Diabetes metabolism abnormality early warning method based on bio-electricity signals

Abstract

The invention discloses a diabetes mellitus metabolic abnormality early warning method based on a bioelectricity signal, and relates to the technical field of medical detection and physiological signal processing, and the method comprises the following steps: collecting an electrocardiosignal, a surface electromyogram signal and a skin electric signal of a detected object, carrying out preprocessing, and carrying out baseline correction to generate a multi-channel bioelectricity signal sequence; establishing a rhythm coupling detection baseline of the individual in a normal physiological state, and performing comparative analysis with a rhythm coupling state corresponding to the current window in the rhythm coupling state time sequence to generate a metabolic abnormality detection result; and performing diabetes mellitus metabolic abnormality risk judgment on the detected object based on the metabolic abnormality detection result, and outputting a diabetes mellitus metabolic abnormality detection conclusion when the diabetes mellitus metabolic abnormality risk meets a preset early warning trigger condition. According to the method, by generating the rhythm coupling state time sequence, advanced signal mode recognition of dynamic coordination modes among different physiological functions is achieved, and the physiological pertinence and application value of metabolic abnormality recognition are improved.

Description

Technical Field

[0001] This invention relates to the field of medical detection and physiological signal processing technology, and in particular to a method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals. Background Technology

[0002] In the field of health monitoring for diabetes and related metabolic disorders, physiological state assessment methods based on bioelectrical signals have been widely studied and applied. Conventional techniques typically analyze human autonomic nervous activity, muscle function, and stress response characteristics by collecting single bioelectrical information such as electrocardiogram (ECG), surface electromyography (EMG), and electrodermal (EDS) signals. These analyses are combined with statistical analysis and signal pattern recognition methods to assist in the judgment of an individual's metabolic state. With the development of wearable sensors and signal processing methods, continuous time window analysis, multi-channel signal collaborative processing, and pattern feature-based analysis frameworks have gradually become important research directions in this field.

[0003] However, in conventional methods, multi-source bioelectric signals are often analyzed as independent features, making it difficult to characterize the synchronization and phase correspondence of different physiological rhythms on the time axis; some methods focus on instantaneous features and static statistical results, and pay insufficient attention to the coupling stability of bioelectric rhythms over time. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method for early warning of metabolic abnormalities in diabetes based on bioelectric signals to solve the problem that multi-source bioelectric signals are difficult to coordinately reflect the temporal correlation characteristics and dynamic coupling changes between different physiological rhythms.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals. The method includes: acquiring and preprocessing electrocardiogram (ECG), surface electromyography (EMG), and electrodermal (EDS) signals of the subject, and performing baseline correction to generate a multi-channel bioelectrical signal sequence; identifying and extracting rhythmic events from the multi-channel bioelectrical signal sequence within a continuous sliding time window, and associating the rhythmic events corresponding to each bioelectrical signal to generate a multi-source bioelectrical rhythmic event sequence; performing rhythmic relationship analysis on the multi-source bioelectrical rhythmic event sequence to identify the synchronization relationship between different bioelectrical rhythms, and determining the phase correspondence and coupling stability between different bioelectrical rhythms by analyzing the relative positional relationship of each rhythmic event on the time axis, generating a rhythmic coupling state time series; establishing a rhythmic coupling detection baseline under normal physiological conditions for the individual, and comparing it with the rhythmic coupling state corresponding to the current window in the rhythmic coupling state time series to generate metabolic abnormality detection results; determining the risk of diabetes metabolic abnormalities in the subject based on the metabolic abnormality detection results, and outputting a diabetes metabolic abnormality detection conclusion when the risk of diabetes metabolic abnormalities meets preset early warning trigger conditions.

[0007] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals according to the present invention, the steps for generating the multi-channel bioelectrical signal sequence are as follows: The preprocessing includes time synchronization and filtering / denoising. Baseline analysis was performed on the preprocessed electrocardiogram, surface electromyography and electrodermal signals to identify relatively stable baseline reference segments for each bioelectric signal and generate a set of baseline reference segments. Based on the baseline reference fragment set, the baseline offset of each bioelectrical signal is calculated, and a baseline correction parameter set is generated; Baseline correction processing was performed on the preprocessed electrocardiogram, surface electromyography, and electrodermal signals using a set of baseline correction parameters to generate multi-channel bioelectrical signal sequences.

[0008] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals described in this invention, the step of identifying and extracting rhythmic events from a multi-channel bioelectrical signal sequence within a continuous sliding time window comprises the following steps: Based on multi-channel bioelectric signal sequences, each bioelectric signal is divided into time segments to form a set of continuous sliding time windows; Within a continuous sliding time window set, cardiac rhythm feature detection is performed on the electrocardiogram signal to identify the time of cardiac occurrence and generate a cardiac rhythm event sequence; Within a continuous sliding time window set, muscle activation state is identified from surface electromyography signals, the start and end times of muscle activation are extracted, and a muscle activation rhythm event sequence is generated. Within a continuously sliding time window set, changes in the skin electrical signal are identified, skin electrical response events are extracted, and a skin electrical response rhythm event sequence is generated.

[0009] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectric signals described in this invention, the multi-source bioelectric rhythm event sequence is generated by correspondingly associating cardiac rhythm event sequences, muscle activation rhythm event sequences, and skin conductance response rhythm event sequences according to their time order and corresponding sliding time windows.

[0010] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectric signals described in this invention, the steps of performing rhythmic relationship analysis on multi-source bioelectric rhythmic event sequences to identify the synchronization relationship between different bioelectric rhythms are as follows: The sequences of cardiac rhythm events, muscle activation rhythm events, and skin conductance response rhythm events are organized according to the time of occurrence of rhythm events to form a multi-source rhythm event alignment set; Based on the multi-source rhythm event alignment set, the consistency of occurrence of different bioelectric rhythm events within the same continuous sliding time window is identified, and a rhythm synchronization feature set is generated.

[0011] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals described in this invention, the steps of determining the phase correspondence and coupling stability between different bioelectrical rhythms by analyzing the relative positional relationship of each rhythmic event on the time axis, and generating a time series of rhythmic coupling states are as follows. Based on the rhythm synchronization feature set, the sequential relationship of different bioelectric rhythm events on the time axis is extracted to generate a rhythm sequence relationship feature set; By performing relative position analysis on the set of rhythm sequence relationship features, the relative position relationship between different bioelectric rhythm events is determined, and a set of rhythm phase relationship features is generated. Based on the rhythm phase relationship feature set and the rhythm synchronization feature set, the changes in different bioelectric rhythm relationships are evaluated, the fluctuation degree and continuous consistency of bioelectric rhythm relationships are identified, and a rhythm coupling stability feature set is generated. By integrating the rhythm synchronization feature set, the rhythm phase relationship feature set, and the rhythm coupling stability feature set, a rhythm coupling state time series is generated.

[0012] As a preferred embodiment of the diabetes metabolic abnormality early warning method based on bioelectrical signals described in this invention, the steps for establishing a rhythm coupling detection baseline under normal physiological conditions are as follows: Based on a preset rhythm coupling fluctuation threshold, the time series of rhythm coupling states are screened for stability, the time periods in a stable physiological detection state are identified, and a stable rhythm coupling state dataset is generated. The rhythm coupling states in the stable rhythm coupling state dataset are statistically summarized to construct a rhythm coupling detection baseline.

[0013] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals according to the present invention, the steps for generating metabolic abnormality detection results by comparing and analyzing the rhythm coupling state corresponding to the current window in the rhythm coupling state time series are as follows. Extract the rhythm coupling state corresponding to the current continuous sliding time window from the rhythm coupling state time series, compare it with the rhythm coupling detection baseline, and calculate the degree of rhythm coupling deviation; Based on the degree of deviation of rhythm coupling, the abnormality of the rhythm coupling state corresponding to the current continuous sliding time window is quantitatively evaluated, and metabolic abnormality detection results are generated.

[0014] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals described in this invention, the steps for determining the risk of diabetic metabolic abnormalities in the tested subjects based on the metabolic abnormality detection results are as follows: Based on the results of metabolic abnormality detection, the direction and duration of change of the degree of metabolic abnormality within a continuous sliding time window are analyzed to generate metabolic abnormality risk evolution characteristics. Based on the evolutionary characteristics of metabolic abnormality risk, the current degree of metabolic abnormality of the tested subjects is classified into risk levels, and metabolic abnormality detection level results are generated.

[0015] As a preferred embodiment of the early warning method for diabetic metabolic abnormalities based on bioelectrical signals according to the present invention, the step of outputting a diabetic metabolic abnormality detection conclusion when the risk of diabetic metabolic abnormality meets the preset early warning triggering conditions is as follows: Based on the results of metabolic abnormality detection levels, a comprehensive judgment is made on the changes in metabolic abnormality risk within a continuous sliding time window to generate a metabolic abnormality risk confirmation result. The system invokes preset warning trigger conditions to determine the results of metabolic abnormality risk confirmation, and outputs the detection conclusion of diabetes metabolic abnormality when the warning trigger conditions are met.

[0016] The beneficial effects of this invention are as follows: by identifying and extracting rhythmic events from multi-channel bioelectrical signal sequences within a continuous sliding time window, a multi-source bioelectrical rhythmic event sequence is generated, realizing the rhythmic expression of multi-source bioelectrical information under a unified time structure, and improving the basic quality and reliability of the temporal features upon which subsequent analysis depends; by generating a rhythmic coupling state time series, advanced signal pattern recognition of dynamic coordination modes between different physiological functions is achieved, improving the physiological specificity and application value of metabolic abnormality identification. Attached Figure Description

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

[0018] Figure 1 This is a flowchart of a method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals.

[0019] Figure 2 This is a flowchart for the identification and association of multi-source bioelectrical rhythm events.

[0020] Figure 3 This is a flowchart for establishing and analyzing the baseline for rhythm coupling detection.

[0021] Figure 4 This is a flowchart for rhythm coupling state analysis and early warning of metabolic abnormalities. Detailed Implementation

[0022] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0023] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0024] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0025] Reference Figures 1-4This is one embodiment of the present invention, which provides a method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals, comprising the following steps: S1. Collect the electrocardiogram (ECG), surface electromyography (EMG), and electrodermal (EDS) signals of the test subjects, perform preprocessing, and perform baseline correction to generate a multi-channel bioelectrical signal sequence.

[0026] S1.1 Preprocessing includes time synchronization and filtering for noise reduction.

[0027] It should be noted that time synchronization refers to resampling the ECG, surface electromyography (EMG), and electrodermal (ED) signals at a uniform sampling frequency (e.g., 250Hz) to align the bioelectrical signals from different channels point by point on the same time axis. Filtering and denoising refers to applying digital bandpass filtering to the ECG, EMG, and EED signals to limit the signal frequency range and remove low-frequency non-physiological drift and high-frequency random noise. A fixed-frequency notch filter (e.g., 50Hz) is then applied to each bioelectrical signal after bandpass filtering to suppress power supply frequency interference, resulting in bioelectrical signals with improved signal-to-noise ratio that primarily retain components related to cardiac activity, muscle activation, and EED responses.

[0028] S1.2. Perform baseline analysis on the preprocessed electrocardiogram, surface electromyography and electrodermal signals to identify baseline reference segments where each bioelectric signal is in a relatively stable position, and generate a set of baseline reference segments.

[0029] Furthermore, the preprocessed ECG, surface electromyography (EMG), and electrodermal (ED) signals are divided into continuous sliding time windows according to time sequence. Within each continuous sliding time window, the amplitude standard deviation of the ECG, EMG, and EED signals is calculated. Time windows with amplitude standard deviations lower than a preset stability threshold are identified as relatively stable baseline reference segments. All baseline reference segments of the ECG, EMG, and EED signals are summarized to generate a baseline reference segment set.

[0030] It should be noted that the stability threshold is the upper limit of the amplitude standard deviation obtained from the statistical analysis of bioelectrical signal data of historical healthy populations. It is set by calculating the high percentile (such as the 95th percentile) of the amplitude standard deviation of electrocardiogram, surface electromyography, and electrodermal signals of healthy populations in the resting state. An exemplary value range is 1.0 to 1.5 microvolts. If it is higher than 1.5 microvolts, too many effective baseline segments may be missed. If it is lower than 1.0 microvolts, noise fluctuations may be misjudged as effective baselines, resulting in a decrease in the representativeness of the baseline reference segment set.

[0031] S1.3 Based on the baseline reference fragment set, calculate the baseline offset of each bioelectric signal and generate a baseline correction parameter set.

[0032] Furthermore, the average amplitude of the central electromyography (ECG), surface electromyography (EMG), and electrodermal (EDA) signals of the baseline reference fragment set within all their respective baseline reference fragments is calculated as the corresponding baseline offset, and these are integrated to generate a set of baseline correction parameters.

[0033] S1.4. Baseline correction processing is performed on the preprocessed electrocardiogram, surface electromyography and electrodermal signals using the baseline correction parameter set to generate a multi-channel bioelectrical signal sequence.

[0034] Furthermore, the baseline offset of the corresponding channel in the baseline correction parameter set is read from the preprocessed ECG, surface myocardiogram and electrodermal signals respectively. Each sampling point of the preprocessed ECG, surface myocardiogram and electrodermal signals is traversed in chronological order. When traversing to each sampling point, the baseline offset of the corresponding channel is applied to the amplitude of the corresponding sampling point to complete the baseline offset compensation process, so as to obtain the baseline-corrected ECG, surface myocardiogram and electrodermal signals. The channels are then encapsulated in chronological order to generate a multi-channel bioelectrical signal sequence.

[0035] S2. Within a continuous sliding time window, rhythmic events are identified and extracted from multi-channel bioelectrical signal sequences, and the rhythmic events corresponding to each bioelectrical signal are correlated to generate a multi-source bioelectrical rhythmic event sequence.

[0036] S2.1 Based on multi-channel bioelectric signal sequences, the time segments of each bioelectric signal are divided to form a set of continuous sliding time windows.

[0037] Furthermore, based on the multi-channel bioelectric signal sequence, the electrocardiogram signal, surface electromyography signal, and skin electrodermal signal are synchronously extracted on the time axis according to a fixed window length and window step size (e.g., 5 seconds as a time segment length and 1 second as the moving interval between adjacent time segments), and the multi-channel bioelectric signal sequence is divided into a series of time segments that overlap and are arranged continuously in time, forming a set of continuous sliding time windows.

[0038] S2.2 Within a continuous sliding time window set, perform cardiac rhythm feature detection on the electrocardiogram signal, identify the time of cardiac occurrence, and generate a cardiac rhythm event sequence.

[0039] Furthermore, within the set of continuous sliding time windows, the amplitude changes of the electrocardiogram (ECG) signal are scanned point by point within each continuous sliding time window. The amplitude difference between adjacent sampling points is calculated to highlight the locations in the ECG signal with larger change rates. Local peak locations with rapid rising edges, peak amplitudes significantly higher than adjacent waveforms, and within the physiologically permissible heartbeat time interval are identified in the ECG signal as waveform locations that conform to heartbeat characteristics. The time points corresponding to the waveform locations that conform to heartbeat characteristics are determined as the heartbeat occurrence time. The heartbeat occurrence times identified within each continuous sliding time window are recorded in chronological order to form a heartbeat rhythm event sequence.

[0040] It should be noted that the physiologically permissible range of heartbeat intervals is determined by collecting electrocardiogram signal data from healthy individuals, statistically analyzing the concentrated intervals between adjacent heartbeats, and combining this with cardiac physiology constraints on the minimum and maximum reasonable heart rates. In practical applications, extreme interval values ​​caused by obvious abnormalities and noise are first eliminated, and then the range of heartbeat intervals covered by most healthy samples is taken as the physiologically permissible range. This ensures that the set heartbeat intervals not only conform to the characteristics of cardiac rhythm under normal autonomic nervous system regulation, but also effectively eliminate the interference of non-physiological fluctuations on the rhythmic event recognition results. In physiological and clinical monitoring, the minimum reasonable heart rate for adults is taken as 40 beats per minute, and the maximum reasonable heart rate is taken as 180 beats per minute.

[0041] S2.3 Within a continuous sliding time window set, identify the muscle activation state of the surface electromyography signal, extract the start and end times of muscle activation, and generate a muscle activation rhythm event sequence.

[0042] Furthermore, within the set of continuously sliding time windows, the amplitude of the surface electromyography (EMG) signal is calculated point by point according to the sampling order, and the absolute value of the EMG signal amplitude is accumulated within the continuously sliding time window to characterize the intensity of muscle activity. When the accumulated result rises continuously from a low level on the time axis and exceeds the background fluctuation range, the corresponding time point is determined as the start time of muscle activation. When the accumulated result falls back to the background fluctuation range, the corresponding time point is determined as the end time of muscle activation. The start and end times corresponding to each muscle activation are recorded in chronological order to generate a muscle activation rhythmic event sequence.

[0043] It should be noted that the background fluctuation range is a statistical interval of amplitude fluctuation calculated based on the surface electromyography (EMG) signal and the electrical skin signal within a time window when no muscle activation and electrical skin response occur, representing the signal fluctuation level under normal resting conditions.

[0044] S2.4 Within a continuous sliding time window set, identify changes in the skin electrical signal response, extract skin electrical response events, and generate a skin electrical response rhythm event sequence.

[0045] Furthermore, within the continuous sliding time window set, the conductivity change between adjacent sampling points of the EDS signal is calculated according to the sampling order, and the conductivity change is accumulated within the continuous sliding time window to characterize the intensity of the EDS signal change. When the EDS signal shows a continuous positive change in a short period of time and the accumulated conductivity change exceeds the background fluctuation range, the start time of the corresponding change process is determined as the time of occurrence of the EDS response event. All EDS response event times are summarized in chronological order to generate a sequence of EDS response rhythm events.

[0046] S2.5. Correlate the cardiac rhythm event sequence, muscle activation rhythm event sequence, and skin conductance response rhythm event sequence according to their time order and corresponding sliding time windows to generate a multi-source bioelectrical rhythm event sequence.

[0047] Furthermore, based on the occurrence time of each rhythmic event in the cardiac rhythmic event sequence, muscle activation rhythmic event sequence, and skin conductance response rhythmic event sequence, each rhythmic event is mapped to the corresponding time window in the continuous sliding time window set. Different types of rhythmic events are then uniformly organized according to the chronological order of their occurrence, so that cardiac rhythmic events, muscle activation rhythmic events, and skin conductance response rhythmic events within the same continuous sliding time window form a corresponding relationship, generating a multi-source bioelectrical rhythmic event sequence.

[0048] S3. Perform rhythm relationship analysis on multi-source bioelectric rhythm event sequences, identify the synchronization relationship between different bioelectric rhythms, and determine the phase correspondence and coupling stability between different bioelectric rhythms by analyzing the relative positional relationship of each rhythm event on the time axis, and generate a rhythm coupling state time series.

[0049] S3.1. Organize the cardiac rhythm event sequence, muscle activation rhythm event sequence, and skin conductance response rhythm event sequence according to the occurrence time of the rhythm events to form a multi-source rhythm event alignment set.

[0050] Furthermore, the occurrence times of all rhythmic events in the cardiac rhythmic event sequence, muscle activation rhythmic event sequence, and skin conductance response rhythmic event sequence are read and uniformly sorted according to chronological order. Rhythmic events within the same continuous sliding time window are grouped under the same time window index. All the sorted rhythmic events within the continuous sliding time window are collected to form a multi-source rhythmic event alignment set.

[0051] S3.2 Based on the multi-source rhythm event alignment set, identify the consistency of occurrence of different bioelectric rhythm events within the same continuous sliding time window, and generate a rhythm synchronization feature set.

[0052] Furthermore, for each continuous sliding time window in the multi-source rhythmic event alignment set, the occurrence frequency of cardiac rhythmic events, muscle activation rhythmic events, and skin conductance response rhythmic events within the time window is counted. When at least two types of rhythmic events exist within the same continuous sliding time window, it is determined that bioelectrical rhythmic events have synchronized within the window, and the continuous sliding time window identifier is recorded. The time interval between different bioelectrical rhythmic events is calculated as a quantitative feature of the synchronization relationship. The synchronization determination results and corresponding quantitative features of all continuous sliding time windows are summarized to generate a rhythmic synchronization feature set.

[0053] S3.3 Based on the rhythm synchronization feature set, extract the sequential relationship of different bioelectric rhythm events on the time axis to generate a rhythm sequence relationship feature set.

[0054] Furthermore, a continuous sliding time window with multiple bioelectrical rhythm events occurring in the rhythm synchronization feature set is selected. The specific occurrence times of cardiac rhythm events, muscle activation rhythm events, and skin conductance response rhythm events within each continuous sliding time window are compared. The chronological relationship of different bioelectrical rhythm events on the time axis is determined based on the order of their occurrence times (e.g., when the cardiac rhythm event occurs earlier than both the muscle activation rhythm event and the skin conductance response rhythm event, the cardiac rhythm event is determined to be in a preceding position on the time axis). The results are then summarized to generate a rhythm chronological relationship feature set.

[0055] S3.4 Perform relative position analysis on the set of rhythm sequence relationship features to determine the relative position relationship between different bioelectric rhythm events and generate a set of rhythm phase relationship features.

[0056] Furthermore, based on the rhythm sequence feature set, paired bioelectric rhythm events (e.g., cardiac rhythm events and muscle activation rhythm events) with a recorded sequence are processed to calculate the time interval between the occurrence times of the paired bioelectric rhythm events. Using the average period of the rhythm to which the earlier event in the paired bioelectric rhythm event belongs (e.g., the average RR interval of the cardiac rhythm) as the reference period, the standardized phase difference characterizing the relative positional relationship is obtained by calculating the ratio of the time interval to the reference period. The standardized phase differences of all paired bioelectric rhythm events are summarized to generate a rhythm phase relationship feature set.

[0057] S3.5. Based on the rhythm phase relationship feature set and the rhythm synchronization feature set, evaluate the changes in different bioelectric rhythm relationships, identify the fluctuation degree and continuous consistency of bioelectric rhythm relationships, and generate a rhythm coupling stability feature set.

[0058] Furthermore, the rhythm phase relationship feature set and the rhythm synchronization feature set are traversed in chronological order. Within each continuous sliding time window, the variance of the standardized phase difference of specific paired bioelectric rhythm events is calculated as a quantitative indicator of the degree of fluctuation in the bioelectric rhythm relationship. The number of times bioelectric rhythm events occur consistently within adjacent continuous sliding time windows is counted and compared to determine whether the consistency pattern is continuous and stable (if at least two types of bioelectric rhythm events are recorded in both a certain continuous sliding time window and the adjacent previous continuous sliding time window, the consistency pattern is considered continuous and stable between these adjacent windows; if no at least two types of bioelectric rhythm events are recorded in one of the adjacent windows, the consistency pattern is considered discontinuous and unstable between these adjacent windows). The quantitative indicator of the degree of fluctuation in the bioelectric rhythm relationship and the judgment result of the continuity and stability of the consistency of bioelectric rhythm events within each continuous sliding time window are integrated to generate a rhythm coupling stability feature set.

[0059] The expression for the quantitative index used to calculate the degree of fluctuation in bioelectrical rhythm relationships is as follows: ; in, It is a quantitative index of the degree of fluctuation in bioelectrical rhythm relationships, used to characterize the degree of fluctuation and consistency of different bioelectrical rhythm relationships within a continuous sliding time window; It represents the number of rhythmic event pairs participating in the statistics within the current continuous sliding time window; It is an index of rhythmic events; It is the first Each rhythmic event corresponds to a value of the rhythmic phase relationship feature; It is the average value of all rhythm phase relationship characteristics within the current continuous sliding time window; It is the first Each rhythmic event takes values ​​for the corresponding rhythmic synchronization feature; It is the average value of all rhythm synchronization features within the current continuous sliding time window.

[0060] S3.6 Integrate the rhythm synchronization feature set, the rhythm phase relationship feature set, and the rhythm coupling stability feature set to generate a rhythm coupling state time series.

[0061] Furthermore, according to the time sequence of the continuous sliding time window, the consistency judgment results and synchronization time interval characteristics of the occurrence of bioelectric rhythm events in the continuous sliding time window are extracted from the rhythm synchronization feature set. The standardized phase difference of each pair of bioelectric rhythm events in the corresponding continuous sliding time window is extracted from the rhythm phase relationship feature set. The quantitative index of the fluctuation degree of bioelectric rhythm relationship and the continuous stability judgment results of the occurrence of bioelectric rhythm events are extracted from the rhythm coupling stability feature set. These are then combined and processed to generate a rhythm coupling state time series.

[0062] S4. Establish a baseline for rhythm coupling detection under normal physiological conditions and compare it with the rhythm coupling state corresponding to the current window in the rhythm coupling state time series to generate metabolic abnormality detection results.

[0063] S4.1. Based on the preset rhythm coupling fluctuation threshold, the stability of the rhythm coupling state time series is screened to identify the time periods in a stable physiological detection state and generate a stable rhythm coupling state dataset.

[0064] Furthermore, the time series of rhythm coupling states is traversed, and the quantitative index of the fluctuation degree of bioelectric rhythm relationship in each continuous sliding time window record is extracted and compared with the preset rhythm coupling fluctuation threshold. At the same time, it is checked whether the continuous stability judgment result of the occurrence of bioelectric rhythm events corresponding to the continuous sliding time window is stable. Continuous sliding time windows with a quantitative index of the fluctuation degree of bioelectric rhythm relationship below the rhythm coupling fluctuation threshold and a continuous stability judgment result of the occurrence of bioelectric rhythm events are marked. In the rhythm coupling state time series, the continuously marked continuous sliding time windows are identified as a time period in a stable physiological detection state, and the rhythm coupling state records corresponding to all time periods in a stable physiological detection state are summarized to generate a stable rhythm coupling state dataset.

[0065] It should be noted that the rhythm coupling fluctuation threshold is set by collecting multi-source bioelectrical signals from a large number of healthy individuals, calculating the distribution of their rhythm coupling deviation, and then taking the high percentile (such as the 95th percentile). An exemplary value range is 1.5 to 2.0. If it is higher than 2.0, the detection sensitivity will be reduced, causing early metabolic abnormalities to be missed. If it is lower than 1.5, the false alarm rate will be increased, and normal physiological fluctuations will be easily misjudged as abnormalities, affecting the accuracy of the early warning.

[0066] S4.2 Statistically summarize the rhythm coupling states in the stable rhythm coupling state dataset to construct a rhythm coupling detection baseline.

[0067] Furthermore, based on each rhythm coupling state record in the stable rhythm coupling state dataset, the synchronization time interval features, standardized phase difference, and fluctuation degree quantification indicators of cardiac rhythm events with muscle activation rhythm events and skin conductance response rhythm events are extracted. The mean and standard deviation of each type of feature data are calculated as statistical benchmarks to characterize the rhythm coupling relationship under normal physiological conditions of an individual, and integrated to construct a rhythm coupling detection baseline.

[0068] It should be noted that the rhythm coupling detection baseline is a quantitative reference standard characterizing the rhythm coupling relationship between an individual's heartbeat, muscle activation, and skin electrical response under normal physiological conditions. It serves as a benchmark for real-time rhythm coupling state comparison to identify whether the current physiological state deviates from the normal pattern.

[0069] S4.3 Extract the rhythm coupling state corresponding to the current continuous sliding time window from the rhythm coupling state time series, compare it with the rhythm coupling detection baseline, and calculate the degree of rhythm coupling deviation.

[0070] Furthermore, in the rhythm coupling state time series, based on the position index of the current continuous sliding time window on the time axis, the rhythm coupling state recorded within the corresponding continuous sliding time window is located and extracted. At the same time, the rhythm coupling state is compared item by item with the corresponding statistical benchmark in the rhythm coupling detection baseline, the deviation is calculated, and the rhythm coupling deviation degree, which characterizes the degree to which the current rhythm coupling relationship deviates from the normal physiological state, is obtained.

[0071] The expression for calculating the degree of rhythm coupling deviation is: ; in, It is the degree of rhythm coupling deviation, used to characterize the overall level of deviation of the rhythm coupling state corresponding to the current continuous sliding time window from the rhythm coupling detection baseline; It is a feature index, used to traverse the various comparative features in the rhythm coupling state; It is the total number of features included in the comparison; It is the first in the rhythm coupling state corresponding to the current continuous sliding time window. Each feature term takes a value; It is the rhythm coupling detection baseline with the first The statistical baseline center value corresponding to each characteristic term; It is the rhythm coupling detection baseline with the first The statistical discrete measure corresponding to each feature term.

[0072] S4.4 Based on the degree of deviation of rhythm coupling, the abnormality of the rhythm coupling state corresponding to the current continuous sliding time window is quantitatively evaluated, and metabolic abnormality detection results are generated.

[0073] Furthermore, the degree of rhythm coupling deviation is compared with a preset abnormality level grading threshold. When the degree of rhythm coupling deviation is greater than the abnormality level grading threshold, the current rhythm coupling state is determined to be "severely abnormal". When the degree of rhythm coupling deviation is within the abnormality level grading threshold, the current rhythm coupling state is determined to be "mildly abnormal". When the degree of rhythm coupling deviation is less than the abnormality level grading threshold, the current rhythm coupling state is determined to be "normal", and metabolic abnormality detection results are generated.

[0074] It should be noted that the abnormality level grading threshold is set by calculating the statistical percentile of the degree of rhythm coupling deviation of historical healthy people in the resting state, taking the high percentile (such as the 95th percentile). An exemplary value range is 1.5 to 2.0. If it is higher than 2.0, it will reduce the detection sensitivity and cause early metabolic abnormalities to be missed. If it is lower than 1.5, it will increase the false alarm rate and misjudge normal physiological fluctuations as abnormalities.

[0075] S5. Based on the results of metabolic abnormality detection, determine the risk of diabetes metabolic abnormality in the tested subjects, and output the detection conclusion of diabetes metabolic abnormality when the risk of diabetes metabolic abnormality meets the preset warning trigger conditions.

[0076] S5.1 Based on the results of metabolic abnormality detection, analyze the direction and duration of change of the degree of metabolic abnormality within a continuous sliding time window, and generate metabolic abnormality risk evolution characteristics.

[0077] Furthermore, the judgment results of adjacent consecutive sliding time windows in the metabolic abnormality detection results are compared. If the abnormality level of the judgment result of the later consecutive sliding time window is higher than that of the previous consecutive sliding time window, a risk increase change direction is recorded; otherwise, a risk decrease change direction is recorded. If the abnormality levels are the same, it is recorded as a stable direction. The number of consecutive sliding time windows with the same change direction (such as consecutive risk increase and consecutive risk decrease) is counted as a measure of the duration of change. Combining the change direction and the duration of change, a metabolic abnormality risk evolution characteristic is generated.

[0078] S5.2 Based on the evolution characteristics of metabolic abnormality risk, classify the current degree of metabolic abnormality of the tested subjects into risk levels and generate metabolic abnormality detection level results.

[0079] Furthermore, the direction of change (continuous risk increase, continuous risk decrease, and stabilization) and the corresponding number of continuous sliding time windows are read from the metabolic abnormality risk evolution characteristics, and combined with the latest metabolic abnormality detection results of the continuous sliding time windows; if the direction of change is continuous risk increase and the number of continuous sliding time windows reaches the threshold of the number of continuous windows required for risk level adjustment (e.g., 8 to 12 continuous time windows), then the abnormality level corresponding to the current judgment result is adjusted up one level (e.g., from "medium risk" corresponding to "mild abnormality" to "high risk"); if the direction of change is continuous risk decrease and the number of continuous sliding time windows reaches the threshold of the number of continuous windows required for risk level adjustment, then the abnormality level is adjusted down one level; if the direction of change is stabilization, then the abnormality level corresponding to the current judgment result is maintained, and a metabolic abnormality detection level result is generated.

[0080] S5.3 Based on the results of metabolic abnormality detection, a comprehensive judgment is made on the changes in metabolic abnormality risk within a continuous sliding time window to generate a metabolic abnormality risk confirmation result.

[0081] Furthermore, based on the metabolic abnormality detection level results, the level results of consecutive sliding time windows are read in chronological order. If the level results of consecutive sliding time windows are all "high risk" and the direction of change is continuous risk increase or stabilization, a metabolic abnormality risk confirmation result of "high risk confirmed" is generated. If the level results of consecutive sliding time windows are all "medium risk" and the direction of change is stable, a metabolic abnormality risk confirmation result of "medium risk confirmed" is generated. If the level results of consecutive sliding time windows are all "low risk" and the direction of change is continuous risk decrease or stabilization, a metabolic abnormality risk confirmation result of "low risk confirmed" is generated. If the level results of consecutive sliding time windows are inconsistent or the direction of change does not conform to the above pattern, a metabolic abnormality risk confirmation result of "risk not confirmed" is generated.

[0082] S5.4. Call the preset warning trigger conditions to trigger the determination of the metabolic abnormality risk confirmation result, and output the diabetes metabolic abnormality detection conclusion when the warning trigger conditions are met.

[0083] Furthermore, the metabolic abnormality risk confirmation result is read. When the metabolic abnormality risk confirmation result is "high risk confirmed", the warning trigger condition is determined to be met. When the metabolic abnormality risk confirmation result is "medium risk confirmed" and the number of continuous sliding time windows reaches the threshold of the number of continuous windows required for warning triggering, the warning trigger condition is determined to be met. When the metabolic abnormality risk confirmation result is "low risk confirmed" or "risk not confirmed", the warning trigger condition is determined not to be met. When the warning trigger condition is met, the diabetes metabolic abnormality detection conclusion containing the current risk level and warning status is output.

[0084] It should be noted that the process for setting the warning trigger conditions is as follows: Multi-source bioelectrical rhythm data of diagnosed early-stage diabetes patients and healthy controls from historical populations are collected. The metabolic abnormality risk confirmation results for both groups are calculated. By analyzing the differences in the distribution of metabolic abnormality risk confirmation results between the two groups, "high-risk confirmation" is set as one of the trigger conditions. Simultaneously, the duration pattern of the "medium-risk confirmation" state in early-stage patients is observed, and the distribution of the duration window number is statistically analyzed. The minimum duration window number that can effectively distinguish early-stage patients from healthy controls is set as the threshold for the duration window number required to trigger "medium-risk confirmation." "Low-risk confirmation" and "risk not confirmed" are set as non-trigger conditions. An exemplary range for the duration window number threshold required for warning triggering is 8 to 12 consecutive sliding time windows. A value higher than 12 consecutive sliding time windows will delay the warning opportunity and may miss the early intervention window. A value lower than 8 consecutive sliding time windows will increase the false positive rate of the warning, leading to unnecessary alerts.

[0085] In summary, this invention achieves rhythmic expression of multi-source bioelectrical information under a unified time structure by: identifying and extracting rhythmic events from multi-channel bioelectrical signal sequences within a continuous sliding time window to generate multi-source bioelectrical rhythmic event sequences, thereby improving the basic quality and reliability of the temporal features upon which subsequent analysis depends; and by generating rhythmic coupling state time series, it achieves advanced signal pattern recognition of dynamic coordination modes between different physiological functions, enhancing the physiological specificity and application value of metabolic abnormality identification.

[0086] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals, characterized in that: include, The electrocardiogram (ECG), surface electromyography (EMG), and electrodermal (EDS) signals of the subjects were collected, preprocessed, and baseline correction was performed to generate multi-channel bioelectrical signal sequences. Within a continuous sliding time window, rhythmic events are identified and extracted from multi-channel bioelectrical signal sequences, and the rhythmic events corresponding to each bioelectrical signal are correlated to generate a multi-source bioelectrical rhythmic event sequence. Rhythm relationship analysis is performed on multi-source bioelectric rhythm event sequences to identify the synchronization relationship between different bioelectric rhythms. By analyzing the relative positional relationship of each rhythm event on the time axis, the phase correspondence and coupling stability between different bioelectric rhythms are determined, and a time series of rhythm coupling states is generated. Establish a baseline for rhythm coupling detection under normal physiological conditions and compare it with the rhythm coupling state corresponding to the current window in the time series of rhythm coupling state to generate metabolic abnormality detection results; Based on the results of metabolic abnormality detection, the risk of diabetes metabolic abnormality is determined in the subjects, and the detection conclusion of diabetes metabolic abnormality is output when the risk of diabetes metabolic abnormality meets the preset warning trigger conditions.

2. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The steps for generating the multi-channel bioelectric signal sequence are as follows: The preprocessing includes time synchronization and filtering / denoising. Baseline analysis was performed on the preprocessed electrocardiogram, surface electromyography and electrodermal signals to identify relatively stable baseline reference segments for each bioelectric signal and generate a set of baseline reference segments. Based on the baseline reference fragment set, the baseline offset of each bioelectrical signal is calculated, and a baseline correction parameter set is generated; Baseline correction processing was performed on the preprocessed electrocardiogram, surface electromyography, and electrodermal signals using a set of baseline correction parameters to generate multi-channel bioelectrical signal sequences.

3. The method for early warning of diabetic metabolic abnormalities based on bioelectrical signals as described in claim 1, characterized in that: The steps for identifying and extracting rhythmic events from multi-channel bioelectrical signal sequences within a continuous sliding time window are as follows: Based on multi-channel bioelectric signal sequences, each bioelectric signal is divided into time segments to form a set of continuous sliding time windows; Within a continuous sliding time window set, cardiac rhythm feature detection is performed on the electrocardiogram signal to identify the time of cardiac occurrence and generate a cardiac rhythm event sequence; Within a continuous sliding time window set, muscle activation state is identified from surface electromyography signals, the start and end times of muscle activation are extracted, and a muscle activation rhythm event sequence is generated. Within a continuously sliding time window set, changes in the skin electrical signal are identified, skin electrical response events are extracted, and a skin electrical response rhythm event sequence is generated.

4. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The multi-source bioelectric rhythm event sequence is generated by associating cardiac rhythm event sequences, muscle activation rhythm event sequences, and skin conductance response rhythm event sequences according to their time sequence and corresponding sliding time windows.

5. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The steps for analyzing the rhythmic relationships of multi-source bioelectrical rhythmic event sequences and identifying the synchronization relationships between different bioelectrical rhythms are as follows. The sequences of cardiac rhythm events, muscle activation rhythm events, and skin conductance response rhythm events are organized according to the time of occurrence of rhythm events to form a multi-source rhythm event alignment set; Based on the multi-source rhythm event alignment set, the consistency of occurrence of different bioelectric rhythm events within the same continuous sliding time window is identified, and a rhythm synchronization feature set is generated.

6. The method for early warning of diabetic metabolic abnormalities based on bioelectrical signals as described in claim 1, characterized in that: The process involves analyzing the relative positions of various rhythmic events on the time axis to determine the phase correspondence and coupling stability between different bioelectrical rhythms, generating a time series of rhythmic coupling states. The steps are as follows: Based on the rhythm synchronization feature set, the sequential relationship of different bioelectric rhythm events on the time axis is extracted to generate a rhythm sequence relationship feature set; By performing relative position analysis on the set of rhythm sequence relationship features, the relative position relationship between different bioelectric rhythm events is determined, and a set of rhythm phase relationship features is generated. Based on the rhythm phase relationship feature set and the rhythm synchronization feature set, the changes in different bioelectric rhythm relationships are evaluated, the fluctuation degree and continuous consistency of bioelectric rhythm relationships are identified, and a rhythm coupling stability feature set is generated. By integrating the rhythm synchronization feature set, the rhythm phase relationship feature set, and the rhythm coupling stability feature set, a rhythm coupling state time series is generated.

7. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The steps for establishing a rhythm coupling detection baseline under normal physiological conditions are as follows: Based on a preset rhythm coupling fluctuation threshold, the time series of rhythm coupling states are screened for stability, the time periods in a stable physiological detection state are identified, and a stable rhythm coupling state dataset is generated. The rhythm coupling states in the stable rhythm coupling state dataset are statistically summarized to construct a rhythm coupling detection baseline.

8. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The metabolic abnormality detection results are generated by comparing and analyzing the rhythm coupling state corresponding to the current window in the rhythm coupling state time series. The steps are as follows: Extract the rhythm coupling state corresponding to the current continuous sliding time window from the rhythm coupling state time series, compare it with the rhythm coupling detection baseline, and calculate the degree of rhythm coupling deviation; Based on the degree of deviation of rhythm coupling, the abnormality of the rhythm coupling state corresponding to the current continuous sliding time window is quantitatively evaluated, and metabolic abnormality detection results are generated.

9. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The steps for determining the risk of diabetic metabolic disorders in test subjects based on metabolic abnormality detection results are as follows. Based on the results of metabolic abnormality detection, the direction and duration of change of the degree of metabolic abnormality within a continuous sliding time window are analyzed to generate metabolic abnormality risk evolution characteristics. Based on the evolutionary characteristics of metabolic abnormality risk, the current degree of metabolic abnormality of the tested subjects is classified into risk levels, and metabolic abnormality detection level results are generated.

10. The method for early warning of metabolic abnormalities in diabetes based on bioelectrical signals as described in claim 1, characterized in that: The steps for outputting a diabetes metabolic abnormality detection conclusion when the risk of diabetes metabolic abnormality meets the preset warning trigger conditions are as follows: Based on the results of metabolic abnormality detection levels, a comprehensive judgment is made on the changes in metabolic abnormality risk within a continuous sliding time window to generate a metabolic abnormality risk confirmation result. The system invokes preset warning trigger conditions to determine the results of metabolic abnormality risk confirmation, and outputs the detection conclusion of diabetes metabolic abnormality when the warning trigger conditions are met.

Images