SCG signal extraction and application method based on linear acceleration of intelligent terminal

Abstract

The application discloses a kind of SCG signal extraction and application method based on intelligent terminal linear acceleration, belong to the technical field of heart health monitoring.Utilize built-in acceleration sensor in mobile terminal to collect linear acceleration signal in the vertical direction of body surface, and obtain heart shock waveform after preprocessing;Through double threshold detection AO peak and construct interbeat interval sequence, calculate cardiac rhythmicity index;Through waveform salience screening AC peak, calculate mechanical work index according to the time difference and signal energy of AO peak and AC peak in the same cardiac cycle;According to the comparison result of two kinds of indexes and preset threshold, output heart state evaluation information.The application does not need special hardware, supports daily heart monitoring under home scene, can evaluate cardiac rhythmicity and mechanical work capacity synchronously, and can provide early warning for arrhythmia, autonomic nervous function disorder, heart failure, aortic valve stenosis and myocardial ischemia.

Description

Technical Field

[0001] This invention relates to the field of cardiac health monitoring technology, and in particular to a method for extracting and applying SCG signals based on linear acceleration of a smart terminal. Background Technology

[0002] Cardiac data is a core indicator for early warning of cardiovascular diseases, sleep quality assessment, and daily chronic disease management. With the significant acceleration of global population aging, the prevalence of chronic cardiovascular diseases such as hypertension and heart failure continues to rise, placing an increasingly heavy burden on social healthcare. Currently, common technologies used for cardiac health monitoring include: electrocardiography (ECG), which records cardiac electrical signals through electrodes on the skin surface, reflecting the heart's electrophysiological activity; photoplethysmography (PPG), which detects changes in blood flow by illuminating the skin, measuring heart rate and blood oxygen saturation, commonly found in smartwatches; echocardiography, which observes cardiac structure and movement through ultrasound imaging, providing the most comprehensive information; and scintigraphy (SCG), a technique that directly and accurately measures the micro-vibrations on the local surface of the chest caused by the mechanical pulsation of the heart, directly reflecting the ejection time of each heartbeat and the opening and closing time of the aortic valve. The mechanical pulsation of the heart is a complex process involving the coupling of fluid mechanics and solid mechanics. When the ventricles contract, the aortic valves open, and blood is pumped into the aorta at high speed. This ejection process generates a strong reaction force, which is transmitted through the intrathoracic tissues to the chest wall surface, forming a weak mechanical vibration wave, i.e., the SCG signal. Compared to the ECG, which only reflects the electrophysiological activity of the heart, the SCG can more directly reflect the purely physical vibration processes such as cardiac ejection and aortic valve opening. This allows for timely warnings of problems such as decreased myocardial contractility that ECG cannot detect. Compared to PPG, SCG can acquire richer cardiac indicators, not only heart rate and blood oxygen, but also calculate indicators such as IBI, LVET, and KE, which can be used for early warning of diseases such as heart failure, myocardial ischemia, and aortic stenosis. Compared to echocardiography, which requires a professional physician to perform imaging examinations of the heart in a hospital using specialized ultrasound equipment, SCG only requires placing an accelerometer flat on the surface of the sternum and recording the mechanical activity of the heart by collecting the tiny vibrations of the chest wall caused by the heartbeat. It does not require professional operators or hospital equipment, the data acquisition process is simple, and it is feasible for daily self-monitoring.

[0003] Currently, SCG signal acquisition primarily relies on dedicated accelerometers (such as the ADXL335). The ADXL335 is placed on the surface of the subject's sternum, and the sensor detects changes in chest wall acceleration in real time and outputs an analog voltage signal. Since the ADXL335 itself lacks digital processing and data storage capabilities, its output analog signal must first be transmitted to a microcontroller development board (such as Arduino). The board's built-in analog-to-digital converter converts the analog signal into a digital signal, which is then transmitted to a computer via a data cable. Dedicated software on the computer handles data recording and subsequent processing. The entire acquisition process must be conducted in a vibration-free indoor environment and operated and supervised by professionals. However, this method faces a high barrier to entry; users need to purchase a microcontroller development board and complete the wiring configuration, making it impossible to perform routine monitoring independently. Therefore, achieving portable, low-barrier-to-entry SCG signal acquisition and cardiovascular index calculation is a significant challenge. Summary of the Invention

[0004] This invention provides a method for extracting and applying SCG signals based on the linear acceleration of a smart terminal, which solves the problems of existing technologies relying on dedicated hardware and having high operational barriers, and realizes the synchronous assessment and early warning of cardiac rhythm and mechanical function in home scenarios.

[0005] This invention provides a method for extracting and applying SCG signals based on linear acceleration of a smart terminal, comprising: The linear acceleration time-series signal along the direction perpendicular to the body surface is acquired by the linear acceleration sensor built into the smart mobile terminal when the smart mobile terminal is attached to the body surface of the subject, and the linear acceleration time-series signal is preprocessed to obtain the cardiac vibration waveform signal. Based on the amplitude threshold and the time threshold, the local maxima of the cardiac waveform signal are detected by iterating through them and recorded as valid AO peak timestamps, thus obtaining the AO peak timestamp set. A cardiac interval sequence is constructed based on the time difference between adjacent AO peak timestamps. Outliers in the cardiac interval sequence are removed to obtain an effective cardiac interval sequence. First-class physiological parameters characterizing cardiac rhythmicity are calculated based on the effective cardiac interval sequence. Within the time window following each AO peak timestamp, local maxima are selected based on waveform prominence and marked as AC peaks. Based on the time difference and signal energy between the AO peak and the AC peak in the same cardiac cycle, a second type of physiological parameter characterizing the mechanical function of the heart is calculated. Based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds, cardiac status assessment information is output.

[0006] In one possible implementation, the preprocessing of the linear acceleration time-series signal to obtain the cardiac waveform signal includes: Discard the data of the first and last preset durations of the linear acceleration timing signal to obtain the effective signal segment; The effective signal segment is subjected to DC removal processing to obtain the DC-removed signal; The signal after DC removal is bandpass filtered to obtain the cardiac waveform signal.

[0007] In one possible implementation, the bandpass filtering process employs a 4th-order Butterworth filter with a lower cutoff frequency of 0.8 Hz and an upper cutoff frequency of 25.0 Hz, and the bandpass filtering process uses a forward-backward bidirectional filtering method.

[0008] In one possible implementation, the step of traversing the cardiac waveform signal to detect local maxima based on amplitude and time thresholds, recording them as valid AO peak timestamps, yields an AO peak timestamp set, including: The preprocessed cardiac waveform signal is divided into multiple consecutive time windows according to a preset duration. The standard deviation of the signal amplitude in each time window is calculated, and the standard deviations of adjacent time windows are linearly interpolated to generate a local standard deviation sequence that corresponds one-to-one with each sampling time. The dynamic amplitude threshold at each sampling time is determined by multiplying the local standard deviation sequence with the preset sensitivity coefficient. The minimum heart rate interval is determined based on the preset maximum heart rate value, and the minimum heart rate interval is converted into the minimum sampling point interval based on the sampling frequency of the cardiac waveform signal. The sampling points of the cardiac waveform signal are traversed sequentially in chronological order. The amplitude of the current sampling point is compared with the dynamic amplitude threshold corresponding to the current sampling point. If the amplitude of the current sampling point is greater than the dynamic amplitude threshold, and the number of sampling points experienced since the last recorded valid AO peak timestamp is not less than the minimum sampling point interval, then it is determined whether the current sampling point is a local maximum value in the predetermined neighborhood. If the current sampling point is determined to be a local maximum, the timestamp of the current sampling point is recorded as a valid AO peak timestamp, and the process continues until all sampling points are processed, thus obtaining an AO peak timestamp set composed of all valid AO peak timestamps.

[0009] In one possible implementation, the step of constructing a cardiac interbeat interval sequence based on the time difference between adjacent AO peak timestamps, removing outliers from the cardiac interbeat interval sequence to obtain a valid cardiac interbeat interval sequence, and calculating a first type of physiological parameter characterizing cardiac rhythmicity based on the valid cardiac interbeat interval sequence includes: Based on the time difference between two adjacent valid AO peak timestamps in the AO peak timestamp set, a cardiac interval sequence is constructed. Based on the median of the cardiac interval sequence, cardiac interval values ​​that deviate from the median by more than a preset percentage are removed to obtain a valid cardiac interval sequence; Based on the effective heartbeat interval sequence, calculate the heart rate value, overall heart rate variability index, parasympathetic activity index, and short-term rhythm fluctuation index respectively; The heart rate value, the overall heart rate variability index, the parasympathetic activity index, and the short-term rhythm fluctuation index together constitute the first type of physiological parameters.

[0010] In one possible implementation, the step of filtering local maxima based on waveform prominence and marking them as AC peaks within a time window after each AO peak timestamp includes: Within the first preset time window after each AO peak timestamp, local maxima are detected, the prominence value of each local maxima is calculated, and local maxima with prominence values ​​greater than a preset prominence threshold are marked as preliminary AC peaks. If there is no local maximum point with a prominence value greater than the preset prominence threshold within the first preset time window, then the local maximum point with the largest amplitude within the first preset time window is marked as the preliminary AC peak. The initial ejaculation time is calculated based on the time difference between the initial AC peak corresponding to each AO peak timestamp and each AO peak timestamp. An initial ejaculation time series is constructed based on all initial ejaculation times, and the median of the initial ejaculation time series is calculated. The upper and lower limits of the second preset time window are determined based on the median. Local maxima are detected again within the second preset time window and the prominence value is calculated. Local maxima with prominence values ​​greater than the preset prominence threshold are marked as the final AC peak. If there is no local maximum point with a prominence value greater than the preset prominence threshold within the second preset time window, then the preliminary AC peak is marked as the final AC peak.

[0011] In one possible implementation, the calculation of a second type of physiological parameter characterizing cardiac mechanical function based on the time difference and signal energy between the AO peak and the AC peak within the same cardiac cycle includes: Based on the time difference between each AO peak timestamp and the corresponding final AC peak, the left ventricular ejection time of each cardiac cycle is calculated, and the left ventricular ejection time of each cardiac cycle is arranged in the time order of the cardiac cycle to obtain the left ventricular ejection time sequence. Based on the energy integral of the cardiac waveform signal between each AO peak timestamp and the corresponding final AC peak, the systolic kinetic energy integral of each cardiac cycle is calculated, and the systolic kinetic energy integrals of each cardiac cycle are arranged in the time order of the cardiac cycles to obtain the systolic kinetic energy integral sequence. Calculate the median of the left ventricular ejection time series and the median of the systolic kinetic energy integral series, respectively. The left ventricular ejection time and systolic kinetic energy integral of each cardiac cycle were scored. Left ventricular ejection time and corresponding systolic kinetic energy integral with quality scores below a preset threshold were removed to obtain effective left ventricular ejection time series and effective systolic kinetic energy integral series. The mean of the effective left ventricular ejection time sequence is calculated as the final left ventricular ejection time, and the mean of the effective systolic kinetic energy integral sequence is calculated as the final systolic kinetic energy integral. The final left ventricular ejection time and the final systolic kinetic energy integral together constitute the second type of physiological parameter characterizing the heart's mechanical function.

[0012] In one possible implementation, the step of outputting cardiac state assessment information based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds includes: The first type of physiological parameter is compared with the first preset threshold range, and cardiac rhythm status assessment information and / or autonomic nerve function status assessment information are output according to the comparison result. And / or, compare the second type of physiological parameter with a second preset threshold range, and output cardiac mechanical function status assessment information based on the comparison result; The cardiac rhythm status assessment information includes arrhythmia warning information and / or normal rhythm information; the autonomic nervous system function status assessment information includes autonomic nervous system dysfunction warning information and / or normal regulation information; the cardiac mechanical function status assessment information includes at least one of heart failure warning information, aortic stenosis warning information, and myocardial ischemia warning information and / or normal mechanical function information.

[0013] One or more technical solutions provided in this invention have at least the following technical effects or advantages: This invention replaces traditional dedicated accelerometers and associated data acquisition equipment with ordinary smart mobile terminals, eliminating the need for additional hardware, reducing equipment costs and operational barriers, and enabling autonomous cardiac monitoring in everyday home settings. By preprocessing the raw signal through DC removal and filtering, interference from sensor fixation bias, respiratory baseline drift, and high-frequency electromagnetic noise on the cardiac waveform is effectively eliminated, improving the signal quality and reliability of subsequent feature extraction. By setting amplitude thresholds and false negative detection time thresholds that dynamically change with signal amplitude, AO peak detection can adapt to signal intensity variations among different subjects and at different times for the same subject, reducing false positive rates when the signal is strong and maintaining detection sensitivity when the signal is weak, ensuring accurate and stable localization of the aortic valve opening time in each cardiac cycle. By eliminating abnormal values ​​in the cardiac interval caused by motion artifacts or occasional interference, the interval data used in heart rate variability analysis is ensured to be pure and reliable, thereby calculating multi-dimensional rhythm indicators such as heart rate, overall heart rate variability, parasympathetic activity, and short-term rhythm fluctuations, achieving quantitative analysis of arrhythmias and autonomic nervous system dysfunction. The system provides comprehensive assessment and early warning. By using waveform prominence rather than a single amplitude as the screening criterion, it accurately locates the weak vibration peak corresponding to the aortic valve closure moment within a limited time window after the AO peak, effectively distinguishing the AC peak from other physiological vibration components within the same cycle. This provides an accurate closure time benchmark for the subsequent precise calculation of left ventricular ejection time. By combining the time interval between aortic valve opening and closing within the same cardiac cycle and the vibration energy of the cardiac waveform during that interval, it calculates the left ventricular ejection time and systolic kinetic energy integral, quantifying the duration of ejection and the mechanical energy generated by myocardial contraction with each heartbeat. This provides a direct mechanical function assessment basis for the early warning of heart failure, aortic stenosis, and myocardial ischemia. Through the joint analysis of cardiac rhythmicity indicators and mechanical work indicators, the quantified parameters are transformed into clinically significant warning information, enabling non-professional users to obtain risk warnings for five heart diseases—arrhythmia, autonomic dysfunction, heart failure, aortic stenosis, and myocardial ischemia—without physician intervention, achieving a complete closed loop from signal acquisition to health status assessment. Attached Figure Description

[0014] Figure 1 This is a flowchart illustrating the steps of the SCG signal extraction and application method based on linear acceleration of a smart terminal provided in an embodiment of the present invention. Figure 2 This is a schematic diagram of the cardiac waveform signal and AO peak annotation provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of an effective cardiac interval sequence provided in an embodiment of the present invention; Figure 4 This is a schematic diagram of the health status determination result based on the first type of physiological parameters provided in an embodiment of the present invention; Figure 5 This is a schematic diagram of the left ventricular ejection time sequence provided in an embodiment of the present invention; Figure 6 This is a schematic diagram of the health status determination result based on the second type of physiological parameters provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0016] This invention provides a method for extracting and applying SCG signals based on linear acceleration of a smart terminal. (See also...) Figure 1 The process includes the following steps S101 to S106.

[0017] S101, acquire the linear acceleration time-series signal along the direction perpendicular to the body surface when the smart mobile terminal is attached to the subject's body surface, collected by the linear acceleration sensor built into the smart mobile terminal, and preprocess the linear acceleration time-series signal to obtain the cardiac vibration waveform signal; Specifically, in step S101, the linear acceleration time-series signal is preprocessed to obtain the cardiac waveform signal, including the following steps S1011 to S1013.

[0018] S1011, discard the data of the first and last preset durations of the linear acceleration timing signal to obtain the effective signal segment; S1012 performs DC removal processing on the effective signal segment to obtain the DC-removed signal; S1013 performs bandpass filtering on the DC-free signal to obtain the oscillation waveform signal. Here, a 4th-order Butterworth filter is used for bandpass filtering. The lower cutoff frequency of the Butterworth filter is 0.8Hz, and the upper cutoff frequency is 25.0Hz. The bandpass filtering process uses a forward-backward bidirectional filtering method.

[0019] For example, this embodiment uses a regular smartphone as the sensing device, with the Phyphox application installed and running. The test subject is in a supine, resting position, maintaining natural breathing. The smartphone is placed horizontally against the lower half of the test subject's sternum (above the precordial region), ensuring that the phone's Z-axis (screen normal direction) is perpendicular to the chest wall to maximize the acquisition of micro-movements caused by cardiac ejection. Specific implementation flow: On your mobile device, tap the "Acceleration (excluding g)" module in the Phyphox app. Figure 2As shown, triaxial data is recorded. In actual data processing, to eliminate the influence of the subject's phone placement and finger tapping on the screen, the algorithm discards the edge data of the first 5 seconds and last 5 seconds of the acquired sequence, retaining only the clean sequence within the middle stable time window. The acquisition process lasts for 60 seconds, obtaining the original linear acceleration sequence perpendicular to the chest wall (Z-axis) after eliminating the influence of gravitational acceleration g, denoted as... ,in, It is a time series.

[0020] After the data acquisition is completed, export a CSV file containing timestamp sequences and triaxial acceleration data, and perform data processing on the exported CSV file on the MATLAB platform.

[0021] For the obtained raw sequence of Z-axis linear acceleration To perform DC baseline removal, the mean of the entire signal segment is calculated and subtracted point by point. The formula is as follows: ; The processed signal fluctuates around zero, which facilitates subsequent filtering and peak detection.

[0022] An adaptive threshold is calculated for the baseline-removed signal for subsequent AO peak detection. The signal is divided into 5-second segments, and the standard deviation of each segment is calculated to measure the overall fluctuation amplitude of the signal in that segment. Since only one standard deviation value is obtained for each segment, linear interpolation is used to smooth the transition between adjacent segments, generating a threshold that dynamically changes with the signal amplitude for each sampling point. ; in, For The standard deviation of the signal within a 5-second window centered on the signal. This is the sensitivity coefficient. The local standard deviation at the current moment is multiplied by 0.5 to form the detection threshold for that moment. The threshold is higher for periods with strong signal amplitude and automatically lowers for periods with weak signal amplitude, always following changes in signal amplitude to ensure accurate detection of AO peaks in all preceding and following periods.

[0023] For the baseline-removed acceleration signal Bandpass filtering is performed to extract the cardiac waveform signal, i.e., the pure SCG signal. The effective frequency band of the SCG signal is 0.8Hz-25Hz, therefore a lower limit frequency is designed. =0.8Hz, upper limit frequency A fourth-order Butterworth bandpass filter with a frequency of 25.0Hz was used, employing both forward and backward filtering to ensure that the peak position after filtering was aligned with the actual heartbeat time, resulting in a pure SCG signal. .

[0024] .

[0025] S102, based on the amplitude threshold and time threshold, traverse the cardiac waveform signal to detect local maxima and record them as valid AO peak timestamps to obtain the AO peak timestamp set; Specifically, in step S102, based on the amplitude threshold and the time threshold, the local maxima of the cardiac waveform signal are detected and recorded as valid AO peak timestamps, thus obtaining the AO peak timestamp set, including the following steps S1021 to S1025.

[0026] S1021, the preprocessed cardiac waveform signal is divided into multiple consecutive time windows according to a preset duration, the standard deviation of the signal amplitude in each time window is calculated, and the standard deviation of adjacent time windows is linearly interpolated to generate a local standard deviation sequence that corresponds one-to-one with each sampling time. S1022, Determine the dynamic amplitude threshold at each sampling time based on the product of the local standard deviation sequence and the preset sensitivity coefficient; S1023, determine the minimum heart rate interval based on the preset maximum heart rate value, and convert the minimum heart rate interval into the minimum sampling point interval based on the sampling frequency of the cardiac waveform signal; S1024, sequentially traverse each sampling point of the cardiac waveform signal in chronological order, compare the amplitude of the current sampling point with the dynamic amplitude threshold corresponding to the current sampling point, if the amplitude of the current sampling point is greater than the dynamic amplitude threshold, and the number of sampling points experienced since the last recorded valid AO peak timestamp is not less than the minimum sampling point interval, then determine whether the current sampling point is a local maximum value in the predetermined neighborhood. S1025, if the current sampling point is determined to be a local maximum value, then the timestamp of the current sampling point is recorded as a valid AO peak timestamp, and the process continues to traverse until all sampling points are processed, thus obtaining an AO peak timestamp set composed of all valid AO peak timestamps.

[0027] For example, in the SCG signal, the physical vibration amplitude generated by aortic valve opening (AO) is the largest. To accurately extract the AO peak of continuous heartbeats, a dual threshold is set: Amplitude threshold: Calculate the standard deviation of the pure SCG signal and set the minimum peak height. ; in, Take 0.5, lower than Fluctuations are considered noise and are not included in the heartbeat candidate points.

[0028] Time threshold: The sampling point interval between two adjacent peaks must be no less than [a certain value]. ,in It is calculated from the minimum heart rate interval corresponding to a maximum heart rate of 130 BPM: ; ; in, This represents the actual sampling rate, with the interval between two adjacent peaks being less than [a certain value]. In this case, the subsequent peak is not recorded as an independent heartbeat to prevent secondary peaks within the same heartbeat cycle from being misjudged.

[0029] Traverse the pure SCG signal, and satisfy Finding local maxima under constraints and recording them as valid AO peak timestamps yields a set of AO peak timestamps, such as... Figure 2 As shown: .

[0030] S103, construct a cardiac interval sequence based on the time difference between adjacent AO peak timestamps, remove outliers from the cardiac interval sequence to obtain an effective cardiac interval sequence, and calculate the first type of physiological parameters characterizing cardiac rhythmicity based on the effective cardiac interval sequence; Specifically, in step S103, a cardiac interbeat sequence is constructed based on the time difference between adjacent AO peak timestamps, outliers in the cardiac interbeat sequence are removed to obtain an effective cardiac interbeat sequence, and the first type of physiological parameters characterizing cardiac rhythmicity are calculated based on the effective cardiac interbeat sequence, including the following steps S1031 to S1034.

[0031] S1031, construct the cardiac interval sequence based on the time difference between two adjacent valid AO peak timestamps in the AO peak timestamp set; S1032, Based on the median of the heartbeat interval sequence, remove heartbeat interval values ​​that deviate from the median by more than a preset percentage to obtain a valid heartbeat interval sequence; S1033, calculate heart rate value, overall heart rate variability index, parasympathetic activity index and short-term rhythm fluctuation index respectively based on effective heartbeat interval sequence; S1034, heart rate value, overall heart rate variability index, parasympathetic activity index and short-term rhythm fluctuation index together constitute the first type of physiological parameter.

[0032] For example, a heartbeat interval sequence, i.e., an IBI sequence, is calculated and constructed, and outliers are removed. The time difference between two adjacent AO peaks is one heartbeat interval. ; Calculate the median of the IBI sequence Outliers deviating more than 20% from the median are removed to obtain valid IBI sequences, such as... Figure 3 As shown: ; Calculation of cardiovascular parameters based on effective IBI sequences: Heart rate (HR): The value for healthy adults is usually between 60 bpm and 120 bpm.

[0033] ; Overall rhythm variability index (SDNN): Calculates the standard deviation of all effective IBIs, reflecting the heart's ability to regulate external stimuli. The value for healthy adults is usually between 30ms and 60ms.

[0034] ; Parasympathetic activity index (RMSSD): Calculated by the root mean square of the difference between adjacent IBI values, reflecting the activity of the parasympathetic nervous system. The value of healthy adults is usually between 20ms and 75ms.

[0035] ; Abnormal jump rate (pNN50): The percentage of times the absolute value of the difference between adjacent IBI values ​​exceeds 50ms. The value for healthy adults is usually between 30% and 60%.

[0036] .

[0037] S104. Within the time window after each AO peak timestamp, local maxima are selected based on waveform prominence and marked as AC peaks. Specifically, in step S104, within the time window after each AO peak timestamp, local maxima are selected based on waveform prominence and marked as AC peaks, including the following steps S1041 to S1045.

[0038] S1041, within the first preset time window after each AO peak timestamp, detect local maxima, calculate the prominence value of each local maxima, and mark the local maxima with prominence values ​​greater than a preset prominence threshold as preliminary AC peaks; S1042, if there is no local maximum point with a prominence value greater than the preset prominence threshold within the first preset time window, then mark the local maximum point with the largest amplitude within the first preset time window as the preliminary AC peak. S1043, calculate the initial ejaculation time based on the time difference between the initial AC peak corresponding to each AO peak timestamp and each AO peak timestamp, construct the initial ejaculation time series based on all initial ejaculation times, and calculate the median of the initial ejaculation time series; S1044, determine the upper and lower limits of the second preset time window based on the median, detect local maxima again within the second preset time window and calculate the prominence value, and mark the local maxima with prominence values ​​greater than the preset prominence threshold as the final AC peak; S1045, if there is no local maximum point with a prominence value greater than the preset prominence threshold within the second preset time window, then the preliminary AC peak is marked as the final AC peak.

[0039] For example, the AC peak corresponds to the aortic valve closure moment and appears after each AO peak. A two-round AC peak detection method is used to locate the AC peak for each heartbeat. The degree of prominence is used as the screening criterion for AC peaks, set at 3% of the global standard deviation. Only peaks with a prominence exceeding this value are considered true AC peak candidates. ; Round 1: Within a window of 200ms-400ms after each AO peak, find the peak whose prominence meets the above conditions, select the peak with the highest prominence as the candidate AC peak, and calculate the preliminary LVET median. If there is no peak that meets the conditions within the window, the absolute maximum value within the window is directly taken as the candidate AC peak.

[0040] Second round: Repositioning using ±15% of the median from the first round as an adaptive window: ; Within a narrower window, the peak is further filtered by prominence. If no peak meeting the criteria is found in the second window, the result from the first round is used. If no result is found in the first round either, the absolute maximum value within the window is taken as a candidate. This yields the accurate AC peak position, and the LVET result is further calculated as follows: Figure 5 As shown.

[0041] S105, based on the time difference and signal energy between the AO peak and AC peak in the same cardiac cycle, calculates the second type of physiological parameter characterizing the mechanical function of the heart. Specifically, in step S105, based on the time difference and signal energy between the AO peak and AC peak in the same cardiac cycle, a second type of physiological parameter characterizing the mechanical function of the heart is calculated, including the following steps S1051 to S1056.

[0042] S1051, Based on the time difference between each AO peak timestamp and the corresponding final AC peak, calculate the left ventricular ejection time of each cardiac cycle, and arrange the left ventricular ejection times of each cardiac cycle according to the time order of the cardiac cycle to obtain the left ventricular ejection time sequence. S1052, calculate the systolic kinetic energy integral of each cardiac cycle based on the energy integral of the cardiac waveform signal between the timestamp of each AO peak and the final AC peak corresponding to each AO peak, and arrange the systolic kinetic energy integrals of each cardiac cycle according to the time order of the cardiac cycle to obtain the systolic kinetic energy integral sequence. S1053, calculate the median of the left ventricular ejection time series and the median of the systolic kinetic energy integral series respectively, based on the median of the left ventricular ejection time series and the median of the systolic kinetic energy integral series; S1054, the quality scores of left ventricular ejection time and systolic kinetic energy integral for each cardiac cycle are calculated, and left ventricular ejection time and corresponding systolic kinetic energy integral with quality scores below a preset threshold are removed to obtain effective left ventricular ejection time series and effective systolic kinetic energy integral series. S1055, calculate the mean of the effective left ventricular ejection time sequence as the final left ventricular ejection time, and calculate the mean of the effective systolic kinetic energy integral sequence as the final systolic kinetic energy integral; S1056, the final left ventricular ejection time, and the final systolic kinetic energy integral together constitute the second type of physiological parameter characterizing the mechanical function of the heart.

[0043] For example, after obtaining the AO peak timestamp sequence and the effective IBI sequence, the AC peak is further extracted to calculate the left ventricular ejection time (LVET) and systolic kinetic energy integral (KE), providing early warning for heart failure, aortic stenosis, and myocardial ischemia.

[0044] Left ventricular ejection time (LVET): The time elapsed from the opening of the aortic valve (AO peak) to the closing of the aortic valve (AC peak), reflecting the duration of each cardiac ejection. A heart with strong myocardial contractility completes ejection within a normal time; in heart failure, the myocardium is weak, resulting in an abnormally shortened ejection time; in aortic stenosis, blood requires more time to squeeze through the narrowed valve orifice, resulting in an abnormally prolonged ejection time. In healthy adults, the value is typically between 248 ms and 336 ms.

[0045] ; Systolic kinetic energy integral (KE): This is the square integral of the SCG signal between the AO and AC peaks, reflecting the magnitude of the mechanical energy generated by the heart during this ejection. Stronger myocardial contractility results in greater chest wall vibration amplitude and a larger integral area; conversely, decreased contractility and a smaller integral area occur during myocardial ischemia or heart failure. There is no uniform absolute normal value for KE; the individual's mean measurement at rest is used as the baseline. Subsequent measurements consistently below 30% of the baseline indicate decreased myocardial contractility.

[0046] .

[0047] S106, based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds, output cardiac status assessment information.

[0048] Specifically, in step S106, cardiac status assessment information is output based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds, including the following steps S1061 to S1062.

[0049] S1061, compare the first type of physiological parameter with the first preset threshold range, and output cardiac rhythm status assessment information and / or autonomic nerve function status assessment information according to the comparison result; And / or, compare the second type of physiological parameter with the second preset threshold range, and output cardiac mechanical function status assessment information based on the comparison result; S1062, cardiac rhythm status assessment information includes arrhythmia warning information and / or normal rhythm information; autonomic nervous system function status assessment information includes autonomic nervous system dysfunction warning information and / or normal regulation information; cardiac mechanical function status assessment information includes at least one of heart failure warning information, aortic stenosis warning information, and myocardial ischemia warning information and / or normal mechanical function information.

[0050] For example, disease warning judgments are made based on the values ​​of various indicators. Arrhythmia warning: A significant irregular jump appears in the IBI sequence, and... This indicates irregular heart rhythm fluctuations; medical examination is recommended. Warning sign of autonomic nervous system dysfunction: and and This suggests a decline in autonomic nervous system regulation; adequate rest and further examination are recommended. The final test results are as follows: Figure 4 As shown, the test subject is in good health.

[0051] In a specific embodiment of the present invention, scoring is performed hop-by-hop, low-quality heartbeats are eliminated, and each valid AC peak candidate is scored based on three conditions, with 1 point awarded for meeting the conditions: The AO peak amplitude is higher than 1.5 times the adaptive threshold of the current position, indicating that the AO peak of this heartbeat is prominent enough and the signal quality is reliable, so 1 point is awarded.

[0052] The deviation of this LVET from the median of all candidate LVETs is no more than 20%, indicating that the ejection time is within the normal range for this subject, and a score of 1 is awarded. ; The kinetic energy integral of this ejection deviates from the median of all candidate KE values ​​by no more than 50%, indicating that the mechanical energy of this ejection is within the normal range, earning 1 point. ; Heartbeats with a total score below 2 are considered low quality and will not be included in subsequent indicator calculations.

[0053] Disease early warning based on the values ​​of various indicators: Heart failure warning: and This indicates decreased myocardial contractility and shortened ejection time; medical examination is recommended.

[0054] Aortic stenosis warning: This indicates that the heart needs to work harder for a long time to squeeze blood through the narrow valve opening, and it is recommended to seek medical attention for examination.

[0055] Myocardial ischemia warning: subsequent measurements The value remained below A blood glucose level of 70% suggests insufficient myocardial blood supply and decreased contractility; medical examination is recommended. Final test results are as follows: Figure 6 As shown, the test subject is in good health.

[0056] The various embodiments described in this specification are presented in a progressive manner. Similar or identical parts between embodiments can be referred to interchangeably. Each embodiment focuses on its differences from other embodiments. All or part of this invention can be used in numerous general-purpose or special-purpose computer system environments or configurations. Examples include: personal computers, server computers, handheld or portable devices, tablet devices, mobile communication terminals, multiprocessor systems, microprocessor-based systems, programmable electronic devices, network PCs, minicomputers, mainframe computers, and distributed computing environments including any of the above systems or devices, etc.

[0057] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the present invention.

Claims

1. A method for extracting and applying SCG signals based on linear acceleration of a smart terminal, characterized in that, include: The linear acceleration time-series signal along the direction perpendicular to the body surface is acquired by the linear acceleration sensor built into the smart mobile terminal when the smart mobile terminal is attached to the body surface of the subject, and the linear acceleration time-series signal is preprocessed to obtain the cardiac vibration waveform signal. Based on the amplitude threshold and the time threshold, the local maxima of the cardiac waveform signal are detected by iterating through them and recorded as valid AO peak timestamps, thus obtaining the AO peak timestamp set. A cardiac interval sequence is constructed based on the time difference between adjacent AO peak timestamps. Outliers in the cardiac interval sequence are removed to obtain an effective cardiac interval sequence. First-class physiological parameters characterizing cardiac rhythmicity are calculated based on the effective cardiac interval sequence. Within the time window following each AO peak timestamp, local maxima are selected based on waveform prominence and marked as AC peaks. Based on the time difference and signal energy between the AO peak and the AC peak in the same cardiac cycle, a second type of physiological parameter characterizing the mechanical function of the heart is calculated. Based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds, cardiac status assessment information is output.

2. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, The preprocessing of the linear acceleration time-series signal to obtain the cardiac waveform signal includes: Discard the data of the first and last preset durations of the linear acceleration timing signal to obtain the effective signal segment; The effective signal segment is subjected to DC removal processing to obtain the DC-removed signal; The signal after DC removal is bandpass filtered to obtain the cardiac waveform signal.

3. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 2, characterized in that, The bandpass filtering process employs a 4th-order Butterworth filter with a lower cutoff frequency of 0.8 Hz and an upper cutoff frequency of 25.0 Hz. The bandpass filtering process uses a forward-backward bidirectional filtering method.

4. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, The process involves iterating through the cardiac waveform signal based on amplitude and time thresholds to detect local maxima, recording them as valid AO peak timestamps, and obtaining a set of AO peak timestamps, including: The preprocessed cardiac waveform signal is divided into multiple consecutive time windows according to a preset duration. The standard deviation of the signal amplitude in each time window is calculated, and the standard deviations of adjacent time windows are linearly interpolated to generate a local standard deviation sequence that corresponds one-to-one with each sampling time. The dynamic amplitude threshold at each sampling time is determined by multiplying the local standard deviation sequence with the preset sensitivity coefficient. The minimum heart rate interval is determined based on the preset maximum heart rate value, and the minimum heart rate interval is converted into the minimum sampling point interval based on the sampling frequency of the cardiac waveform signal. The sampling points of the cardiac waveform signal are traversed sequentially in chronological order. The amplitude of the current sampling point is compared with the dynamic amplitude threshold corresponding to the current sampling point. If the amplitude of the current sampling point is greater than the dynamic amplitude threshold, and the number of sampling points experienced since the last recorded valid AO peak timestamp is not less than the minimum sampling point interval, then it is determined whether the current sampling point is a local maximum value in the predetermined neighborhood. If the current sampling point is determined to be a local maximum, the timestamp of the current sampling point is recorded as a valid AO peak timestamp, and the process continues until all sampling points are processed, thus obtaining an AO peak timestamp set composed of all valid AO peak timestamps.

5. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, The process involves constructing a cardiac interval sequence based on the time difference between adjacent AO peak timestamps, removing outliers from the cardiac interval sequence to obtain a valid cardiac interval sequence, and calculating a first type of physiological parameter characterizing cardiac rhythmicity based on the valid cardiac interval sequence, including: Based on the time difference between two adjacent valid AO peak timestamps in the AO peak timestamp set, a cardiac interval sequence is constructed. Based on the median of the cardiac interval sequence, cardiac interval values ​​that deviate from the median by more than a preset percentage are removed to obtain a valid cardiac interval sequence; Based on the effective heartbeat interval sequence, calculate the heart rate value, overall heart rate variability index, parasympathetic activity index, and short-term rhythm fluctuation index respectively; The heart rate value, the overall heart rate variability index, the parasympathetic activity index, and the short-term rhythm fluctuation index together constitute the first type of physiological parameters.

6. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, Within the time window following each AO peak timestamp, local maxima are selected based on waveform prominence and marked as AC peaks, including: Within the first preset time window after each AO peak timestamp, local maxima are detected, the prominence value of each local maxima is calculated, and local maxima with prominence values ​​greater than a preset prominence threshold are marked as preliminary AC peaks. If there is no local maximum point with a prominence value greater than the preset prominence threshold within the first preset time window, then the local maximum point with the largest amplitude within the first preset time window is marked as the preliminary AC peak. The initial ejaculation time is calculated based on the time difference between the initial AC peak corresponding to each AO peak timestamp and each AO peak timestamp. An initial ejaculation time series is constructed based on all initial ejaculation times, and the median of the initial ejaculation time series is calculated. The upper and lower limits of the second preset time window are determined based on the median. Local maxima are detected again within the second preset time window and the prominence value is calculated. Local maxima with prominence values ​​greater than the preset prominence threshold are marked as the final AC peak. If there is no local maximum point with a prominence value greater than the preset prominence threshold within the second preset time window, then the preliminary AC peak is marked as the final AC peak.

7. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, The calculation of a second type of physiological parameter characterizing cardiac mechanical function based on the time difference and signal energy between the AO peak and the AC peak in the same cardiac cycle includes: Based on the time difference between each AO peak timestamp and the corresponding final AC peak, the left ventricular ejection time of each cardiac cycle is calculated, and the left ventricular ejection time of each cardiac cycle is arranged in the time order of the cardiac cycle to obtain the left ventricular ejection time sequence. Based on the energy integral of the cardiac waveform signal between each AO peak timestamp and the corresponding final AC peak, the systolic kinetic energy integral of each cardiac cycle is calculated, and the systolic kinetic energy integrals of each cardiac cycle are arranged in the time order of the cardiac cycles to obtain the systolic kinetic energy integral sequence. Calculate the median of the left ventricular ejection time series and the median of the systolic kinetic energy integral series, respectively. The left ventricular ejection time and systolic kinetic energy integral of each cardiac cycle were scored. Left ventricular ejection time and corresponding systolic kinetic energy integral with quality scores below a preset threshold were removed to obtain effective left ventricular ejection time series and effective systolic kinetic energy integral series. The mean of the effective left ventricular ejection time sequence is calculated as the final left ventricular ejection time, and the mean of the effective systolic kinetic energy integral sequence is calculated as the final systolic kinetic energy integral. The final left ventricular ejection time and the final systolic kinetic energy integral together constitute the second type of physiological parameter characterizing the heart's mechanical function.

8. The method for extracting and applying SCG signals based on linear acceleration of a smart terminal according to claim 1, characterized in that, The step of outputting cardiac status assessment information based on the comparison results of the first type of physiological parameters and / or the second type of physiological parameters with preset thresholds includes: The first type of physiological parameter is compared with the first preset threshold range, and cardiac rhythm status assessment information and / or autonomic nerve function status assessment information are output according to the comparison result. And / or, compare the second type of physiological parameter with a second preset threshold range, and output cardiac mechanical function status assessment information based on the comparison result; The cardiac rhythm status assessment information includes arrhythmia warning information and / or normal rhythm information; the autonomic nervous system function status assessment information includes autonomic nervous system dysfunction warning information and / or normal regulation information; the cardiac mechanical function status assessment information includes at least one of heart failure warning information, aortic stenosis warning information, and myocardial ischemia warning information and / or normal mechanical function information.

Images