when the calculated signal quality index exceeds a preset threshold.
9. A health early warning device based on multi-state recognition, cross-state dynamic threshold active verification, characterized in that, include: The physiological data acquisition module is used to collect multimodal physiological data in real time through a smart device worn by the user, including exercise data, heart rate data, respiratory rate data, sleep data, blood oxygen data, body surface temperature data, ambient temperature data, pulse wave data, and blood pressure data. The physiological state recognition module is used to identify the user's current physiological state based on the multimodal physiological data and using a pre-trained classification model built into the smart device. The physiological state includes at least daytime activity state and sleep state. The warning threshold retrieval module is used to retrieve a set of warning thresholds corresponding to the current physiological state from a preset dynamic threshold library based on the identified current physiological state; wherein, the dynamic threshold library sets independent warning thresholds for the same physiological indicator under different physiological states; The tiered response initiation module is used to initiate a tiered response process when the user's physiological parameters monitored in real time exceed the currently invoked warning threshold; the tiered response process includes: Initial warning stage: A gentle reminder is issued and the user is prompted to rest. At the same time, a rest mode of a preset duration is initiated to conduct key measurements to obtain and verify physiological data. During the sleep state, the sampling time is extended and silent verification is performed. Verification and judgment stage: Based on the verified physiological data, determine whether an abnormality is confirmed; Tiered response phase: If the review and judgment confirm an anomaly, the corresponding response action is triggered according to the anomaly level, which includes low risk, medium risk and high risk.