Night physiological state monitoring method, device and equipment based on multi-mode time sequence physiological signals and medium

Abstract

The invention discloses a nighttime physiological state monitoring method, device and equipment based on a multi-mode time sequence physiological signal and a medium, and relates to the technical field of artificial intelligence, and the method comprises the steps: extracting the time sequence characteristics of each mode under each sub-window from the physiological signal of each mode collected under a nighttime time window, and determining the characteristic change direction between the adjacent sub-windows; determining a first linkage feature and a second linkage feature of a preset time sequence coupling linkage mode in a preset time synchronization linkage mode, and determining an association relationship among the fluctuation features of each mode as a third linkage feature of a preset stability linkage mode to obtain linkage mode features including the first, second and third linkage features; scoring the first, second and third linkage features to obtain an evaluation result; and aggregating the time sequence features, the linkage mode features and the evaluation result to obtain a corresponding night physiological status monitoring result. And carrying out night physiological linkage relationship analysis on the multi-modal time sequence physiological data to obtain a more accurate night physiological state.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence technology, and in particular to a method, apparatus, device, and medium for monitoring nighttime physiological state based on multimodal temporal physiological signals. Background Technology

[0002] With the rapid development of wearable devices, in-hospital monitoring equipment, and home health monitoring systems, various physiological signals such as blood pressure, heart rate, respiratory parameters, and sleep structure can be continuously or intermittently collected in scenarios such as nighttime sleep. Moreover, individual behavior is relatively stable and there is less external interference during the nighttime period, making the collected data more conducive to reflecting the true physiological regulatory characteristics. It is an important window for observing the coordinated work of the cardiovascular system, autonomic nervous system, and respiratory regulation system.

[0003] Currently, there are significant bottlenecks in the comprehensive utilization of multimodal physiological data at night. The analysis of nighttime physiological data is usually centered on a single modality. Independent analysis of various indicators cannot form an overall representation of the nighttime physiological state, making it difficult to reflect the coordination or disorder of the physiological system. Existing output formats are mostly single indicator values, scores, or text descriptions, lacking a unified and structured representation of the state.

[0004] In summary, how to analyze the nocturnal physiological linkages of multimodal time-series physiological data to obtain more accurate nocturnal physiological states is a problem that needs to be solved in this field. Summary of the Invention

[0005] In view of this, the purpose of this invention is to provide a method, device, equipment, and medium for monitoring nighttime physiological states based on multimodal temporal physiological signals, and to analyze the nighttime physiological linkage relationship of multimodal temporal physiological data to obtain a more accurate nighttime physiological state. The specific solution is as follows: In a first aspect, this application discloses a method for monitoring nighttime physiological states based on multimodal temporal physiological signals, including: The physiological signals of the target object in each modality are received by the physiological signal acquisition device during the nighttime time window, and the nighttime time window is divided into multiple sliding sub-windows; The temporal features of each modality under each sliding sub-window are extracted from the physiological signals after time stamp alignment, and the direction of feature change between the temporal features of the same modality under adjacent sliding sub-windows is determined. Based on the characteristic change direction of different modes under the same sliding sub-window, a first linkage feature between physiological signals under a preset time synchronization linkage mode is determined. Based on the characteristic change direction of different modes under different sliding sub-windows, a second linkage feature between physiological signals under a preset time coupling linkage mode is determined. Based on the physiological signals under all sliding sub-windows, the fluctuation features of each mode are determined. The correlation between the fluctuation features of each mode is determined as a third linkage feature of a preset stability linkage mode, so as to obtain a linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature. The first linkage feature, the second linkage feature, and the third linkage feature are scored to obtain the evaluation result; The temporal features, the linkage pattern features, and the evaluation results are aggregated into a physiological state vector under the nighttime time window to obtain the corresponding nighttime physiological state monitoring results.

[0006] Optionally, before the physiological signal acquisition device collects physiological signals of the target object in each modality during the nighttime time window, it further includes: The historical collection records of the target object are obtained from a preset historical record storage system, and the start and end times of the nighttime time window are determined based on the historical collection records; wherein, the preset historical record storage system can be any one of a mobile operating system, a wearable device log system, or a medical information system; Alternatively, the start and end times of the nighttime time window can be predicted based on historical physiological signals collected by physiological signal acquisition equipment. Alternatively, the nighttime period received through a preset human-computer interaction interface can be determined as the start and end time of the nighttime time window; Record the start and end times of the nighttime window, the method for determining the nighttime window, and the confidence or quality indicator of the nighttime window; Accordingly, the physiological signals of the target object in each modality collected by the physiological signal acquisition device during the nighttime time window include: The device receives physiological signals from at least two modalities of the target subject, including blood pressure, heart rate, respiration, and sleep signals, collected by a physiological signal acquisition device during a nighttime time window.

[0007] Optionally, the step of extracting the temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals includes: Outlier removal is performed on each of the physiological signals, and the removed physiological signals of each modality are timestamped to obtain the aligned physiological signals of each modality. Temporal features of each modality under each sliding sub-window are extracted from the aligned physiological signals.

[0008] Optionally, the direction of feature change includes an upward direction, a downward direction, and a stable direction; determining the first linkage feature between physiological signals under the preset time synchronization linkage mode based on the direction of feature change of different modalities under the same sliding sub-window includes: Determine whether the direction of feature change is the same for different modalities under the current sliding sub-window; If the changes in the features are in the same direction, then the first linkage feature between the physiological signals under the current sliding sub-window preset time synchronization linkage mode is determined to be used to characterize any one of the linkage relationships among the same upward linkage, same downward linkage, and same stable linkage. Accordingly, determining the second linkage feature between physiological signals under the preset temporal coupling linkage mode based on the feature change direction of different modalities under different sliding sub-windows includes: If the feature change direction of the current mode under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current mode; Determine whether there are other modalities whose feature change direction is upward or downward within a preset number of sliding sub-windows after the leading window. If so, determine the sliding sub-windows whose feature change direction is upward or downward as lagging windows. Based on the leading window and the lagging window, a coupling change event is established, and the number of events under all the sliding sub-windows is counted. The coupling stability score is determined according to the number of events, and a second linkage feature between the physiological signals under a preset temporal coupling linkage mode is generated according to the coupling stability score and each coupling change event.

[0009] Optionally, determining whether the feature change directions of different modalities under the current sliding sub-window are the same includes: Determine the current sliding sub-window, and identify available modes that meet preset availability conditions from each mode based on the physiological signal corresponding to the current sliding sub-window; Determine whether the direction of feature change is the same for different available modalities under the current sliding sub-window; Accordingly, if the feature change direction of the current modality under the current sliding sub-window is an unstable and invariant direction, then marking the current sliding sub-window as the leading window of the current modality includes: If the feature change direction of the currently available modality under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current modality.

[0010] Optionally, determining the correlation between the fluctuation characteristics of each mode as the third linkage characteristic of the preset stability linkage mode includes: Arbitrarily select two modes from each mode as the first mode and the second mode; If the fluctuation characteristic of the first mode is greater than the first preset threshold and the fluctuation characteristic of the second mode is greater than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the preset high fluctuation linkage relationship between the first mode and the second mode. If the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the existence of a preset low fluctuation linkage relationship between the first mode and the second mode. If the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is greater than the second preset threshold, or the fluctuation characteristic of the first mode is greater than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the preset asymmetric fluctuation linkage relationship between the first mode and the second mode.

[0011] Optionally, after aggregating the temporal features, the linkage pattern features, and the evaluation results into a physiological state vector under the nighttime time window, the method further includes: The first physiological state vector in each nighttime time window within a preset time period is determined as the baseline physiological state vector. The physiological state change trend is constructed based on the comparison results between the baseline physiological state vector and the other physiological state vectors within the preset time period.

[0012] Secondly, this application discloses a nighttime physiological state monitoring device based on multimodal temporal physiological signals, comprising: The sub-window division module is used to receive physiological signals of various modalities of the target object collected by the physiological signal acquisition device under the night time window, and divide the night time window into multiple sliding sub-windows; The change direction determination module is used to extract the temporal features of each modality under each sliding sub-window from each of the timestamp-aligned physiological signals, and to determine the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows; The linkage feature determination module is used to determine the first linkage feature between physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modes under the same sliding sub-window, determine the second linkage feature between physiological signals under a preset time coupling linkage mode based on the feature change direction of different modes under different sliding sub-windows, determine the fluctuation feature of each mode based on the physiological signals under all sliding sub-windows, and determine the correlation between the fluctuation features of each mode as the third linkage feature of a preset stability linkage mode, so as to obtain a linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature; The feature comprehensive evaluation module is used to score the first linkage feature, the second linkage feature and the third linkage feature to obtain the evaluation result; The physiological state acquisition module is used to aggregate the temporal features, the linkage mode features and the evaluation results into a physiological state vector under the nighttime time window, so as to obtain the corresponding nighttime physiological state monitoring results.

[0013] Thirdly, this application discloses an electronic device, including: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the aforementioned disclosed method for monitoring nighttime physiological states based on multimodal temporal physiological signals.

[0014] Fourthly, this application discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the aforementioned disclosed method for monitoring nighttime physiological states based on multimodal temporal physiological signals.

[0015] The beneficial effects of this application are as follows: This application receives physiological signals of various modalities of a target object collected by a physiological signal acquisition device under a nighttime time window, divides the nighttime time window into multiple sliding sub-windows; extracts the temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals, and determines the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows; determines the first linkage feature between the physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modalities under the same sliding sub-window, and determines the preset time coupling linkage based on the feature change direction of different modalities under different sliding sub-windows. The second linkage feature between the physiological signals in the mode is determined based on the physiological signals under all the sliding sub-windows to determine the fluctuation features of each mode, and the correlation between the fluctuation features of each mode is determined as the third linkage feature of the preset stable linkage mode, so as to obtain the linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature; the first linkage feature, the second linkage feature and the third linkage feature are scored to obtain the evaluation result; the temporal feature, the linkage mode feature and the evaluation result are aggregated into the physiological state vector under the night time window to obtain the corresponding night physiological state monitoring result. Therefore, this application achieves fine-grained temporal feature extraction of multimodal physiological signals by dividing the data into sliding sub-windows. Based on the feature change direction of adjacent sub-windows, it accurately captures the synchronous linkage relationship and cross-window temporal coupling relationship between different modalities. It combines all sub-window data to extract modal fluctuation features and mine their correlations, forming a complete linkage pattern feature set containing three types of linkage features. The evaluation results are then obtained through scoring and aggregated into a physiological state vector. This effectively integrates the temporal coordination information and fluctuation correlation information of multimodal physiological signals, avoiding the limitations of single-modal isolated analysis and fixed-window coarse-grained analysis, and improving the completeness and accuracy of nighttime physiological state representation. Through standardized linkage pattern features and hierarchically aggregated physiological state vectors, a structured presentation of multimodal physiological linkage relationships is achieved. The linkage pattern recognition logic based on the correlation between feature change direction and fluctuation features enhances the systematicness and interpretability of multimodal physiological signal linkage relationship modeling, while ensuring the consistency and reliability of physiological state evaluation results. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only 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] Figure 1This is a flowchart of a nighttime physiological state monitoring method based on multimodal temporal physiological signals disclosed in this application; Figure 2 This is a schematic diagram of the structure of a nighttime physiological state monitoring device based on multimodal temporal physiological signals disclosed in this application; Figure 3 This is a structural diagram of an electronic device disclosed in this application. Detailed Implementation

[0018] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] With the rapid development of wearable devices, in-hospital monitoring equipment, and home health monitoring systems, various physiological signals such as blood pressure, heart rate, respiratory parameters, and sleep structure can be continuously or intermittently collected in scenarios such as nighttime sleep. Moreover, individual behavior is relatively stable and there is less external interference during the nighttime period, making the collected data more conducive to reflecting the true physiological regulatory characteristics. It is an important window for observing the coordinated work of the cardiovascular system, autonomic nervous system, and respiratory regulation system.

[0020] Currently, there are significant bottlenecks in the comprehensive utilization of multimodal physiological data at night. The analysis of nighttime physiological data is usually centered on a single modality. Independent analysis of various indicators cannot form an overall representation of the nighttime physiological state, making it difficult to reflect the coordination or disorder of the physiological system. Existing output formats are mostly single indicator values, scores, or text descriptions, lacking a unified and structured representation of the state.

[0021] Therefore, this application provides a nighttime physiological state monitoring scheme based on multimodal temporal physiological signals, which performs nighttime physiological linkage analysis on multimodal temporal physiological data to obtain a more accurate nighttime physiological state.

[0022] See Figure 1 As shown in the figure, this application discloses a method for monitoring nighttime physiological state based on multimodal temporal physiological signals, including: Step S11: Receive physiological signals of the target object in each modality collected by the physiological signal acquisition device under the night time window, and divide the night time window into multiple sliding sub-windows.

[0023] In this embodiment, before receiving the physiological signals of the target object in each modality collected by the physiological signal acquisition device under the nighttime time window, the method further includes: obtaining the historical acquisition records of the target object from a preset historical record storage system, and determining the start and end times of the nighttime time window based on the historical acquisition records; wherein, the preset historical record storage system is any one of a mobile operating system, a wearable device log system, or a medical information system; or, predicting the start and end times of the nighttime time window based on the historical physiological signals collected by the physiological signal acquisition device; or, determining the nighttime period received through a preset human-computer interaction interface as the start and end times of the nighttime time window; and recording the start and end times of the nighttime time window, the method of determining the nighttime time window, and the confidence or quality indicator of the nighttime time window.

[0024] There are three main ways to determine the nighttime time window for a target object. The first is to obtain the target object's historical data collection records from a preset historical data storage system (this system can be any of the mobile operating system, wearable device log system, or medical information system). Based on the historical data collection records, the start and end times of the nighttime time window are determined. In other words, it is read from an external system, such as from the sleep recording API (Application Programming Interface) of a smartphone operating system. The system obtains the user's sleep time period through an application programming interface (API). It reads the sleep start and end times identified by the device from the wearable device's sleep logs, or obtains the patient's rest time markings from the hospital monitoring system. The second method uses historical physiological signals collected by physiological signal acquisition devices. By analyzing the composite characteristics of multimodal signals such as activity level, body position changes, heart rate stability, and respiratory patterns, the system predicts the start and end times of the nighttime time window. When the external system does not provide sleep time period information, this system can estimate the start and end boundaries of the nighttime time window based on the composite characteristics of multimodal signals such as activity level, body position changes, heart rate stability, and respiratory patterns, and output the confidence level of the window boundaries. The third method directly determines the user-configured nighttime time period received through a preset human-computer interaction interface as the start and end times of the nighttime time window, and simultaneously records the start and end times of the nighttime time window, i.e., using a fixed time window. When the first or second method is unavailable or the confidence level is lower than a preset threshold, a fixed nighttime time window (e.g., 22:00 to 07:00 the next day) or a user-configured time period is used as the analysis window.

[0025] When determining the nighttime time window, it is also necessary to record the method of determining the nighttime time window (including three types: obtaining from the preset historical record storage system, prediction based on historical physiological signals, and receiving user configuration through the human-computer interaction interface), as well as the corresponding confidence index (for the case of prediction based on historical physiological signals) or quality index. That is to say, it includes the start time, end time, window source type (externally provided / system estimated / fixed window), confidence index (for the case of system estimation), and data quality index.

[0026] In this embodiment, the physiological signals of the target object collected by the receiving physiological signal acquisition device under the nighttime time window include at least two of the following physiological signals collected by the receiving physiological signal acquisition device under the nighttime time window: blood pressure signal, heart rate signal, respiratory signal, and sleep signal of the target object.

[0027] The system receives physiological signals from various modalities of the target subject collected by a physiological signal acquisition device during a nighttime time window. Specifically, it receives physiological signals from at least two modalities among the target subject's blood pressure, heart rate, respiration, and sleep signals. Data sources can include various physiological monitoring devices or wearable devices such as continuous non-invasive blood pressure monitoring devices, ECG monitoring devices, breathing belt sensors, and accelerometers. Specifically, the blood pressure signal includes continuous blood pressure waveforms, intermittent blood pressure measurements (systolic pressure, diastolic pressure, and mean arterial pressure), and blood pressure variability indicators. Data sources can include timed measurement records from continuous non-invasive blood pressure monitoring devices, home electronic blood pressure monitors, or blood pressure estimates based on photoplethysmography (PPG). Heart rate signals include instantaneous heart rate time series, time-domain indicators of heart rate variability, and frequency-domain indicators of heart rate variability. Time-domain indicators of heart rate variability include RMSSD (Root Mean Square of Successive Differences), SDNN (Standard Deviation of Normal-to-Normal Intervals), and pNN50 (Percentage of NN Intervals Differing by More Than 50 ms). Frequency-domain indicators of heart rate variability include LF (Low Frequency), HF (High Frequency), and the LF / HF ratio. Data sources can include ECG monitoring, PPG sensors, or heart rate recordings from wearable devices. Respiratory signals include respiratory rate, respiratory amplitude, apnea event markers, and blood oxygen saturation. Data sources can include respiratory belt sensors, pulse oximeters, or respiratory signal extraction based on heart rate variability. Sleep signals include body movement signals, sleep stage estimates (light sleep, deep sleep, and REM sleep), and the number of awakenings. Data sources can include accelerometers, body position sensors, or sleep staging algorithms based on multimodal signals.

[0028] Within a defined individualized nighttime time window, it is divided into multiple consecutive sliding sub-windows according to configurable parameters. The length of the sliding sub-window is set to 5 to 30 minutes, and the sliding step size is set to 50% of the window length (i.e., there is a 50% overlap between adjacent sliding sub-windows). This setting can form a sequence of continuously covered and partially overlapping sliding sub-windows within the nighttime period. For example, when the window length is 10 minutes and the step size is 5 minutes, an 8-hour nighttime period can be divided into approximately 95 sliding sub-windows, providing fine-grained time unit support for subsequent temporal feature extraction of various modalities and linkage pattern recognition.

[0029] Step S12: Extract the temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals, and determine the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows.

[0030] In this embodiment, the step of extracting the temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals includes: removing outliers from each physiological signal and aligning the timestamps of the removed physiological signals of each modality to obtain the aligned physiological signals of each modality; and extracting the temporal features of each modality under each sliding sub-window from the aligned physiological signals.

[0031] First, data quality control is performed on each of the physiological signals to detect and mark outliers, sensor detachment segments, and segments with poor signal quality to complete the outlier removal. Then, the physiological signals of each modality after outlier removal are timestamped, aligned to a unified time coordinate system, and the problem of inconsistent sampling frequencies is handled to obtain the aligned physiological signals of each modality.

[0032] Subsequently, temporal features of each modality are extracted from each sliding sub-window. Blood pressure modality features include the mean, standard deviation, maximum, minimum, trend (slope of linear fitting), and number of abrupt changes (number of times the change exceeds a threshold) of systolic and diastolic blood pressure within the window. Heart rate modality features include the mean, standard deviation, heart rate variability indices (RMSSD, SDNN), and heart rate trend within the window. Respiratory modality features include the mean respiratory rate, respiratory rate variability, number of apnea events, and mean and minimum blood oxygen saturation within the window. Sleep modality features include the number of body movements within the window, dominant sleep stage (the sleep stage with the highest percentage within the window), and wakefulness markers. The output of feature extraction is a set of feature vectors for each sliding window, denoted as {F_BP(t), F_HR(t), F_Resp(t), F_Sleep(t)}, where t represents the center time of the window.

[0033] For each modality, the change in the corresponding temporal feature of that modality under two adjacent sliding sub-windows is calculated. For example, the mean systolic blood pressure BP_mean(t) of the previous window t minus the mean systolic blood pressure BP_mean(t-1) of the next window t-1, the mean heart rate HR_mean(t) of the previous window t minus the mean heart rate HR_mean(t-1) of the next window t-1, etc. The direction of feature change is determined based on a preset threshold. If the change is greater than a preset rising threshold, it is determined to be "rising". If the change is less than a preset falling threshold, it is determined to be "falling". If the change is between the preset rising threshold and the preset falling threshold, it is determined to be "stable". The preset thresholds for different modalities can be configured or adaptively adjusted according to the device accuracy, population characteristics or application scenario, so as to determine the direction of feature change between temporal features of the same modality under adjacent sliding sub-windows.

[0034] Step S13: Determine the first linkage feature between physiological signals under the preset time synchronization linkage mode based on the feature change direction of different modes under the same sliding sub-window; determine the second linkage feature between physiological signals under the preset time coupling linkage mode based on the feature change direction of different modes under different sliding sub-windows; determine the fluctuation feature of each mode based on the physiological signals under all sliding sub-windows; and determine the correlation between the fluctuation features of each mode as the third linkage feature of the preset stability linkage mode, so as to obtain the linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature.

[0035] In an embodiment of the time-synchronized linkage mode, the direction of feature change includes an upward direction, a downward direction, and a stable direction. Determining the first linkage feature between physiological signals under the preset time-synchronized linkage mode based on the direction of feature change of different modalities under the same sliding sub-window includes: determining whether the direction of feature change of different modalities under the current sliding sub-window is the same; if the direction of feature change is the same, then determining that the first linkage feature between physiological signals under the preset time-synchronized linkage mode under the current sliding sub-window is used to characterize any one of the linkage relationships among upward linkage, downward linkage, and stable linkage in the same direction.

[0036] The time-synchronized linkage mode refers to the simultaneous change in physiological parameters of multiple modalities within the same sliding time window, forming a joint change pattern. Characteristic change directions include upward, downward, and stable directions. Based on the characteristic change directions of different modalities under the same sliding sub-window, the first linkage feature between physiological signals under the preset time-synchronized linkage mode is determined. Specifically, it is determined whether the characteristic change directions of different modalities (such as combinations of blood pressure and heart rate, respiration and sleep, etc.) under the current sliding sub-window are the same. If all modalities are in an upward direction, the first linkage feature under the preset time-synchronized linkage mode under the current sliding sub-window is used to characterize upward linkage in the same direction. If all modalities are in a downward direction, the first linkage feature is used to characterize downward linkage in the same direction. If all modalities are in a stable direction, the first linkage feature is used to characterize stable linkage in the same direction. For example, if blood pressure and heart rate change in the same direction (rising or falling simultaneously), it is identified as linkage in the same direction; if blood pressure and heart rate change in opposite directions, it is identified as linkage in opposite directions; if at least one is stable, there is no obvious synchronous linkage in that window. The synchronization strength can be further quantified by methods such as normalizing the product of change amplitudes or the correlation of time series within the window, forming a complete time-synchronized linkage mode feature. To quantify the synchronization strength, `Score_sync` is defined as a correlation measure of the normalized product of change amplitudes or the rate of change, ranging from 0 to 1, with higher values ​​indicating stronger synchronization. Finally, this mode outputs a time-synchronized linkage mode object, containing fields such as: window time, linkage type (same direction / opposite direction / none), synchronization strength, and a list of involved modalities.

[0037] In this embodiment, determining whether the feature change directions of different modalities under the current sliding sub-window are the same includes: determining the current sliding sub-window, and identifying available modalities that meet preset availability conditions from each modality based on the physiological signal corresponding to the current sliding sub-window; and determining whether the feature change directions of different available modalities under the current sliding sub-window are the same.

[0038] The current sliding sub-window is determined, and combined with the physiological signals of each modality corresponding to the sliding sub-window (blood pressure related signals, heart rate or heart rate variability signals, respiratory related parameters, sleep structure parameters, etc.), the modalities corresponding to outliers, sensor detachment segments, and poor signal quality segments are detected and excluded. The usable modalities that meet the minimum requirement for the number of data points, meet the signal quality standard, and have no missing markers are identified. Then, based on the preset threshold (which can be configured or adaptively adjusted according to the device accuracy, population characteristics, or application scenario), the characteristic change direction (rising, falling, or stable) of each usable modality under the current sliding sub-window is determined, and then it is determined whether the characteristic change directions of different usable modalities are the same.

[0039] Furthermore, the feature mapping method under the time-synchronized linkage mode is as follows: When the system determines that a time-synchronized linkage mode is formed between blood pressure-related parameters and other modal parameters such as heart rate and respiration within a certain nighttime sliding time window or multiple consecutive windows, that is, after determining the first linkage feature, it will further analyze the stability of the blood pressure drop rate between adjacent windows under the synchronous linkage condition. The calculation method of the variability of the nighttime blood pressure drop rate is as follows: First, clarify the linkage preconditions (which must be met simultaneously), that is, within the individualized nighttime time window, the nighttime period is divided into consecutive sliding time windows. In adjacent windows, if the blood pressure-related parameters and heart rate or respiration-related parameters are consistent in the direction of change, or if the blood pressure change amplitude is significantly correlated with the change amplitude of other modal parameters within the same window, then the time-synchronized linkage mode is determined to be established, and the variability of the nighttime blood pressure drop rate is calculated only in the set of windows where the mode is established; then, define the nighttime blood pressure drop rate. In window i, the nighttime blood pressure drop rate Di = (BP_baseline - BP_i) / BP_baseline, where BP_baseline is the reference blood pressure value determined for the individual before the start of the nighttime time window or in the early night (selectable from the statistical value of the early nighttime window, the daytime reference value, or the historical nighttime baseline), and BP_i is the average blood pressure value within the i-th sliding window; then, variability is calculated, and a blood pressure drop rate sequence {Di|i∈W_sync} is constructed within the set W_sync of all windows that meet the conditions for the establishment of the time synchronization linkage mode. The variability of the nighttime blood pressure drop rate is defined as the normalized dispersion of this sequence, i.e., V_BP_sync = StdDev(Di) / [Mean(|Di|) + [ε], where ε is a very small positive number to prevent the denominator from being zero. This variability reflects the stability of blood pressure reduction behavior under the background of multimodal synchronous regulation. Its technical meaning is that if V_BP_sync is low, it means that blood pressure reduction is consistent during synchronous regulation. If V_BP_sync is high, it means that blood pressure reduction has significant instability under the background of synchronous linkage. When it is detected that the blood pressure reduction rate fluctuates significantly between different windows while the variation amplitude of other modal parameters remains relatively stable under the continuous existence of synchronous linkage mode, the system maps the variability of the nighttime blood pressure reduction rate to synchronous linkage stability feature, which is used to indicate the inconsistency of the degree of synchronous regulation of blood pressure regulation behavior with other physiological systems.

[0040] In the temporal coupling linkage mode embodiment, determining the second linkage feature between physiological signals under the preset temporal coupling linkage mode based on the feature change direction of different modalities under different sliding sub-windows includes: if the feature change direction of the current modality under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current modality; determining whether there are other modalities whose feature change direction is rising or falling within the subsequent preset number of sliding sub-windows of the leading window, and if so, determining the sliding sub-windows whose feature change direction is rising or falling as lagging windows; establishing coupling change events based on the leading window and the lagging window, and counting the number of coupling change events under all sliding sub-windows, determining the coupling stability score based on the number of events, and generating the second linkage feature between physiological signals under the preset temporal coupling linkage mode based on the coupling stability score and each coupling change event.

[0041] The temporal coupling linkage mode refers to the situation where, in multiple consecutive time windows, the change of mode A statistically leads or lags behind the change of mode B, and this leading or lagging relationship remains stable across multiple windows. By observing multiple consecutive time windows (e.g., the observation window sequence t-2, t-1, t, t+1, t+2), it is possible to detect whether the change of mode A systematically precedes or lags the change of mode B, and to identify linkage relationships with causal or responsive characteristics. If the characteristic change direction of the current modality (such as breathing, sleep structure, etc.) under the current sliding sub-window is an unstable and invariant direction (i.e., an upward or downward direction), then the current sliding sub-window is marked as the leading window of the current modality; it is determined whether there are other modalities (such as heart rate, blood pressure, etc.) with characteristic change directions of an upward or downward direction within the preset number (1 to 3, configurable) sliding sub-windows after the leading window. If so, the sliding sub-windows with characteristic change directions of an upward or downward direction for other modalities are determined as lagging windows, and the typical time delay from the leading window to the lagging window is recorded (in terms of window number or minutes); a coupled change event is established based on the leading window and the lagging window; for example, for N consecutive windows (N≥5), the following patterns are detected: heart rate response caused by apnea, detecting whether the heart rate rises within 1-3 windows (windows t+1, t+2, t+3) after the apnea event (window t); blood pressure drop caused by deep sleep transition, detecting whether blood pressure drops within the window after the sleep stage changes from light sleep to deep sleep (window t). In this way, the number of coupled change events under all sliding sub-windows is counted, and the total number of events in the current mode whose feature change direction is not stable and unchanged during the entire nighttime period is counted. The ratio of the number of coupled change events to the total number of events is used as the coupling stability score (if the score exceeds a preset ratio, such as 30%, a stable temporal coupling linkage mode is confirmed to exist). Then, combined with the typical time delay of each coupled change event and the information of the leading and lagging modes involved, a second linkage feature between physiological signals under the preset temporal coupling linkage mode is generated. This feature includes quantitative information such as coupling mode type, typical time delay, stability score, and frequency of occurrence.

[0042] In this embodiment, the step of marking the current sliding sub-window as the leading window of the current modality if the feature change direction of the current modality under the current sliding sub-window is an unstable and invariant direction includes: if the feature change direction of the currently available modality under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current modality.

[0043] First, data quality control is performed on the physiological signals of each modality under the current sliding sub-window. Abnormal values, sensor detachment segments, and segments with poor signal quality are detected and marked. The currently available modality is identified if the number of data points meets the minimum requirement, the signal quality meets the standard, and there are no missing data markers. If the characteristic change direction of the currently available modality under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the currently available modality.

[0044] The feature mapping method under the time-series coupling linkage mode is as follows: After the system determines that there is a stable leading or lagging relationship between blood pressure-related parameters and heart rate or respiratory-related parameters and constructs a time-series coupling linkage mode, it will analyze the consistency of blood pressure drop rate changes across windows in the time window sequence where the coupling relationship is established. The calculation method of the variability of nighttime blood pressure drop rate is as follows: First, clarify the linkage precondition, that is, in multiple consecutive nighttime sliding windows, if it is detected that blood pressure changes are statistically leading or lagging behind changes in heart rate or respiratory-related parameters, and the leading or lagging relationship remains stable in multiple windows, then the time-series coupling linkage mode is determined to be established; then, construct the time-aligned drop rate sequence. In a preferred embodiment, for the time delay Δt of blood pressure leading or lagging, the blood pressure drop rate sequence is time-aligned with the corresponding modal change sequence to construct a coupling window set W_couple. In each coupling window j, the blood pressure drop rate is defined as D_j_aligned = (BP_baseline - BP_(j+Δt)) / BP_baseline (where BP_baseline is the reference blood pressure value determined for the individual before the start of the nighttime time window or in the early night, which can be selected from the statistical value of the early nighttime window, the daytime reference value, or the historical nighttime baseline, and BP_(j+Δt) is the average blood pressure value within the corresponding delayed window); then, variability calculation is performed (preferred method). In all aligned windows that satisfy the time-series coupling linkage mode, the variability of the nighttime blood pressure decrease rate is defined as V_BP_couple = StdDev(D_j_aligned) / [Mean(|D_j_aligned|]. +ε], where ε is a very small positive number to prevent the denominator from being zero; its technical meaning is that this variability is used to characterize the stability of blood pressure regulation at the time response level. Low variability indicates that blood pressure changes have a consistent time structure in response to other modalities, while high variability indicates that the blood pressure response delay is unstable, indicating an abnormal time coupling state. When it is detected that while blood pressure changes maintain a leading or lagging relationship, its rate of decline exhibits irregular fluctuations in different windows, and this fluctuation characteristic does not match the time rhythm of heart rate or respiratory signals, the system maps the variability of the nighttime blood pressure decline rate as a time coupling stability deviation feature, which is used to characterize the unstable state of blood pressure regulation at the time response level.

[0045] In an embodiment of the stability linkage mode, determining the correlation between the fluctuation characteristics of each mode as the third linkage characteristic of the preset stability linkage mode includes: arbitrarily selecting two modes from each mode as the first mode and the second mode; if the fluctuation characteristic of the first mode is greater than a first preset threshold and the fluctuation characteristic of the second mode is greater than a second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize a preset high fluctuation linkage relationship between the first mode and the second mode; if the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize a preset low fluctuation linkage relationship between the first mode and the second mode; if the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is greater than the second preset threshold, or the fluctuation characteristic of the first mode is greater than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize a preset asymmetric fluctuation linkage relationship between the first mode and the second mode.

[0046] The stability linkage mode refers to the correlation between the fluctuation characteristics (such as variability, rhythm stability, and mutation frequency) of different modalities, rather than a direct numerical correlation. Two modalities are randomly selected from various modalities such as blood pressure, heart rate, respiration, and sleep structure as the first and second modalities. First, fluctuation characteristics of each modality are extracted based on sliding window data throughout the nighttime period, such as blood pressure variability (CV_BP), heart rate variability (CV_HR), respiratory rate variability, sleep stage transition frequency, and mutation frequency of each modality. The coefficient of variation is calculated as the ratio of the standard deviation to the mean of the corresponding modality's time-series characteristic sequence, and the mutation frequency is the percentage of windows whose time-series characteristics exceed the overall mean ± 2 standard deviations. If the fluctuation characteristics of the first modality are greater than a first preset threshold and the fluctuation characteristics of the second modality are greater than a second preset threshold (the preset thresholds can be configured or adaptively adjusted according to device accuracy, population characteristics, or application scenarios), then the third linkage of the preset stability linkage mode is determined. The features are used to characterize a preset high-fluctuation linkage relationship (i.e., double high-fluctuation linkage) between the first mode and the second mode; if the fluctuation feature of the first mode is less than the first preset threshold and the fluctuation feature of the second mode is less than the second preset threshold, then the third linkage feature of the preset stability linkage mode is used to characterize a preset low-fluctuation linkage relationship (i.e., double low-fluctuation linkage) between the first mode and the second mode; if the fluctuation feature of the first mode is less than the first preset threshold and the fluctuation feature of the second mode is greater than the second preset threshold, or the fluctuation feature of the first mode is greater than the first preset threshold and the fluctuation feature of the second mode is less than the second preset threshold, then the third linkage feature of the preset stability linkage mode is used to characterize a preset asymmetric fluctuation linkage relationship (i.e., asymmetric stability mode) between the first mode and the second mode. Specifically, firstly, fluctuation features, i.e., second-order features, are extracted. For example, blood pressure variability is calculated by calculating the coefficient of variation of systolic blood pressure (CV_BP = StdDev(BP_all) / Mean(BP_all)) throughout the night; heart rate variability is calculated by calculating the coefficient of variation of heart rate (CV_HR) throughout the night, or by using standard HRV (Heart Rate Variability) indicators such as SDNN (Standard Deviation of Normal-to-Normal Intervals); and mutation frequency is counted by counting the number of mutations in blood pressure or heart rate that exceed the normal fluctuation range (e.g., mean ± 2 standard deviations) during the night. Next, second-order feature association is determined. If blood pressure variability is high (CV_BP > threshold) and heart rate variability is also high (CV_HR > threshold), it is identified as a dual-high fluctuation linkage. If blood pressure mutations are frequent and heart rate mutations are also frequent, it is identified as a synchronous unstable linkage. If one side has high fluctuations while the other side has low fluctuations, it is identified as an asymmetric stability pattern.Simultaneously, by quantifying the correlation strength between different modal fluctuation characteristics, a complete third linkage feature can be formed. This feature is used to reveal the synergistic or dysregulatory state of the regulatory stability of different physiological systems. In other words, the stability linkage mode object includes fields such as linkage type (double high fluctuation / synchronous instability / asymmetry, etc.), variability index of each modality, correlation strength, and clinical interpretation prompts.

[0047] The feature mapping method under the stable linkage mode is as follows: When the system constructs a stable linkage mode and performs second-order feature analysis on the volatility or rhythm stability of different modal signals, the variability of the nocturnal blood pressure drop rate, as one of the volatility features of the blood pressure mode, is included in the stable linkage feature set. The calculation method of the variability of the nocturnal blood pressure drop rate is as follows: First, clarify the linkage preconditions. In the stable linkage mode, the absolute changes within a single time window are not compared, but the correlation between the volatility features of different modes is analyzed within the entire nighttime time window. Then, construct the nocturnal blood pressure drop rate sequence. Within the entire nighttime time window, construct the blood pressure drop rate sequence {Dk} corresponding to each sliding time window, k = 1, 2, ..., N (where each Dk is the blood pressure drop rate within the corresponding sliding time window, Dk = (BP_baseline - BP_k) / BP_baseline is the reference blood pressure value determined for the individual before the start of the nighttime time window or in the early night. It can be selected from one of the statistical values ​​of the early nighttime window, the daytime reference value, or the historical nighttime baseline; BP_k is the average blood pressure value within the k-th sliding window); then, variability calculation is performed (preferred method). In a preferred embodiment, the variability of the nighttime blood pressure drop rate is defined as V_BP_stable = StdDev(Dk) / [Mean(|Dk|]. +ε] (where ε is a very small positive number to prevent the denominator from being zero), and this variability is used as a blood pressure fluctuation feature, which together with the fluctuation features of other modalities (such as heart rate variability, respiratory rhythm stability indicators, etc.) constitutes a stability linkage feature vector; the technical meaning is that if V_BP_stable increases at the same time as the fluctuation features of other modalities, it indicates that the overall regulatory stability has decreased. If only the blood pressure-related variability increases, it indicates an independent unstable state of the blood pressure regulation system. In this mode, the system does not compare the absolute magnitude of the blood pressure drop rate, but rather analyzes its variability feature together with second-order features such as heart rate variability and respiratory rhythm stability. When the variability of the nocturnal blood pressure drop rate increases significantly and shows an abnormal correlation with the stability features of other modalities, the system maps this feature to a multimodal stability linkage abnormal feature to reflect the downward trend of the overall stability of nocturnal physiological regulation.

[0048] In practical applications, some modal data may be missing during certain periods. For example, a user may not wear a certain sensor for part of the night, or a sensor may temporarily malfunction. This invention provides a modal missing tolerance mechanism to ensure that the system can still output valid linkage modes and state vectors even when data is incomplete. The degradation processing strategy is as follows: 1) Trimodal to bimodal: When one of the four modalities is completely missing (e.g., respiratory modal data is missing), linkage modes are constructed based on the remaining three modalities. For example, synchronous linkage and temporal coupling modes are identified based on blood pressure, heart rate, and sleep modalities. 2) Bimodal to monomodal: When only two modalities are available, the linkage relationship between these two modalities can still be identified, but a comprehensive linkage mode for multiple modalities cannot be constructed. 3) Partial time period missing: When a modality is missing only during certain time periods, the modality does not participate in linkage mode identification during the missing time periods; for the time periods with complete data, linkage modes are identified normally.

[0049] The overall process of constructing linkage patterns is as follows: For each nighttime period, the construction of linkage patterns is executed according to the following process: Process 1: Traverse all sliding windows and identify time-synchronized linkage patterns for each window to form a sequence of synchronized linkage patterns. That is, as the nighttime period progresses, the sliding window moves continuously, moving one step at a time (e.g., 5 minutes). For a new window position, feature extraction and linkage pattern identification are repeated. The technical significance of continuous sliding is to capture the dynamic evolution of linkage patterns, that is, the same linkage pattern may exhibit different intensities or characteristics at different times of the night, and this evolution process can be tracked through continuous sliding; improve the temporal resolution of identification, that is, compared with analyzing the entire nighttime as a single window, the sliding window can provide a more granular characterization of linkage relationships; support real-time or near-real-time processing: in practical applications, the sliding window and linkage patterns can be updated in real time during the nighttime data streaming process, without waiting for the entire night's data collection to be completed. Process 2: Analyze the continuous window sequence, identify time-coupled linkage patterns, and record the lead-lag relationship and stability. Process 3: Based on the data of the entire nighttime period, calculate the second-order features and identify stable linkage patterns. Step 4: Summarize the identification results of the three types of linkage modes to form a set of nighttime physiological linkage mode features.

[0050] Through the above mapping method, the variability of nighttime blood pressure drop rate is not used as an independent output indicator, but rather as a feature dimension of the time-synchronized linkage mode, time-series coupled linkage mode, or stable linkage mode, respectively. This variability is encoded into the nighttime physiological linkage mode feature vector and participates in the hierarchical representation and structured output of nighttime physiological states. In this way, nighttime blood pressure regulation features are always interpreted within the context of multimodal physiological linkage, avoiding isolated analysis of single physiological parameters, thereby improving the stability, consistency, and engineering usability of nighttime physiological state assessment results. In specific implementation, the calculation of the variability of nighttime blood pressure drop rate as a feature dimension of the linkage mode does not use a single fixed formula, but rather, according to the linkage mode type, features are extracted and encoded in an engineering-feasible manner within the corresponding time window set. In a preferred embodiment of the present invention, the above three types of nighttime blood pressure drop rate variability features are not output separately, but are encoded into the nighttime physiological linkage mode feature vector according to the corresponding linkage mode type, for example: F_link = [V_BP_sync, V_BP_couple, V_BP_stable, ...]; Used for subsequent hierarchical characterization and structured output of nocturnal physiological states.

[0051] In this embodiment, the variability of nighttime blood pressure drop rate is not output as an independent physiological indicator, but rather as a feature dimension embedded in the nighttime physiological linkage mode feature vector under different nighttime physiological linkage modes. To ensure the feasibility and stability of the technical solution, in a preferred embodiment, different calculation methods are used for the variability of nighttime blood pressure drop rate for different types of nighttime physiological linkage modes. Through the above calculation rules, the variability of nighttime blood pressure drop rate is encoded as a numerical feature dimension in the linkage mode feature vector under different linkage mode conditions. This feature vector serves as part of the intermediate or final output of the nighttime physiological state stratification and is provided to the upper-level system in structured data form. It should be noted that the above variability calculation is not an unconditional statistical analysis of all nighttime data, but is only performed within a subset of windows where a specific linkage mode is established. This feature reflects the stability of blood pressure regulation behavior in a multimodal collaborative context, rather than the statistical fluctuation of a single physiological indicator.

[0052] Step S14: Score the first linkage feature, the second linkage feature, and the third linkage feature to obtain the evaluation result.

[0053] The evaluation result is obtained by comprehensively scoring the first, second, and third linkage features. The specific scoring rules are as follows: First, a comprehensive state layer scoring system is constructed in the three-layer structured state vector. The linkage coordination score (base score of 100 points) is calculated based on the proportion of the number of reverse linkage windows to the number of unidirectional linkage windows in the first linkage feature (deduct 15 points if it exceeds 50%), the stability score of the temporal coupling mode in the second linkage feature (deduct 10 points if it is below 0.3), and the stability linkage type in the third linkage feature (deduct 20 points if it is a double high fluctuation or synchronous unstable type). The nighttime stability score (base score of 100 points) is combined with the modal mutation frequency of each feature in the first linkage feature (deduct 10 points if it exceeds 5%). The scores are calculated based on the following factors: blood pressure coefficient of variation (CV_BP>0.08, deduct 20 points); heart rate coefficient of variation (CV_HR>0.12, deduct 15 points); and the total number of apnea events (more than 10, deduct 15 points) in the third linkage feature. The final assessment result includes a linkage coordination score (0-100 points), a nighttime stability score (0-100 points), and abnormal linkage indicators (marked according to the abnormal performance of the three linkage features as excessive reverse linkage, unstable temporal coupling, double high fluctuations, etc.). This assessment result is a structured summary of the three linkage features, used to describe the synergistic state and stability of the nighttime physiological multi-system regulation, and does not constitute a medical diagnosis or risk assessment.

[0054] Step S15: Aggregate the time sequence features, the linkage mode features, and the evaluation results into a physiological state vector under the nighttime time window to obtain the corresponding nighttime physiological state monitoring results.

[0055] Specifically, the identified linkage patterns are mapped into a three-layer structured state vector to obtain the corresponding nighttime physiological state monitoring results. The details are as follows: Layer 1 - Original Feature Layer: The original feature layer contains the basic statistical features (also known as time-series features) of each modality throughout the entire nighttime period, serving as the basic dimension of the state vector, as follows: Blood pressure characteristics: nighttime mean systolic blood pressure, mean diastolic blood pressure, blood pressure variability, and blood pressure drop (relative to daytime baseline). Heart rate characteristics: nighttime average heart rate, heart rate variability indicators (RMSSD, SDNN), resting heart rate (lowest 5th percentile); Respiratory characteristics: mean respiratory rate, total number of apnea events, lowest blood oxygen saturation; Sleep characteristics: total sleep duration, percentage of each sleep stage, number of awakenings; Layer 2 - Linkage Mode Layer: The linkage mode layer contains the identification results and quantitative features of three types of linkage modes: Time synchronization linkage characteristics (i.e., the first linkage characteristic): number of linkage windows in the same direction, number of linkage windows in opposite directions, average synchronization strength, and strongest synchronization period identifier; Temporal coupling linkage features (i.e., second linkage features): the type of coupling mode identified (such as respiratory-heart rate coupling), typical time delay, stability score, and frequency of occurrence; Stability linkage characteristics (i.e., the third linkage characteristic): linkage type identifier (such as dual high fluctuations / synchronous instability), variability index of each mode, and correlation strength; Layer 3 - Integrated State Layer: The integrated state layer aggregates multi-dimensional linked features into a holistic assessment of nighttime physiological state: Coordination score: Based on the comprehensive performance of three types of coordination modes, it quantifies the degree of coordination of multiple physiological systems, with a value from 0 to 100; Nighttime stability score: Based on fluctuation linkage patterns and abrupt events, assesses the stability of nighttime physiological state, with a value ranging from 0 to 100; Abnormal linkage indicator: Indicates whether an abnormal linkage pattern has been detected (such as reverse linkage of modes that should be coordinated, frequent sudden changes in periods that should be stable, etc.). The final output format is: The state vector is output in a structured data format (such as JSON) and includes the following fields: Timestamp information: start and end times of the nighttime time window, and the data collection time range; Window type identifier: Individualized window / Fixed window / System estimation window; Three-layer state vector: Layer 1 original feature array, Layer 2 linkage mode feature array, Layer 3 comprehensive score and label; Data quality indicators: data integrity level, complete modality list, missing modality list, confidence level (window for system estimation); The comprehensive score described above is a structured summary of the linkage characteristics, used to describe the characteristics of state changes, and does not constitute a medical diagnosis or risk assessment.

[0056] In this embodiment, after aggregating the temporal features, the linkage mode features, and the evaluation results into a physiological state vector under the nighttime time window, the method further includes: determining the first physiological state vector among the physiological state vectors under each of the nighttime time windows within a preset time period as the baseline physiological state vector; and constructing a physiological state change trend based on the comparison results between the baseline physiological state vector and the remaining physiological state vectors within the preset time period.

[0057] The beneficial effects of this application are as follows: This application receives physiological signals of various modalities of a target object collected by a physiological signal acquisition device under a nighttime time window, divides the nighttime time window into multiple sliding sub-windows; extracts the temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals, and determines the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows; determines the first linkage feature between the physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modalities under the same sliding sub-window, and determines the preset time coupling linkage based on the feature change direction of different modalities under different sliding sub-windows. The second linkage feature between the physiological signals in the mode is determined based on the physiological signals under all the sliding sub-windows to determine the fluctuation features of each mode, and the correlation between the fluctuation features of each mode is determined as the third linkage feature of the preset stable linkage mode, so as to obtain the linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature; the first linkage feature, the second linkage feature and the third linkage feature are scored to obtain the evaluation result; the temporal feature, the linkage mode feature and the evaluation result are aggregated into the physiological state vector under the night time window to obtain the corresponding night physiological state monitoring result. Therefore, this application achieves fine-grained temporal feature extraction of multimodal physiological signals by dividing the data into sliding sub-windows. Based on the feature change direction of adjacent sub-windows, it accurately captures the synchronous linkage relationship and cross-window temporal coupling relationship between different modalities. It combines all sub-window data to extract modal fluctuation features and mine their correlations, forming a complete linkage pattern feature set containing three types of linkage features. The evaluation results are then obtained through scoring and aggregated into a physiological state vector. This effectively integrates the temporal coordination information and fluctuation correlation information of multimodal physiological signals, avoiding the limitations of single-modal isolated analysis and fixed-window coarse-grained analysis, and improving the completeness and accuracy of nighttime physiological state representation. Through standardized linkage pattern features and hierarchically aggregated physiological state vectors, a structured presentation of multimodal physiological linkage relationships is achieved. The linkage pattern recognition logic based on the correlation between feature change direction and fluctuation features enhances the systematicness and interpretability of multimodal physiological signal linkage relationship modeling, while ensuring the consistency and reliability of physiological state evaluation results.

[0058] See Figure 2 As shown in the figure, this application discloses a nighttime physiological state monitoring device based on multimodal temporal physiological signals, comprising: The sub-window division module 11 is used to receive physiological signals of various modalities of the target object collected by the physiological signal acquisition device under the night time window, and divide the night time window into multiple sliding sub-windows; The change direction determination module 12 is used to extract the temporal features of each modality under each sliding sub-window from each of the timestamp-aligned physiological signals, and to determine the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows; The linkage feature determination module 13 is used to determine the first linkage feature between physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modes under the same sliding sub-window, determine the second linkage feature between physiological signals under a preset time coupling linkage mode based on the feature change direction of different modes under different sliding sub-windows, determine the fluctuation feature of each mode based on the physiological signals under all sliding sub-windows, and determine the correlation between the fluctuation features of each mode as the third linkage feature of a preset stability linkage mode, so as to obtain a linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature; The feature comprehensive evaluation module 14 is used to score the first linkage feature, the second linkage feature and the third linkage feature to obtain the evaluation result; The physiological state acquisition module 15 is used to aggregate the temporal features, the linkage mode features and the evaluation results into a physiological state vector under the nighttime time window, so as to obtain the corresponding nighttime physiological state monitoring results.

[0059] Furthermore, embodiments of this application also provide an electronic device. Figure 3 This is a structural diagram of an electronic device 20 according to an exemplary embodiment. The content of the diagram should not be construed as limiting the scope of this application.

[0060] Figure 3 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Specifically, it may include: at least one processor 21, at least one memory 22, a power supply 23, a communication interface 24, an input / output interface 25, and a communication bus 26. The memory 22 stores a computer program, which is loaded and executed by the processor 21 to implement the relevant steps in the nighttime physiological state monitoring method based on multimodal temporal physiological signals disclosed in any of the foregoing embodiments.

[0061] In this embodiment, the power supply 23 is used to provide operating voltage for various hardware devices on the electronic device; the communication interface 24 can create a data transmission channel between the electronic device and external devices, and the communication protocol it follows can be any communication protocol applicable to the technical solution of this application, and is not specifically limited here; the input / output interface 25 is used to acquire external input data or output data to the outside world, and its specific interface type can be selected according to specific application needs, and is not specifically limited here.

[0062] The processor 21 may include one or more processing cores, such as a quad-core processor or an octa-core processor. The processor 21 may be implemented using at least one hardware form selected from DSP (Digital Signal Processing), FPGA (Field-Programmable Gate Array), and PLA (Programmable Logic Array). The processor 21 may also include a main processor and a coprocessor. The main processor, also known as a CPU (Central Processing Unit), is used to process data in the wake-up state; the coprocessor is a low-power processor used to process data in the standby state. In some embodiments, the processor 21 may integrate a GPU (Graphics Processing Unit), which is responsible for rendering and drawing the content to be displayed on the screen. In some embodiments, the processor 21 may also include an AI (Artificial Intelligence) processor, which is used to handle computational operations related to machine learning.

[0063] In addition, the memory 22, as a carrier for resource storage, can be a read-only memory, random access memory, disk or optical disk, etc. The resources stored on it include operating system 221, computer program 222 and data 223, etc., and the storage method can be temporary storage or permanent storage.

[0064] The operating system 221 manages and controls the various hardware devices and computer programs 222 on the electronic device to enable the processor 21 to perform calculations and processing on the massive amounts of data 223 in the memory 22. The operating system can be Windows, Unix, Linux, etc. The computer program 222, in addition to including a computer program capable of performing the nighttime physiological state monitoring method based on multimodal temporal physiological signals disclosed in any of the foregoing embodiments, may further include computer programs capable of performing other specific tasks. The data 223 may include data received by the electronic device from external devices, as well as data collected by its own input / output interface 25.

[0065] Furthermore, this application also discloses a computer-readable storage medium for storing a computer program; wherein, when the computer program is executed by a processor, it implements the aforementioned disclosed method for monitoring nighttime physiological states based on multimodal temporal physiological signals. Specific steps of this method can be found in the corresponding content disclosed in the foregoing embodiments, and will not be repeated here.

[0066] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to in the method section.

[0067] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. 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 implementation should not be considered beyond the scope of this application. The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein can be implemented directly in hardware, software modules executed by a processor, or a combination of both. The software module may be located in random access memory (RAM), memory, read-only memory (ROM), electrically programmable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), register, hard disk, removable disk, CD-ROM (Compact Disc Read-Only Memory), or any other form of storage medium known in the art.

[0068] Finally, it should be noted that in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0069] The foregoing has provided a detailed description of a method, apparatus, device, and medium for monitoring nighttime physiological states based on multimodal temporal physiological signals provided by the present invention. Specific examples have been used to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only intended to help understand the method and core ideas of the present invention. At the same time, for those skilled in the art, there will be changes in specific implementation methods and application scope based on the ideas of the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A method for monitoring nighttime physiological states based on multimodal temporal physiological signals, characterized in that, include: The physiological signals of the target object in each modality are received by the physiological signal acquisition device during the nighttime time window, and the nighttime time window is divided into multiple sliding sub-windows; The temporal features of each modality under each sliding sub-window are extracted from the physiological signals after time stamp alignment, and the direction of feature change between the temporal features of the same modality under adjacent sliding sub-windows is determined. Based on the characteristic change direction of different modes under the same sliding sub-window, a first linkage feature between physiological signals under a preset time synchronization linkage mode is determined. Based on the characteristic change direction of different modes under different sliding sub-windows, a second linkage feature between physiological signals under a preset time coupling linkage mode is determined. Based on the physiological signals under all sliding sub-windows, the fluctuation features of each mode are determined. The correlation between the fluctuation features of each mode is determined as a third linkage feature of a preset stability linkage mode, so as to obtain a linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature. The first linkage feature, the second linkage feature, and the third linkage feature are scored to obtain the evaluation result; The temporal features, the linkage pattern features, and the evaluation results are aggregated into a physiological state vector under the nighttime time window to obtain the corresponding nighttime physiological state monitoring results.

2. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to claim 1, characterized in that, Before the physiological signal acquisition device collects physiological signals of the target object in each modality during the nighttime time window, it also includes: The historical collection records of the target object are obtained from a preset historical record storage system, and the start and end times of the nighttime time window are determined based on the historical collection records; wherein, the preset historical record storage system can be any one of a mobile operating system, a wearable device log system, or a medical information system; Alternatively, the start and end times of the nighttime time window can be predicted based on historical physiological signals collected by physiological signal acquisition equipment. Alternatively, the nighttime period received through a preset human-computer interaction interface can be determined as the start and end time of the nighttime time window; Record the start and end times of the nighttime window, the method for determining the nighttime window, and the confidence or quality indicator of the nighttime window; Accordingly, the physiological signals of the target object in each modality collected by the physiological signal acquisition device during the nighttime time window include: The device receives physiological signals from at least two modalities of the target subject, including blood pressure, heart rate, respiration, and sleep signals, collected by a physiological signal acquisition device during a nighttime time window.

3. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to claim 1, characterized in that, The extraction of temporal features of each modality under each sliding sub-window from the timestamp-aligned physiological signals includes: Outlier removal is performed on each of the physiological signals, and the removed physiological signals of each modality are timestamped to obtain the aligned physiological signals of each modality. Temporal features of each modality under each sliding sub-window are extracted from the aligned physiological signals.

4. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to claim 1, characterized in that, The direction of the characteristic change includes the upward direction, the downward direction, and the stable and unchanged direction; The step of determining the first linkage feature between physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modalities under the same sliding sub-window includes: Determine whether the direction of feature change is the same for different modalities under the current sliding sub-window; If the changes in the features are in the same direction, then the first linkage feature between the physiological signals under the current sliding sub-window preset time synchronization linkage mode is determined to be used to characterize any one of the linkage relationships among the same upward linkage, same downward linkage, and same stable linkage. Accordingly, determining the second linkage feature between physiological signals under the preset temporal coupling linkage mode based on the feature change direction of different modalities under different sliding sub-windows includes: If the feature change direction of the current mode under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current mode; Determine whether there are other modalities whose feature change direction is upward or downward within a preset number of sliding sub-windows after the leading window. If so, determine the sliding sub-windows whose feature change direction is upward or downward as lagging windows. Based on the leading window and the lagging window, a coupling change event is established, and the number of events under all the sliding sub-windows is counted. The coupling stability score is determined according to the number of events, and a second linkage feature between the physiological signals under a preset temporal coupling linkage mode is generated according to the coupling stability score and each coupling change event.

5. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to claim 4, characterized in that, The step of determining whether the feature change directions of different modalities under the current sliding sub-window are the same includes: Determine the current sliding sub-window, and identify available modes that meet preset availability conditions from each mode based on the physiological signal corresponding to the current sliding sub-window; Determine whether the direction of feature change is the same for different available modalities under the current sliding sub-window; Accordingly, if the feature change direction of the current modality under the current sliding sub-window is an unstable and invariant direction, then marking the current sliding sub-window as the leading window of the current modality includes: If the feature change direction of the currently available modality under the current sliding sub-window is an unstable and invariant direction, then the current sliding sub-window is marked as the leading window of the current modality.

6. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to claim 1, characterized in that, The step of determining the correlation between the fluctuation characteristics of each mode as the third linkage characteristic of the preset stability linkage mode includes: Arbitrarily select two modes from each mode as the first mode and the second mode; If the fluctuation characteristic of the first mode is greater than the first preset threshold and the fluctuation characteristic of the second mode is greater than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the preset high fluctuation linkage relationship between the first mode and the second mode. If the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the existence of a preset low fluctuation linkage relationship between the first mode and the second mode. If the fluctuation characteristic of the first mode is less than the first preset threshold and the fluctuation characteristic of the second mode is greater than the second preset threshold, or the fluctuation characteristic of the first mode is greater than the first preset threshold and the fluctuation characteristic of the second mode is less than the second preset threshold, then the third linkage characteristic of the preset stability linkage mode is determined to characterize the preset asymmetric fluctuation linkage relationship between the first mode and the second mode.

7. The method for monitoring nighttime physiological state based on multimodal temporal physiological signals according to any one of claims 1 to 6, characterized in that, After aggregating the temporal features, the linkage pattern features, and the evaluation results into a physiological state vector under the nighttime time window, the method further includes: The first physiological state vector in each nighttime time window within a preset time period is determined as the baseline physiological state vector. The physiological state change trend is constructed based on the comparison results between the baseline physiological state vector and the other physiological state vectors within the preset time period.

8. A nighttime physiological state monitoring device based on multimodal temporal physiological signals, characterized in that, include: The sub-window division module is used to receive physiological signals of various modalities of the target object collected by the physiological signal acquisition device under the night time window, and divide the night time window into multiple sliding sub-windows; The change direction determination module is used to extract the temporal features of each modality under each sliding sub-window from each of the timestamp-aligned physiological signals, and to determine the feature change direction between the temporal features of the same modality under adjacent sliding sub-windows; The linkage feature determination module is used to determine the first linkage feature between physiological signals under a preset time synchronization linkage mode based on the feature change direction of different modes under the same sliding sub-window, determine the second linkage feature between physiological signals under a preset time coupling linkage mode based on the feature change direction of different modes under different sliding sub-windows, determine the fluctuation feature of each mode based on the physiological signals under all sliding sub-windows, and determine the correlation between the fluctuation features of each mode as the third linkage feature of a preset stability linkage mode, so as to obtain a linkage mode feature including the first linkage feature, the second linkage feature and the third linkage feature; The feature comprehensive evaluation module is used to score the first linkage feature, the second linkage feature and the third linkage feature to obtain the evaluation result; The physiological state acquisition module is used to aggregate the temporal features, the linkage mode features and the evaluation results into a physiological state vector under the nighttime time window, so as to obtain the corresponding nighttime physiological state monitoring results.

9. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor is configured to execute the computer program to implement the steps of the nighttime physiological state monitoring method based on multimodal temporal physiological signals as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store a computer program; wherein, when the computer program is executed by a processor, it implements the steps of the nighttime physiological state monitoring method based on multimodal temporal physiological signals as described in any one of claims 1 to 7.

Images