Noninvasive troponin monitoring system

Abstract

The invention discloses a non-invasive troponin monitoring system, which comprises a signal acquisition module, a filtering processing module, a feature extraction module, a clustering result generation module, a clustering result generation module, a clustering result generation module, a clustering result generation module, a clustering result generation module, a clustering result generation module and a clustering result generation module, wherein the signal acquisition module collects bio-electricity signals and the like to generate a data set, and the filtering processing module calculates a signal-to-noise ratio and adjusts filter bandwidth to output filtering signals. The data fusion module synchronizes the electrocardiogram period to integrate the feature vector, and the calibration output module monitors the environment change and generates a calibrated feature vector. According to the non-invasive troponin monitoring system, an original data set is generated by collecting multi-dimensional signals of a human body and environmental parameters. A dynamic filtering technology is utilized to optimize a signal-to-noise ratio, accurately extract time domain and frequency domain features and perform clustering analysis, an electrocardiogram period is synchronously analyzed to realize data fusion, multi-dimensional feature vectors are generated, calibration is performed in combination with environment monitoring data, interference is effectively eliminated, and high precision and stability of a troponin monitoring result are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of troponin monitoring, and in particular to a non-invasive troponin monitoring system. Background Art

[0002] The field of troponin monitoring technology encompasses in vitro diagnostic methods for detecting cardiomyocyte-specific protein markers in the blood. The core content of this field covers immunoassay technology, miniaturized detection equipment, trace blood sample processing, rapid detection reagents, precise quantitative detection, clinical result interpretation, and disease risk assessment. This technology has developed into a diverse and systematic application system. Among them, the non-invasive troponin monitoring system refers to a technical solution for detecting human troponin levels in a non-invasive manner. The system includes optical detection methods, skin surface signal capture, biosensors, signal conversion and processing, and data output modules. The specific content of its participation is to use optical principles or biosensing methods to obtain human body surface signals, convert the signals into electrical signals through sensors, and then amplify and filter them through circuits, and finally calculate the troponin concentration value through a data processing unit and display the output.

[0003] Existing technologies have many limitations in the troponin monitoring process. First, the data acquisition dimension is limited, only blood samples or single skin surface signals are collected, and multi-dimensional information such as electrocardiogram waveforms, respiratory waveforms and environmental parameters are ignored, resulting in single data and difficulty in fully reflecting the impact of human physiological state on troponin levels. Second, the filtering process lacks adaptability and uses fixed filter parameters, which makes it difficult to cope with complex and changeable signal characteristics and noise environments, causing some troponin signal features to be lost or interfered with by noise. Third, feature extraction is not in-depth and comprehensive enough, only limited time domain or frequency domain features are extracted, and the correlation between feature parameters is not fully explored, making it impossible to fully characterize the troponin signal characteristics. In addition, data fusion is not synchronized with the electrocardiogram cycle analysis, resulting in inaccurate and incomplete feature extraction. Finally, the calibration link ignores compensation for changes in environmental parameters, causing the monitoring results to be affected by environmental interference and affect the accuracy. Summary of the Invention

[0004] The main purpose of the present invention is to provide a non-invasive troponin monitoring system that can effectively solve the limitations of existing technologies in troponin monitoring, such as insufficient multidimensional data collection, filtering that is not suitable for complex environments, insufficient feature extraction, asynchronous data fusion, and lack of environmental compensation in calibration. These problems result in monitoring results being affected by environmental interference, inaccurate feature extraction, and difficulty in fully reflecting troponin levels.

[0005] To achieve the above-mentioned object, the technical solution adopted by the present invention is: a non-invasive troponin monitoring system, the system comprising: The signal acquisition module collects bioelectric signals from the human skin surface, obtains electrocardiogram waveforms and respiratory waveforms, measures ambient temperature and humidity, and electrode contact pressure, and generates a raw signal data set; The filtering processing module analyzes the original signal data set to calculate the signal-to-noise ratio, dynamically adjusts the filter bandwidth after comparing it with a preset threshold, and generates a filtered signal; The feature extraction module extracts the time domain peak, rise time and frequency domain main frequency of the filtered signal, calculates the correlation coefficient between the feature parameters, and generates a feature parameter clustering result; The data fusion module synchronously analyzes the electrocardiogram cycle, extracts the characteristic parameters from the characteristic parameter clustering results, and integrates them to generate a multi-dimensional fusion feature vector; The calibration output module monitors changes in environmental parameters, calculates compensation values ​​and maps them to the multi-dimensional fusion feature vector to generate a calibrated feature vector.

[0006] Preferably, the original signal data set includes bioelectric signals, electrocardiogram waveforms, respiratory waveforms, and environmental parameters; the filtered signals include filtered troponin signals, signal-to-noise ratio data, and filter bandwidth parameters; the feature parameter clustering results include time domain feature parameters, frequency domain feature parameters, and feature correlation coefficients; the multi-dimensional fusion feature vector includes electrocardiogram cycle features, fusion feature vectors, and feature weight coefficients; and the calibrated feature vector includes environmental parameter compensation values, calibrated feature vectors, and feature calibration models.

[0007] Preferably, the signal acquisition module includes a signal acquisition submodule, a waveform acquisition submodule, and an environment monitoring submodule; The signal acquisition submodule collects bioelectric signals from the surface of human skin and generates bioelectric signal data; The waveform acquisition submodule acquires the electrocardiogram waveform and the respiratory waveform, and generates the electrocardiogram waveform data and the respiratory waveform data; The environmental monitoring submodule measures ambient temperature, humidity and electrode contact pressure to generate environmental parameter data.

[0008] Preferably, the filtering processing module includes a signal-to-noise ratio calculation submodule and a filter adjustment submodule; The signal-to-noise ratio calculation submodule analyzes the original signal data set to calculate the signal-to-noise ratio and generate signal-to-noise ratio data; The filter adjustment submodule dynamically adjusts the filter bandwidth after comparing with the preset signal-to-noise ratio threshold to generate a filtered signal.

[0009] Preferably, the feature extraction module includes a time domain feature extraction submodule and a frequency domain feature extraction submodule; The time domain feature extraction submodule extracts the time domain peak and rise time of the filtered signal and generates time domain feature parameters; The frequency domain feature extraction submodule extracts the frequency domain main frequency and power spectrum density, calculates the correlation coefficient between feature parameters, and generates feature parameter clustering results.

[0010] Preferably, the characteristic parameter clustering result adopts the formula: ; in, Representative Time domain characteristic parameters, Representative frequency domain characteristic parameters, Represents the clustering results of feature parameters.

[0011] Preferably, the data fusion module includes a periodic synchronization analysis submodule and a feature fusion submodule; The cycle synchronization analysis submodule synchronously analyzes the ECG cycle and generates ECG cycle features.

[0012] The feature fusion submodule extracts the feature parameters within each cycle and integrates them to generate a multi-dimensional fusion feature vector.

[0013] Preferably, the calibration output module includes an environmental monitoring submodule and a feature calibration submodule; Environmental monitoring submodule monitors changes in environmental parameters and generates compensation values ​​for these parameters; The feature calibration submodule calculates the compensation value and maps it to the fused feature vector to generate the calibrated feature vector.

[0014] Compared with the prior art, the present invention has the following beneficial effects: By comprehensively collecting bioelectric signals from the human skin surface, electrocardiogram waveforms, respiratory waveforms, and environmental parameters, a raw signal dataset containing rich information is generated, providing a solid foundation for subsequent precise processing. During the filtering process, the signal-to-noise ratio is calculated in real time and the filter bandwidth is dynamically adjusted to effectively preserve the characteristics of the troponin signal, improving signal purity and key information retention. The feature extraction process not only obtains the time domain peak and rise time, but also deeply extracts features such as the main frequency in the frequency domain, calculates the correlation coefficient between feature parameters, and generates feature parameter clustering results, achieving a precise multi-dimensional characterization of the troponin signal. During data fusion, the electrocardiogram cycle is synchronously analyzed and feature parameters are extracted, integrated to form a multi-dimensional fused feature vector, which greatly restores the characteristics of troponin level changes. Finally, by monitoring changes in environmental parameters, compensation values ​​are calculated and mapped to the fused feature vector to generate a calibrated feature vector, effectively eliminating interference from environmental factors and ensuring accurate and reliable monitoring results. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a flow chart of signal processing and feature analysis calibration of the present invention; Figure 2 This is a diagram of the sub-module parallel process architecture of the present invention. DETAILED DESCRIPTION

[0016] In order to make the technical means, creative features, objectives and effects achieved by the present invention easier to understand, the present invention is further described below in conjunction with specific implementation methods.

[0017] See also Figure 1 The present invention provides a technical solution: a non-invasive troponin monitoring system, the system comprising: The signal acquisition module collects bioelectric signals from the human skin surface, obtains electrocardiogram waveforms and respiratory waveforms, measures ambient temperature and humidity, and electrode contact pressure, and generates a raw signal data set. The raw signal data set includes bioelectric signals, electrocardiogram waveforms, respiratory waveforms, and environmental parameters. The filtering processing module analyzes the original signal data set to calculate the signal-to-noise ratio, dynamically adjusts the filter bandwidth after comparing it with the preset threshold, and generates a filtered signal. The filtered signal includes the filtered troponin signal, the signal-to-noise ratio data, and the filter bandwidth parameters; The feature extraction module extracts the time domain peak, rise time and frequency domain main frequency of the filtered signal, calculates the correlation coefficient between the feature parameters, and generates the feature parameter clustering results. The feature parameter clustering results include time domain feature parameters, frequency domain feature parameters, and feature correlation coefficients. The data fusion module synchronously analyzes the ECG cycle, extracts the characteristic parameters within each cycle, and integrates them to generate a multi-dimensional fusion feature vector. The multi-dimensional fusion feature vector includes the ECG cycle characteristics, the fusion feature vector, and the feature weight coefficient; The calibration output module monitors the changes in environmental parameters, calculates compensation values ​​and maps them to the fused feature vector, and generates a calibrated feature vector. The calibrated feature vector includes the environmental parameter compensation value, the calibrated feature vector, and the feature calibration model.

[0018] See also Figure 2 ,The signal acquisition module includes a signal acquisition submodule, a waveform acquisition submodule, and an ,environmental monitoring submodule; The signal acquisition submodule uses a high-precision bioelectrical signal sensor attached to the surface of the human chest skin. The electrode spacing of the sensor is set to the standard 5 cm. The sensor collects bioelectrical signals at a sampling rate of 250Hz and generates discrete bioelectrical signal data points. For example, collecting a signal for 1 second can obtain 250 data points, and the amplitude of the bioelectric signal ranges from 0 to 5 millivolts.

[0019] The waveform acquisition submodule synchronously collects the electrocardiogram waveform and respiratory waveform. The electrocardiogram uses a standard lead connection method and converts the analog signal into a digital signal through an analog-to-digital converter. The heart rate range of the acquired electrocardiogram waveform data is 60 to 100 beats per minute, and the frequency range of the respiratory waveform is 12 to 20 breaths per minute. For example, in an adult in a calm state, the electrocardiogram shows a heart rate of approximately 75 beats / minute, and the respiratory waveform shows a respiratory rate of approximately 16 beats / minute. The identification of the P wave, QRS complex, and T wave of the electrocardiogram waveform is determined by the amplitude and time width characteristics. The P wave amplitude is usually 0.05 to 0.25 millivolts, and the QRS complex amplitude is 0.5 to 2.5 millivolts.

[0020] The environmental monitoring submodule uses environmental sensors to monitor ambient temperature, humidity, and electrode contact pressure. The monitoring range of ambient temperature is 10 to 40 degrees Celsius, the humidity range is 20% to 90% relative humidity, and the electrode contact pressure range is 0 to 50 kPa. For example, in an indoor environment, the ambient temperature is 25 degrees Celsius, the humidity is 45% relative humidity, and the electrode contact pressure is 10 kPa. The environmental sensor sends data to the data processing unit via wireless transmission.

[0021] The data collected by each submodule is converted to digital form and stored in the data buffer to generate a raw signal data set. The raw signal data set is stored in a time series format, with each data point containing a timestamp, biopotential signal amplitude, electrocardiogram waveform data, respiratory waveform data, ambient temperature, humidity, and electrode contact pressure. For example, the data recorded at a certain moment is: timestamp 12:34:56.789, bioelectric signal amplitude 2.3 mV, electrocardiogram waveform data 78 times / minute, respiratory waveform data 16 times / minute, ambient temperature 25 degrees Celsius, humidity 45% relative humidity, and electrode contact pressure 10 kPa.

[0022] Through the above acquisition and storage process, the original signal data set is finally generated. The original signal data set includes bioelectric signals, electrocardiogram waveforms, respiratory waveforms, and environmental parameters, providing a data basis for the subsequent filtering processing module.

[0023] See also Figure 2 ,The filtering processing module includes a signal-to-noise ratio calculation submodule and a filter adjustment submodule; The signal-to-noise ratio calculation submodule receives the original signal data set and first extracts the signal amplitude and noise amplitude in the bioelectric signal data. The signal amplitude is obtained by calculating the average amplitude of 10 consecutive data points; Assuming that the amplitudes of 10 consecutive data points are 2.1mV, 2.3mV, 2.2mV, 2.4mV, 2.3mV, 2.2mV, 2.1mV, 2.0mV, 2.1mV, and 2.2mV respectively, the signal amplitude is (2.1+2.3+2.2+2.4+2.3+2.2+2.1+2.0+2.1+2.2) / 10=2.2mV. The noise amplitude is obtained by calculating the average of the absolute values ​​of the differences between adjacent data points. The adjacent differences are 0.2, -0.1, 0.2, -0.1, If the signal amplitude is -0.1, -0.1, -0.1, 0.1, and 0.1, the noise amplitude is (0.2 + 0.1 + 0.2 + 0.1 + 0.1 + 0.1 + 0.1 + 0.1 + 0.1) / 9 ≈ 0.122 mV. The signal-to-noise ratio (SNR) is calculated as follows: SNR = 20 × log10 (signal amplitude / noise amplitude) = 20 × log10 (2.2 / 0.122) ≈ 20 × 1.907 ≈ 38.14 dB. The preset SNR threshold is 35 dB. The current SNR of 38.14 dB is greater than the preset threshold of 35 dB.

[0024] The filter adjustment submodule dynamically adjusts the filter bandwidth. The initial filter bandwidth is 10 Hz. Based on the amplitude of the signal-to-noise ratio exceeding the threshold (38.14-35=3.14 dB), the filter bandwidth is adjusted by 0.5 Hz for every 1 dB exceeding the threshold. The adjusted filter bandwidth is 10 + 3.14 × 0.5 ≈ 11.57 Hz. The original signal is filtered using the adjusted filter. The amplitude range of the filtered troponin signal is reduced to 0.5 to 1.5 mV, the signal-to-noise ratio data is updated to 38.14 dB, and the filter bandwidth parameter is updated to 11.57 Hz. The filtered signal is generated, which includes the filtered troponin signal, signal-to-noise ratio data, and filter bandwidth parameter.

[0025] See also Figure 2 ,The feature extraction module includes a time domain feature extraction submodule and a frequency domain feature extraction submodule; The time-domain feature extraction submodule receives the filtered signal and extracts the time-domain peak by detecting the local maximum of the signal amplitude. The detection window is set to 5 consecutive data points. In a certain section of the filtered signal, the data point amplitudes are 0.8mV, 0.9mV, 1.0mV, 0.9mV, and 0.8mV, respectively. The middle data point of 1.0mV is identified as the time-domain peak. The rise time is extracted, and the time difference between the signal amplitude rising from the baseline (0.5mV) to the peak (1.0mV) is calculated. Assuming the rise spans 4 data points and the sampling rate is 250 Hz, the rise time is 4 / 250 = 0.016 seconds.

[0026] The frequency domain feature extraction submodule detects peaks and troughs from the filtered troponin signal and determines the time domain peak value between each peak and trough ( ) and rise time, perform fast Fourier transform (FFT) on the filtered signal, calculate the main frequency and power spectrum density of the spectrum graph, and determine the frequency domain characteristic parameters ( ), calculate the correlation between the time domain characteristic parameters and the frequency domain characteristic parameters, and use the formula for calculation: ; For example, suppose that in a set of parameters, the first time domain characteristic parameter , the first frequency domain characteristic parameter , substitute into the formula to calculate: ; This result shows that the characteristic parameter clustering results achieved the highest correlation.

[0027] in, Representative A time domain characteristic parameter, specifically the time domain peak value between the peak and the trough; Representative Frequency domain characteristic parameters, specifically the main frequency and power spectrum density, Represents the characteristic parameter clustering result, which is used to measure the correlation between the time domain and frequency domain characteristic parameters, and the sum symbol Indicates the accumulation of all characteristic parameters, square root symbol Indicates taking the square root of the accumulated square value.

[0028] See also Figure 2 ,The data fusion module includes a periodic synchronization analysis submodule and a ,feature fusion submodule; The cycle synchronization analysis submodule receives ECG waveform data and identifies the R-wave apex of the ECG waveform as the cycle starting point. An ECG cycle is defined as the period between two consecutive R-wave apexes. The duration of an ECG cycle ranges from 0.6 to 1.2 seconds and varies dynamically with heart rate. For example, at a heart rate of 75 beats per minute, the ECG cycle duration is 0.8 seconds. R-wave apex identification is achieved using an amplitude threshold method, which is set to 70% of the average amplitude of the troponin signal.

[0029] The feature fusion submodule extracts time-domain and frequency-domain feature parameters within each ECG cycle and performs weighted fusion according to preset feature weight coefficients. The feature weight coefficients are set based on the sensitivity of the features to troponin concentration, with a time-domain feature weight of 0.6 and a frequency-domain feature weight of 0.4. The fusion formula is: fused feature value = (time-domain feature value × time-domain weight) + (frequency-domain feature value × frequency-domain weight).

[0030] For example, the time domain eigenvalue is 2.0 millivolts, the frequency domain eigenvalue is 1.2 Hz, and the fusion eigenvalue is calculated as: 2.0×0.6+1.2×0.4=1.2+0.48=1.68. The setting of the feature weight coefficient is obtained through experimental calibration to ensure the accuracy of the fusion result.

[0031] See Figure 2 ,The calibration output module includes an environmental monitoring submodule and a feature calibration submodule; The environmental monitoring submodule monitors changes in ambient temperature, humidity, and electrode contact pressure in real time, updating data once a second. The change in environmental parameters is calculated by the difference between the initial environmental parameters. For example, the initial ambient temperature is 25 degrees Celsius, the current temperature is 26 degrees Celsius, and the temperature change is +1 degree Celsius; the initial humidity is 45% relative humidity, the current humidity is 47% relative humidity, and the humidity change is +2% relative humidity; the initial electrode contact pressure is 10 kPa, the current pressure is 10.5 kPa, and the pressure change is +0.5 kPa. The monitoring of environmental parameters is achieved through high-precision sensors to ensure the real-time and accuracy of the data.

[0032] The feature calibration submodule calculates the compensation value based on the change in environmental parameters. The compensation formula is: compensation value = k1 × temperature change + k2 × humidity change + k3 × pressure change, where k1, k2, and k3 are compensation coefficients for environmental parameters, obtained through experimental calibration, and are 0.02, 0.01, and 0.03, respectively. For example, substituting the above changes into the calculation, the compensation value is calculated as follows: 0.02×1+0.01×2+0.03×0.5=0.02+0.02+0.015=0.055. The setting of the compensation coefficient is obtained through experimental calibration to ensure the rationality of the compensation value.

[0033] The compensation value is mapped to each dimension of the fused feature vector using a linear mapping method: calibrated feature value = fused feature value + compensation value × fused feature value; For example, the fused eigenvalue in the fused eigenvector is 1.68, and the eigenvalue after calibration = 1.68 + 0.055 × 1.68 = 1.68 + 0.0924 = 1.7724. The linear mapping is achieved through simple mathematical operations to ensure the accuracy of the calibration results.

[0034] The calibrated feature vector includes the environmental parameter compensation value (such as the temperature compensation value of 0.055), the calibrated feature vector (such as [1.7724, 0.8]), and the feature calibration model (such as the linear mapping model parameters). Through the above monitoring and calibration process, the calibrated feature vector is finally generated. The calibrated feature vector includes the environmental parameter compensation value, the calibrated feature vector, and the feature calibration model, completing the data processing flow of the entire non-invasive troponin monitoring system.

[0035] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A non-invasive troponin monitoring system, characterized in that: The system comprises: The signal acquisition module collects bioelectric signals from the human skin surface, obtains electrocardiogram waveforms and respiratory waveforms, measures ambient temperature and humidity, and electrode contact pressure, and generates a raw signal data set; The filtering processing module analyzes the original signal data set to calculate the signal-to-noise ratio, dynamically adjusts the filter bandwidth after comparing it with a preset threshold, and generates a filtered signal; The feature extraction module extracts the time domain peak, rise time and frequency domain main frequency of the filtered signal, calculates the correlation coefficient between the feature parameters, and generates a feature parameter clustering result; The data fusion module synchronously analyzes the electrocardiogram cycle, extracts the characteristic parameters from the characteristic parameter clustering results, and integrates them to generate a multi-dimensional fusion feature vector; The calibration output module monitors changes in environmental parameters, calculates compensation values ​​and maps them to the multi-dimensional fusion feature vector to generate a calibrated feature vector.

2. The non-invasive troponin monitoring system according to claim 1, characterized in that: The original signal data set includes bioelectric signals, electrocardiogram waveforms, respiratory waveforms, and environmental parameters; the filtered signals include filtered troponin signals, signal-to-noise ratio data, and filter bandwidth parameters; the feature parameter clustering results include time domain feature parameters, frequency domain feature parameters, and feature correlation coefficients; the multi-dimensional fusion feature vector includes electrocardiogram cycle features, fusion feature vectors, and feature weight coefficients; and the calibrated feature vector includes environmental parameter compensation values, calibrated feature vectors, and feature calibration models.

3. The non-invasive troponin monitoring system according to claim 1, characterized in that: The signal acquisition module includes a signal acquisition submodule, a waveform acquisition submodule, and an environment monitoring submodule; The signal acquisition submodule collects bioelectric signals from the surface of human skin and generates bioelectric signal data; The waveform acquisition submodule acquires the electrocardiogram waveform and the respiratory waveform, and generates the electrocardiogram waveform data and the respiratory waveform data; The environmental monitoring submodule measures ambient temperature, humidity and electrode contact pressure to generate environmental parameter data.

4. The non-invasive troponin monitoring system according to claim 1, characterized in that: The filtering processing module includes a signal-to-noise ratio calculation submodule and a filter adjustment submodule; The signal-to-noise ratio calculation submodule analyzes the original signal data set to calculate the signal-to-noise ratio and generate signal-to-noise ratio data; The filter adjustment submodule dynamically adjusts the filter bandwidth after comparing with the preset signal-to-noise ratio threshold to generate a filtered signal.

5. The non-invasive troponin monitoring system according to claim 1, characterized in that: The feature extraction module includes a time domain feature extraction submodule and a frequency domain feature extraction submodule; The time domain feature extraction submodule extracts the time domain peak and rise time of the filtered signal and generates time domain feature parameters; The frequency domain feature extraction submodule extracts the frequency domain main frequency and power spectrum density, calculates the correlation coefficient between feature parameters, and generates feature parameter clustering results.

6. The non-invasive troponin monitoring system according to claim 5, characterized in that: The characteristic parameter clustering result adopts the formula: ； in, Representative Time domain characteristic parameters, Representative frequency domain characteristic parameters, Represents the clustering results of feature parameters.

7. The non-invasive troponin monitoring system according to claim 5, characterized in that: The data fusion module includes a periodic synchronization analysis submodule and a feature fusion submodule; The cycle synchronization analysis submodule synchronously analyzes the ECG cycle and generates ECG cycle features.

8. Feature fusion submodule, extracts the feature parameters within each cycle and integrates them to generate a multi-dimensional fusion feature vector.

9. The non-invasive troponin monitoring system according to claim 7, characterized in that: The calibration output module includes an environmental monitoring submodule and a feature calibration submodule; The environmental monitoring submodule monitors changes in environmental parameters and generates compensation values ​​for these parameters; The feature calibration submodule calculates the compensation value and maps it to the fused feature vector to generate the calibrated feature vector.

10. The non-invasive troponin monitoring system according to claim 8, characterized in that: The characteristic parameter clustering result includes time domain characteristic parameters, frequency domain characteristic parameters, and characteristic correlation coefficients; the multi-dimensional fusion characteristic vector includes electrocardiogram cycle characteristics, fusion characteristic vectors, and characteristic weight coefficients; the calibrated characteristic vector includes environmental parameter compensation values, calibrated characteristic vectors, and characteristic calibration models.

Images