Signal detection method and device, computer device, storage medium and program product

Abstract

The application relates to a signal detection method and device, computer equipment, a storage medium and a program product. The method comprises the following steps: obtaining a target photoplethysmogram signal of a to-be-detected object; the target photoplethysmogram signal comprises multiple sub-signals; performing quality assessment on each sub-signal to obtain a quality assessment result; the quality assessment result comprises any one of a high-quality signal, a general signal and a poor-quality signal; according to the quality assessment results of the multiple sub-signals, performing feature extraction on the multiple sub-signals to obtain multiple feature information; each sub-signal corresponds to one feature information; and according to the quality assessment results of all the sub-signals and the multiple feature information, determining an abnormal condition of the target photoplethysmogram signal. The signal detection method provided by the application can improve the accuracy of detecting the abnormal condition of the target photoplethysmogram signal, thereby improving the accuracy of detecting whether the to-be-detected object has an atrial fibrillation event.

Description

Technical Field

[0001] This application relates to the field of signal processing technology, and in particular to a signal detection method, apparatus, computer equipment, storage medium, and program product. Background Technology

[0002] Atrial fibrillation (AF), also known as atrial fibrillation, is the most common and serious cardiac arrhythmia. It not only increases the risk of stroke by 5 times, but also induces heart failure, worsens myocardial ischemia, and doubles the risk of death. Therefore, it is necessary to test patients for atrial fibrillation.

[0003] In traditional techniques, atrial fibrillation detection based on photoplethysmography (PPG) has become a research hotspot due to its advantages such as non-invasiveness, low cost, and ease of integration. However, existing PPG atrial fibrillation detection methods suffer from low detection accuracy. Summary of the Invention

[0004] Therefore, it is necessary to provide a signal detection method, apparatus, computer equipment, storage medium, and program product that can improve the detection accuracy in response to the above-mentioned technical problems.

[0005] In a first aspect, this application provides a signal detection method, the method comprising:

[0006] Acquire the target photoplethysmography (PPG) signal of the object to be detected; the target PPG signal includes multiple frames of sub-signals;

[0007] Each frame of sub-signal is evaluated for quality, and the quality evaluation result is obtained. The quality evaluation result includes any one of the following: the sub-signal is a high-quality signal, a normal signal, or a poor-quality signal.

[0008] Based on the quality assessment results of the multi-frame sub-signals, feature extraction is performed on the multi-frame sub-signals to obtain multiple feature information; each frame sub-signal corresponds to one feature information.

[0009] Based on the quality assessment results of all frame sub-signals and multiple feature information, the anomalies of the target photoplethysmography signal are determined.

[0010] In one embodiment, a quality assessment is performed on multiple frames of sub-signals to obtain a quality assessment result, including:

[0011] For each sub-signal frame, determine the median of all pulse intervals in the sub-signal, and determine the first evaluation index based on each pulse interval and the median in the sub-signal.

[0012] The sample entropy of the sub-signal is determined to obtain the second evaluation index;

[0013] Determine the peak information of the sub-signal, and determine the third evaluation index based on the peak information;

[0014] The quality assessment result is determined based on at least one of the first assessment indicator, the second assessment indicator, and the third assessment indicator.

[0015] In one embodiment, the third evaluation metric is determined based on the peak information, including:

[0016] Based on the peak information, determine the maximum slope sequence corresponding to the sub-signal, and calculate the variance of the maximum slope sequence to obtain the first variance;

[0017] The pulse width sequence corresponding to the sub-signal is determined based on the peak information, and the variance of the pulse width sequence is calculated to obtain the second variance;

[0018] Based on the peak information, determine the peak amplitude sequence corresponding to the sub-signal, and calculate the variance of the peak amplitude sequence to obtain the third variance;

[0019] The third evaluation metric is determined based on at least one of the first variance, the second variance, and the third variance.

[0020] In one embodiment, based on the quality assessment results of the multi-frame sub-signals, feature extraction is performed on the multi-frame sub-signals to obtain multiple feature information, including:

[0021] Based on the quality assessment results of multi-frame sub-signals, the target feature extraction method is determined; the target feature extraction method includes envelope extraction, downsampling processing, power spectrum calculation, and frequency domain feature calculation.

[0022] Based on the target feature extraction method, feature extraction is performed on the multi-frame sub-signals to obtain multiple feature information; each feature information includes at least one of the following: main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability.

[0023] In one embodiment, the target feature extraction method is determined based on the quality assessment results of multi-frame sub-signals, including:

[0024] If a sub-signal with a consecutive preset number of frames in a multi-frame sub-signal is determined to be a high-quality signal, the first feature extraction method is determined as the target feature extraction method.

[0025] The first feature extraction method includes a first envelope extraction, a first downsampling process, a first power spectrum calculation, and a first frequency domain feature calculation; the first envelope extraction uses the Hilbert transform method, and the first power spectrum calculation uses the Gossel algorithm for detecting a first preset number of specific frequencies.

[0026] In one embodiment, the target feature extraction method is determined based on the quality assessment results of multi-frame sub-signals, including:

[0027] If it is determined that the sub-signals of a consecutive preset number of frames in the multi-frame sub-signals are general signals, the second feature extraction method is determined as the target feature extraction method;

[0028] The second feature extraction method includes second envelope extraction, second downsampling processing, second power spectrum calculation, and second frequency domain feature calculation. The second envelope extraction adopts the moving root mean square method, the second power spectrum calculation adopts the Gossel algorithm with a second preset number of specific frequency detections, the sampling frequency of the second downsampling processing is less than the sampling frequency of the first downsampling processing, and the second preset number is less than the first preset number.

[0029] In one embodiment, the target feature extraction method is determined based on the quality assessment results of multi-frame sub-signals, including:

[0030] If it is determined that a sub-signal with a consecutive preset number of frames in a multi-frame sub-signal is a low-quality signal, the third feature extraction method is determined as the target feature extraction method.

[0031] The third feature extraction method includes third envelope extraction, third downsampling processing, third power spectrum calculation, and third frequency domain feature calculation. The third envelope extraction adopts the peak detection and connection method, the third power spectrum calculation adopts the Gossel algorithm with a third preset number of specific frequency detections, the sampling frequency of the third downsampling processing is less than the sampling frequency of the second downsampling processing, and the third preset number is less than the second preset number.

[0032] In one embodiment, feature extraction is performed on multiple frames of sub-signals according to a target feature extraction method to obtain multiple feature information, including:

[0033] The target feature extraction method is used to extract features from multiple frames of sub-signals following a preset number of consecutive frames, resulting in multiple feature information.

[0034] In one embodiment, based on the quality assessment results of all frame sub-signals and multiple feature information, anomalies in the target photoplethysmography signal are determined, including:

[0035] The target feature value is determined based on the feature information of each frame of sub-signal;

[0036] Based on the quality assessment results of all frame sub-signals and the target characteristic values, anomalies in the target photoplethysmography signal are determined.

[0037] In one embodiment, based on the quality assessment results of all frame sub-signals and the target feature value, anomalies in the target photoplethysmography signal are determined, including:

[0038] For each frame of sub-signal, if the sub-signal is a high-quality signal, the target feature value of the sub-signal is compared with the first feature threshold to obtain the first comparison result;

[0039] When the sub-signal is a general signal, the target feature value of the sub-signal is compared with the second feature threshold to obtain a second comparison result; if the first feature threshold is less than the second feature threshold,

[0040] When the sub-signal is a poor-quality signal, the target feature value of the sub-signal is compared with the third feature threshold to obtain the third comparison result; the second feature threshold is less than the third feature threshold.

[0041] Based on any one of the first comparison result, second comparison result, and third comparison result corresponding to all frame sub-signals, determine the abnormality of the target photoplethysmography signal.

[0042] In one embodiment, acquiring the photoplethysmography signal of the object to be detected includes:

[0043] Acquire the initial photoplethysmography signal of the object to be detected;

[0044] The initial photoplethysmography signal is preprocessed to obtain the target photoplethysmography signal; the preprocessing includes filtering and framing.

[0045] Secondly, one embodiment of this application provides a signal detection device, the device comprising:

[0046] The acquisition module is used to acquire the target photoplethysmography (PPG) signal of the object to be detected; the target PPG signal includes multiple frames of sub-signals.

[0047] The evaluation module is used to perform quality evaluation on each frame of sub-signal and obtain the quality evaluation result; the quality evaluation result includes any one of the following: the sub-signal is a high-quality signal, a normal signal, and a poor-quality signal;

[0048] The extraction module is used to extract features from the multi-frame sub-signals based on the quality assessment results of the multi-frame sub-signals, and obtain multiple feature information; each frame sub-signal corresponds to one feature information.

[0049] The determination module is used to determine the anomalies in the target photoplethysmography signal based on the quality assessment results of all frame sub-signals and multiple feature information.

[0050] Thirdly, one embodiment of this application provides a computer device including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps of the method provided in the first aspect above.

[0051] Fourthly, one embodiment of this application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the method provided in the first aspect above.

[0052] Fifthly, one embodiment of this application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the method provided in the first aspect above.

[0053] The aforementioned signal detection method, apparatus, computer equipment, storage medium, and program product acquire the target photoplethysmography (PPG) signal of the object to be detected. The target PPG signal includes multiple frames of sub-signals. Each frame of sub-signal undergoes quality evaluation to obtain a quality evaluation result. The quality evaluation result includes the sub-signal being classified as a high-quality signal, a general signal, or a low-quality signal. Based on the quality evaluation results of the multiple frames of sub-signals, features are extracted from the multiple frames of sub-signals to obtain multiple feature information. Each frame of sub-signal corresponds to one feature information. Based on the quality evaluation results of all frames of sub-signals and the multiple feature information, anomalies in the target PPG signal are determined. In this embodiment, by performing quality evaluation on each frame of the acquired target PPG signal, it is convenient to use different feature extraction methods to obtain feature information for subsequent judgment processes based on sub-signals of different qualities. This improves the reliability of the target PPG signal, thereby increasing the accuracy of determining anomalies in the target PPG signal and ultimately improving the accuracy of detecting atrial fibrillation events in the object to be detected. Attached Figure Description

[0054] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0055] Figure 1 This is a diagram illustrating the application environment of the signal detection method in one embodiment;

[0056] Figure 2 This is a flowchart illustrating the steps of a signal detection method in one embodiment;

[0057] Figure 3 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0058] Figure 4 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0059] Figure 5 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0060] Figure 6 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0061] Figure 7 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0062] Figure 8 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0063] Figure 9 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0064] Figure 10 This is a flowchart illustrating the steps of a signal detection method in another embodiment;

[0065] Figure 11 This is a schematic diagram of the signal detection device in one embodiment;

[0066] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0067] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0068] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the solutions, or any combination of multiple solutions, and the terms "at least one" or "at least one" refer to one of the solutions, or any combination of multiple solutions.

[0069] First, before introducing the technical solutions of the embodiments disclosed in this application, the background technology or technological evolution on which the embodiments of this application are based will be introduced. In the field of biomedical signal processing and wearable monitoring, atrial fibrillation (AF) is the most common and serious arrhythmia. It not only increases the risk of stroke by 5 times, but also induces heart failure, aggravates myocardial ischemia, and increases the mortality rate by 2 times. Therefore, it is necessary to detect AF in patients. Traditional AF diagnosis relies on electrocardiogram (ECG) examination, but ECG equipment has certain limitations. First, it requires operation by professional medical personnel and is not suitable for long-term continuous monitoring; second, the equipment is bulky, resulting in poor patient compliance; and finally, it cannot capture paroxysmal AF. AF detection technology based on photoplethysmography (PPG) has become a research hotspot due to its advantages such as non-invasiveness, low cost, and easy integration. PPG AF detection technology detects AF by processing and analyzing the patient's PPG signal. However, existing PPG atrial fibrillation detection methods suffer from poor signal reliability and low accuracy due to the susceptibility of PPG signals to factors such as motion artifacts, ambient light interference, and changes in skin blood perfusion. To address this, this application provides a signal detection method that processes PPG signals to improve the accuracy of atrial fibrillation detection.

[0070] The signal detection method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or located in the cloud or on other network servers. Terminal 102 can be, but is not limited to, various personal computers, laptops, smartphones, tablets, and portable wearable devices. Portable wearable devices can be smartwatches, smart bracelets, head-mounted devices, etc. Head-mounted devices can be virtual reality (VR) devices, augmented reality (AR) devices, smart glasses, etc. Server 104 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.

[0071] The technical solution of this application and how the technical solution of this application solves the technical problem are described in detail below with specific embodiments.

[0072] In one embodiment, such as Figure 2As shown, a signal detection method is provided, which can be applied to... Figure 1 Taking a terminal as an example, this method includes the following steps:

[0073] Step 200: Obtain the target photoplethysmography signal of the object to be detected; the target photoplethysmography signal includes multiple frames of sub-signals.

[0074] Photoplethysmography (PPG) signals can be acquired using a PPG sensor and transmitted to the terminal. The PPG sensor acquires the PPG signal of the subject by placing it on the finger, earlobe, or other suitable location. The sampling rate of the PPG sensor can be set from 100Hz to 256Hz to ensure the capture of detailed pulse waveforms of the subject.

[0075] In an optional embodiment, the sampling rate of the PPG sensor for acquiring PPG signals is set to 125Hz. The PPG sensor includes a light source (red light and infrared light) and a photodetector (i.e., a receiver). The light source emits red light and infrared light towards the object to be detected, and the photodetector receives the light signals after the red light and infrared light have passed through the object to be detected, which are used to obtain the absorption of red light and infrared light by the object to be detected, thus obtaining the PPG signal.

[0076] After acquiring the photoplethysmography (PPG) signal from the PPG sensor, the terminal can perform frame segmentation processing to obtain multiple sub-frames, i.e., the target PPG signal. This embodiment does not limit the specific method of frame segmentation processing, as long as it achieves the desired function.

[0077] Step 210: Perform quality assessment on each frame of sub-signal to obtain quality assessment results; the quality assessment results include any one of the following: sub-signal is a high-quality signal, a normal signal, or a low-quality signal.

[0078] After receiving multiple frames of sub-signals, the terminal performs a quality assessment on each frame to determine whether it is a high-quality signal, a normal signal, or a low-quality signal, thus obtaining the quality type of each frame. Under the same quality assessment standard, the highest quality is considered a high-quality signal, the medium quality is considered a normal signal, and the lowest quality is considered a low-quality signal.

[0079] In an optional embodiment, threshold ranges for different quality types can be preset. For each frame of sub-signal, a quality assessment value is obtained by performing a quality evaluation on the sub-signal. This quality assessment value is then compared with the threshold ranges, and the quality type corresponding to the threshold range in which the quality assessment value falls is determined as the quality type of the sub-signal. The quality assessment of the sub-signal can be performed using at least one of the following quality standards: complexity, consistency, and pulse rate stability. This embodiment does not limit the quality assessment standards or specific quality assessment methods, as long as the functionality is achieved.

[0080] Step 220: Based on the quality assessment results of the multi-frame sub-signals, perform feature extraction on the multi-frame sub-signals to obtain multiple feature information; each frame sub-signal corresponds to one feature information;

[0081] After determining the quality assessment results of multiple sub-frames, the terminal performs feature extraction on the sub-frames based on these results, obtaining the feature information corresponding to each sub-frame. In other words, different quality assessment results for the sub-frames require different complexity feature extraction methods; the higher the quality of the sub-frame, the more complex the feature extraction method used.

[0082] Step 230: Based on the quality assessment results of all frame sub-signals and multiple feature information, determine the abnormality of the target photoplethysmography signal.

[0083] Abnormalities in the target photoplethysmography (PPG) signal are used to characterize whether the subject under test has atrial fibrillation. An abnormal PPG signal indicates that the subject under test has atrial fibrillation, i.e., an atrial fibrillation event exists; a normal PPG signal indicates that the subject under test has not atrial fibrillation, i.e., no atrial fibrillation event exists.

[0084] After obtaining the quality assessment results and feature information of all frames of sub-signals in the target photoplethysmography (PPG), the terminal determines the anomalies in the target PPG signal based on the quality assessment results and multiple feature information. The anomalies in the target PPG signal are determined based on the anomalies of multiple consecutive frames of sub-signals. That is, if multiple consecutive frames of sub-signals are abnormal, the target PPG signal is determined to be abnormal, indicating that the object under test has atrial fibrillation. Different quality assessment results for sub-signals are used to compare the feature information of the sub-signal with different feature thresholds to determine the anomaly of that sub-signal.

[0085] In an optional embodiment, if the terminal determines that the target photoplethysmography signal is abnormal, it issues a prompt message to indicate that an atrial fibrillation event has occurred in the object being detected. The prompt message may be an alarm bell ringing, an alarm light illuminating, or both an alarm light illuminating and an alarm bell ringing; alternatively, it may be a pop-up prompt box on the terminal's display screen showing the words "Atrial fibrillation event present".

[0086] The signal detection method provided in this application involves acquiring the target photoplethysmography (PPG) signal of the object to be detected. The PPG signal includes multiple frames of sub-signals. Each frame of sub-signal undergoes a quality assessment to obtain a quality assessment result. The quality assessment result includes the sub-signal being classified as a high-quality signal, a general signal, or a low-quality signal. Based on the quality assessment results of the multiple frames of sub-signals, features are extracted from the multiple frames of sub-signals to obtain multiple feature information. Each frame of sub-signal corresponds to one feature information. Based on the quality assessment results of all frames of sub-signals and the multiple feature information, anomalies in the target PPG signal are determined. In this embodiment, by performing a quality assessment on each frame of the acquired target PPG signal, it is convenient to use different feature extraction methods to obtain feature information for subsequent judgment processes based on sub-signals of different qualities. This improves the reliability of the target PPG signal, thereby increasing the accuracy of determining anomalies in the target PPG signal and ultimately improving the accuracy of detecting atrial fibrillation events in the object to be detected. Furthermore, by employing feature extraction methods of varying complexity for sub-signals of different qualities, we can avoid using either highly complex or low-complexity methods for all frame sub-signals. This allows us to improve the accuracy of feature extraction while reducing its complexity. Compared to existing high-performance algorithms that rely on deep learning models or complex signal processing, this method requires fewer computational resources and enables real-time detection on power-constrained embedded devices, making the signal detection method more practical.

[0087] In one embodiment, such as Figure 3 As shown, this relates to an implementation method for quality assessment of multi-frame sub-signals to obtain quality assessment results. The steps of this implementation method include:

[0088] Step 300: For each sub-signal frame, determine the median of all pulse intervals in the sub-signal, and determine the first evaluation index based on each pulse interval and median in the sub-signal.

[0089] The pulse wave interval refers to the time interval between the peaks of the pulse wave in a sub-signal. Each frame of the sub-signal includes multiple pulse wave peaks, and therefore includes multiple pulse wave intervals. The median pulse wave interval is the pulse wave interval that is located in the middle position after all pulse wave intervals in the sub-signal are sorted by size.

[0090] The terminal determines a first evaluation index, i.e., an index for evaluating the time-domain signal of the sub-signal, based on each pulse interval and the median of all pulse intervals in the sub-signal. This first evaluation index characterizes the stability of the pulse rate corresponding to the sub-signal. This embodiment does not limit the specific methods for determining the pulse intervals and the median of all pulse intervals in the sub-signal, as long as the function is achieved.

[0091] In an optional embodiment, the specific method for determining the first evaluation quality includes: determining the peak position of the sub-signal; determining all pulse intervals in the sub-signal based on the peak position, thereby forming a pulse interval sequence; sorting all pulse intervals in the pulse interval sequence in ascending order; calculating the median of all pulse intervals; and calculating the average error between all pulse intervals in the pulse interval sequence and the median to obtain the first evaluation index.

[0092] For example, the terminal calculates the first-order difference of the sub-signal and determines the peak position of the sub-signal by using the positive zero-crossing point of the first-order difference. The formula for calculating the first-order difference is as follows: ,in, Represents the first-order difference value. This represents the pulse wave value at the nth sampling point in the sub-signal. This represents the pulse wave value at the (n-1)th sampling point in the sub-signal, where n is greater than or equal to 2 and less than or equal to N, and N is the total length of the sub-signal. The first evaluation metric can be expressed as: Where L represents the total length of the pulse wave interval sequence. Indicates the pulse wave interval sequence number 1 The interval between pulse waves This represents the median of all pulse intervals; This represents the average error.

[0093] Step 310: Determine the sample entropy of the sub-signal to obtain the second evaluation index.

[0094] Sample entropy (SE) is a nonlinear characteristic used to evaluate the complexity of a sub-signal. It primarily analyzes the complexity of a sub-signal at a set scale. Under normal circumstances, the complexity of a sub-signal is relatively stable. However, if the sub-signal is interfered with or poorly acquired, its complexity will change significantly. By determining the sample entropy of the sub-signal, changes in its complexity can be captured. The terminal obtains a second evaluation metric by determining the sample entropy of the sub-signal. This second evaluation metric characterizes the complexity of the sub-signal.

[0095] In an optional embodiment, the process of determining the sample entropy of the sub-signal includes: firstly resampling the sub-signal to obtain a first signal; extracting two sub-sequences of different lengths from the first signal; determining the number of similar elements in each sub-sequence, and determining the sample entropy based on the number of similar elements in each sub-sequence.

[0096] For example, the terminal resamples the sub-signal to obtain the signal at a set scale. , It is a scale factor. Assuming that sample entropy is used to capture changes in the sub-signal over a time scale of 2 to determine the complexity changes of the molecular signal, then... Option 2. The formula for resampling the sub-signal is expressed as: ,in, Represents the mesoscale of the first signal after resampling. The i-th element below, Let represent the j-th element in the sub-signal. The terminal extracts two subsequences (embedding dimensions) of length m and m+1 from the first signal. The two subsequences can be represented as follows: , For subsequences For each element in the array, count the number of elements whose distance to it is less than the tolerance value r, denoted as r. Calculate all average For subsequences For each element in the array, count the number of elements whose distance to it is less than the tolerance value r, denoted as r. Calculate all average The second evaluation metric, Samp_SQ, or sample entropy SampEn, can be expressed as... .

[0097] Step 320: Determine the peak information of the sub-signal and determine the third evaluation index based on the peak information.

[0098] The peak information of the sub-signal includes the peak amplitude and pulse width of the sub-signal's pulse wave. After receiving the sub-signal, the terminal calculates the peak information of the sub-signal and determines the third evaluation index based on this peak information. The third evaluation index is used to evaluate the consistency of the sub-signal.

[0099] In one embodiment, such as Figure 4 As shown, this involves an implementation method for determining a third evaluation index based on peak information. The steps of this implementation method include:

[0100] Step 400: Determine the maximum slope sequence corresponding to the peak information signal and calculate the variance of the maximum slope sequence to obtain the first variance.

[0101] The peak information includes the position of the pulse wave peaks of the sub-signal. The terminal obtains the maximum slope sequence by determining the maxima of the first-order difference between the pulse wave peaks in the sub-signal. The maximum slope sequence contains the first... Each element can be represented as ,in, This represents the x-th first-order difference value of the sub-signal. This represents the (x+1)th first-order difference value of the sub-signal. After obtaining the maximum slope sequence, the terminal calculates the variance of the maximum slope sequence to obtain the first variance. The first variance can be expressed as: ,in, It refers to the total number of elements in the sequence with the maximum slope. This represents the mean of all elements corresponding to the sequence with the maximum slope.

[0102] Step 410: Determine the pulse width sequence corresponding to the sub-signal based on the peak information, and calculate the variance of the pulse width sequence to obtain the second variance.

[0103] After determining the peak information of the sub-signal, the terminal determines the corresponding pulse width sequence based on this peak information. Specifically, the pulse width of the sub-signal is defined as the distance between the positions of half the peak amplitude on either side of the pulse wave peak. The pulse width sequence can be represented as Pw. K []. After determining the pulse width sequence corresponding to the sub-signal, the terminal calculates the variance of the pulse width sequence to obtain the second variance. The second variance can be expressed as: Where K refers to the total number of elements in the pulse width sequence. This represents the mean of all elements corresponding to the pulse width sequence. This represents the k-th element in the pulse width sequence.

[0104] Step 420: Determine the peak amplitude sequence corresponding to the sub-signal based on the peak information, and calculate the variance of the peak amplitude sequence to obtain the third variance.

[0105] After obtaining the peak information of the sub-signal, the terminal determines the peak amplitude sequence corresponding to the sub-signal based on the peak information, that is, it determines the peak amplitude of each pulse peak based on the peak information, and obtains the peak amplitude sequence Pa. F []. After obtaining the peak amplitude sequence, the terminal calculates the variance of the peak amplitude sequence to obtain the third difference. The third difference can be expressed as: Where F refers to the total number of elements in the peak amplitude sequence. This represents the mean of all elements corresponding to the pulse width sequence. This represents the f-th element in the peak amplitude sequence.

[0106] Step 430: Determine the third evaluation index based on at least one of the first variance, the second variance, and the third variance.

[0107] After obtaining the first variance, second variance, and third variance, the terminal determines a third evaluation index based on at least one of these three variances. The terminal can use any one of the first variance, second variance, and third variance as the third evaluation index. Alternatively, the terminal can obtain the third evaluation index by taking a weighted average of any two combinations of the first variance, second variance, and third variance.

[0108] In an optional embodiment, if a third evaluation index is determined based on the first variance, the second variance, and the third variance, then the third evaluation index can be expressed as follows: ,in, , and These are the weights for the first, second, and third variances, respectively. These weights can be set by the user based on experience. Optionally, = = =1 / 3.

[0109] Step 330: Determine the quality assessment result based on at least one of the first assessment indicator, the second assessment indicator, and the third assessment indicator.

[0110] After obtaining the first evaluation index, the second evaluation index, and the third evaluation index, the terminal determines the quality evaluation result based on at least one of the three evaluation indicators. The terminal can use any one of the three evaluation indicators as a target evaluation index and compare it with a first preset threshold and a second preset threshold to determine the quality evaluation result; the first preset threshold is greater than the second preset threshold. If the target evaluation index is greater than the first preset threshold, the quality evaluation result is determined to be a high-quality signal; if the target evaluation index is less than the first preset threshold but greater than the second preset threshold, the quality evaluation result is determined to be a normal signal; if the target evaluation index is less than the second preset threshold, the quality evaluation result is determined to be a low-quality signal.

[0111] The terminal can also take a weighted average of any two combinations of the first, second, and third evaluation indicators and use this as the target evaluation indicator. This average is then compared with the third and fourth preset thresholds to determine the quality evaluation result. The third preset threshold is greater than the fourth preset threshold. If the target evaluation indicator is greater than the third preset threshold, the sub-signal is determined to be a high-quality signal; if the target evaluation indicator is less than the third preset threshold but greater than the fourth preset threshold, the sub-signal is determined to be a normal signal; if the target evaluation indicator is less than the fourth preset threshold, the sub-signal is determined to be a poor-quality signal.

[0112] The terminal can also perform a weighted evaluation of the first, second, and third evaluation indicators, and use this as the target evaluation indicator. This target indicator is then compared with the fifth and sixth preset thresholds to determine the quality evaluation result. The fifth preset threshold is greater than the sixth preset threshold. If the target evaluation indicator is greater than the fifth preset threshold, the sub-signal is determined to be a high-quality signal; if the target evaluation indicator is less than the fifth preset threshold but greater than the sixth preset threshold, the sub-signal is determined to be a normal signal; if the target evaluation indicator is less than the sixth preset threshold, the sub-signal is determined to be a poor-quality signal. Specifically, the target evaluation indicator can be expressed as... ,in, , and These are the weights for the first, second, and third evaluation indicators, respectively. These weights can be set by the user based on experience. Optionally, = = =1 / 3.

[0113] In an optional embodiment, the first evaluation index, the second evaluation index, and the third evaluation index may be normalized before weighted summation.

[0114] In this embodiment, the final quality assessment result is obtained by determining a first assessment index, a second assessment index, and a third assessment index for the quality assessment of the sub-signal. This approach, which assesses the quality of the sub-signal from multiple dimensions (stability, complexity, and consistency of pulse rate), improves the accuracy of the sub-signal quality assessment and, consequently, the accuracy of identifying anomalies in the target photoplethysmography signal.

[0115] In one embodiment, such as Figure 5 As shown, this involves an implementation method for extracting features from multiple sub-frame signals based on the quality assessment results of these sub-frame signals to obtain multiple feature information. The steps of this implementation method include:

[0116] Step 500: Based on the quality assessment results of the multi-frame sub-signals, determine the target feature extraction method; the target feature extraction method includes envelope extraction, downsampling processing, power spectrum calculation, and frequency domain feature calculation.

[0117] After obtaining the quality assessment result of the sub-signal, the terminal determines the target feature extraction method for extracting features from the sub-signal based on the quality assessment result.

[0118] In an optional embodiment, the terminal pre-stores a correspondence between quality assessment results and feature extraction methods. After obtaining the quality assessment result of a sub-signal, the terminal determines the target feature extraction method for extracting features from the sub-signal based on this correspondence.

[0119] Step 510: Extract features from the multi-frame sub-signals using the target feature extraction method to obtain multiple feature information; each feature information includes at least one of the following: main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability.

[0120] Feature information refers to the frequency domain features of the sub-signal, including at least one of the following: dominant frequency energy proportion, harmonic ratio, spectral entropy, and dominant frequency instability. The dominant frequency energy proportion is used to characterize the energy concentration and purity of the sub-signal; the harmonic ratio is used to characterize the distortion of the pulse wave of the object being detected; the spectral entropy is used to characterize the disorder of the spectral distribution of the sub-signal; and the dominant frequency instability is used to characterize the stability of the dominant frequency of the sub-signal.

[0121] After obtaining the target feature extraction method corresponding to multiple frames of sub-signals, the terminal performs feature extraction on the multiple frames of sub-signals according to the target feature extraction method to obtain the feature information of the multiple frames of sub-signals. The target feature extraction method includes envelope extraction, downsampling processing, power spectrum calculation, and frequency domain feature calculation. That is to say, the process of the terminal extracting features from the sub-signals according to the target feature extraction method includes: first, extracting the envelope of the sub-signals to obtain the envelope signal; then, downsampling the envelope signal to obtain the downsampled signal; finally, determining the power spectrum of the downsampled signal, and calculating the frequency domain features corresponding to the sub-signals based on the power spectrum.

[0122] In this embodiment, after determining the quality assessment results of the multi-frame sub-signals, the target feature extraction method corresponding to the quality assessment results is determined to extract features from the multi-frame signals to obtain feature information. In this way, for the quality assessment results of the multi-frame sub-signals, a suitable feature extraction method corresponding to them can be selected to extract features from the multi-frame sub-signals, which can improve the accuracy of feature extraction from the multi-frame sub-signals. This can improve the accuracy of subsequent determination of anomalies in the target photoplethysmography signal, making the signal detection method more practical.

[0123] In one embodiment, such as Figure 6As shown, this relates to an implementation of a target feature extraction method based on the quality assessment results of multiple sub-frame signals. This implementation includes:

[0124] Step 600: If it is determined that a sub-signal with a consecutive preset number of frames in a multi-frame sub-signal is a high-quality signal, the first feature extraction method is determined as the target feature extraction method.

[0125] The first feature extraction method includes a first envelope extraction, a first downsampling process, a first power spectrum calculation, and a first frequency domain feature calculation; the first envelope extraction uses the Hilbert transform method, and the first power spectrum calculation uses the Gossel algorithm for detecting a first preset number of specific frequencies.

[0126] The preset number of frames can be pre-set by the user and stored in the terminal. Optionally, the preset number of frames is 10 frames. The first preset data is pre-set by the user and stored in the terminal. Optionally, the first preset number is 32. The envelope extraction method in the first feature extraction method is the first envelope extraction method, namely the Hilbert transform method, and the power spectrum calculation method is the first power spectrum calculation method, namely the Gossel algorithm using the first preset number of specific frequency detections.

[0127] After obtaining the quality assessment results of all frame sub-signals, if the terminal determines that the quality assessment results of the sub-signals for a consecutive preset number of frames are all high-quality signals, then the first feature extraction method is determined as the target feature extraction method.

[0128] Please continue reading Figure 6 In one embodiment, another implementation of the target feature extraction method is involved, based on the quality assessment results of multiple sub-frame signals. This implementation includes:

[0129] Step 610: If it is determined that the sub-signals of a consecutive preset number of frames in the multi-frame sub-signals are general signals, the second feature extraction method is determined as the target feature extraction method.

[0130] The second feature extraction method includes second envelope extraction, second downsampling processing, second power spectrum calculation, and second frequency domain feature calculation. The second envelope extraction adopts the moving root mean square method, the second power spectrum calculation adopts the Gossel algorithm with a second preset number of specific frequency detections, the sampling frequency of the second downsampling processing is less than the sampling frequency of the first downsampling processing, and the second preset number is less than the first preset number.

[0131] The second preset quantity can be pre-set by the user and stored in the terminal, and the second preset quantity is less than the first preset quantity. Optionally, the second preset quantity is 24. The envelope extraction method in the second feature extraction method is the second envelope extraction method, that is, the moving root mean square method, and the power spectrum calculation method is the second power spectrum calculation method, that is, the Gossel algorithm for specific frequency detection of the second preset quantity. The computational complexity of the second envelope extraction method is less than that of the first envelope extraction method, and the computational complexity of the second power spectrum calculation method is less than that of the first power spectrum calculation method. The sampling frequency of the second downsampling process is less than that of the first downsampling process. Optionally, the sampling frequency of the first downsampling process is 8Hz, and the sampling frequency of the second downsampling process is 5Hz.

[0132] Please continue reading Figure 6 In one embodiment, another implementation of the target feature extraction method is involved, based on the quality assessment results of multiple sub-frame signals. This implementation includes:

[0133] Step 620: If it is determined that a sub-signal with a consecutive preset number of frames in a multi-frame sub-signal is a low-quality signal, the third feature extraction method is determined as the target feature extraction method.

[0134] The third feature extraction method includes third envelope extraction, third downsampling processing, third power spectrum calculation, and third frequency domain feature calculation. The third envelope extraction adopts the peak detection and connection method, the third power spectrum calculation adopts the Gossel algorithm with a third preset number of specific frequency detections, the sampling frequency of the third downsampling processing is less than the sampling frequency of the second downsampling processing, and the third preset number is less than the second preset number.

[0135] The third preset quantity can be pre-set by the user and stored in the terminal, and the third preset quantity is less than the second preset quantity. Optionally, the third preset quantity is 16. The envelope extraction method in the third feature extraction method is the third envelope extraction method, that is, the peak detection and connection method. The power spectrum calculation method is the third power spectrum calculation method, that is, the Gossel algorithm with specific frequency detection of the third preset quantity. The computational complexity of the third envelope extraction method is less than that of the second envelope extraction method. The computational complexity of the third power spectrum calculation method is less than that of the second power spectrum calculation method. The sampling frequency of the third downsampling process is less than that of the second downsampling process. Optionally, the sampling frequency of the second downsampling process is 4Hz.

[0136] In this embodiment, when the quality assessment results of a series of sub-signals within a predetermined number of consecutive frames are all classified as high-quality, average, or low-quality, a corresponding target feature extraction method is determined. The target feature extraction method for high-quality signals has higher computational complexity, the method for average signals has medium computational complexity, and the method for low-quality signals has lower computational complexity. This avoids using either highly complex or low-complexity methods for all frames of sub-signals, thereby improving the accuracy of feature extraction while reducing its complexity. Furthermore, the higher the quality of the multi-frame sub-signals, the more computationally complex the target feature extraction method, which improves the reliability (confidence) of the obtained feature information, thus enhancing the reliability of identifying anomalies in the subsequently determined target photoplethysmography signal.

[0137] In one embodiment, an implementation method is provided for extracting features from multiple frames of sub-signals according to a target feature extraction method to obtain multiple feature information. This implementation method includes:

[0138] The target feature extraction method is used to extract features from multiple frames of sub-signals following a preset number of consecutive frames, resulting in multiple feature information.

[0139] After determining the target feature extraction method based on the quality assessment results of sub-signals for a consecutive preset number of frames, the terminal uses this target feature extraction method to extract features from each sub-signal after the preset number of frames, obtaining the corresponding feature information. Assuming that all q consecutive frames of sub-signals are high-quality signals, the target feature extraction method (first feature extraction method) corresponding to the high-quality signals is used to extract features from the (q+1)th frame of sub-signals.

[0140] In this embodiment, a defined target feature extraction method is used to extract features from multiple frames of sub-signals following a preset number of consecutive frames, resulting in multiple feature information. This improves the accuracy of feature extraction, thereby enhancing the accuracy of determining the abnormality of the target photoplethysmography signal based on the feature information, and ultimately enabling accurate detection of atrial fibrillation events in the target object.

[0141] In an optional embodiment, if the target feature extraction method is a first feature extraction method, then the process of using the first feature extraction method to extract features from each sub-signal in multiple frames of sub-signals following a consecutive preset number of sub-signals, and obtaining the corresponding feature information, includes:

[0142] (1) The envelope is extracted using the Hilbert transform method to obtain the first envelope signal;

[0143] Assuming that each subsequent sub-signal after a predetermined number of frames is y(t), and the Hilbert transform of this sub-signal is performed, the transform formula is as follows: ,in, Let H be the Hilbert analytic signal, and H be the Hilbert operator. The instantaneous amplitude of the Hilbert analytic signal is calculated to obtain the envelope curve, i.e., the first envelope signal. .

[0144] Envelope extraction using the Hilbert transform method yields an extremely smooth envelope signal with complete phase information, providing an ideal foundation for subsequent frequency domain analysis to obtain frequency domain features. The Hilbert transform can accurately extract the instantaneous amplitude and phase information of the pulse wave in the sub-signal, maintaining the temporal relationship of the pulse wave group. This is crucial for analyzing the variability of the pulse wave (RR) interval and the irregularity of heart rhythm. The core characteristic of atrial fibrillation is that irregular atrial excitation leads to irregular ventricular responses, and this irregularity is directly reflected in the fluctuation pattern of the envelope signal.

[0145] (2) The first envelope signal is resampled to obtain the first sampled signal.

[0146] The sampling rate corresponding to the first envelope signal is 125Hz. This first envelope signal is resampled (downsampled), reducing the sampling rate to 8Hz to obtain the first sampled signal. The choice of an 8Hz sampling rate is based on two considerations: firstly, an 8Hz sampling rate is sufficient to capture heart rate signals up to 240 bpm (corresponding to 4Hz), satisfying the Nyquist sampling theorem; secondly, compared to the standard 5Hz downsampling, 8Hz sampling provides higher temporal resolution, enabling more precise capture of instantaneous changes in heart rate. An 8Hz sampling rate satisfies the Nyquist sampling theorem requirement for a heart rate up to 240 bpm while avoiding redundant information from excessively high sampling rates. From a cardiovascular physiology perspective, heart rate typically does not exceed 200 bpm (3.3Hz), therefore an 8Hz sampling rate is sufficient to capture all heart rate variability information while reducing computational burden.

[0147] In an optional embodiment, an FIR anti-aliasing filter with a cutoff frequency of 3.5 Hz can be applied before resampling the first envelope signal to ensure no spectral aliasing.

[0148] (3) The first power spectrum is calculated by using the Goertzel algorithm with a first preset number of specific frequency detections to obtain the first power spectrum.

[0149] The Goertzel algorithm is essentially an IIR filter for a specific frequency point, whose frequency response peaks at that specific frequency. Since atrial fibrillation detection only concerns a limited frequency band within the heart rate band (0.5Hz-4Hz) and does not require calculation of the full spectrum, the Goertzel algorithm is more efficient than the Fourier transform.

[0150] The specific process of the Goertzel algorithm is as follows:

[0151] For each specific frequency Calculate intermediate variables ,in, The sampling rate is given by the initial condition. .

[0152] After processing T elements in the subsignal, at a specific frequency The power at that point can be expressed as:

[0153] .

[0154] The first power spectrum can be constructed based on multiple specific frequencies and their corresponding powers.

[0155] In one optional embodiment, the specific frequencies are selected by non-uniformly distributing 32 frequency points within the 0.5Hz-4Hz range. A denser distribution of frequency points (0.05Hz interval) is used in the low heart rate region (0.5Hz-1.2Hz), while a sparser distribution (0.1Hz-0.15Hz interval) is used in the high heart rate region (1.2Hz-4Hz). This better reflects the characteristics of heart rate distribution and improves frequency resolution at low heart rates.

[0156] After obtaining the first power spectrum, a fine post-processing is performed: First, a 5-point sliding window smoothing filtering algorithm based on polynomial fitting (Savitzky-Golay, SG) (polynomial order 2) is used to smooth the first power spectrum, suppressing random fluctuations while maintaining the shape of the spectral peaks; then, the background noise floor is estimated based on morphological operations, and the background noise floor is subtracted from the first power spectrum to enhance the significance of the effective components.

[0157] (4) Perform first frequency domain feature calculation on the first power spectrum to obtain feature information. Feature information includes the dominant frequency energy ratio, elastic harmonic ratio, normalized spectral entropy, and multi-scale dominant frequency instability.

[0158] 1) Dominant frequency energy spectrum proportion Calculation: Determine the frequency corresponding to the largest amplitude in the first power spectrum as the dominant frequency, and use this dominant frequency... Centered on the frequency band, the sum of all power spectra within a bandwidth of ±0.25Hz is determined, which is the integrated energy within the bandwidth of ±0.25Hz, thus obtaining the proportion of the dominant frequency energy spectrum. The proportion of the dominant frequency energy spectrum provides a more robust reflection of energy concentration. In normal sinus rhythm, cardiac electrical activity is highly synchronized, resulting in a sharp dominant peak in the PPG signal. During atrial fibrillation, atrial activity is disordered, and ventricular responses are irregular, leading to dispersed spectral energy and a decrease in dominant peak energy. Because heart rate itself experiences normal physiological fluctuations, calculating the dominant frequency band energy rather than single-point energy is more consistent with physiological reality.

[0159] 2) Elastic harmonic ratio Calculation: For the first power spectrum, at the precise second harmonic frequency... The actual maximum power point is searched within a range of ±0.2Hz. This maximum power point adapts to the harmonic frequency transformation represented by the actual heart rate fluctuation. The ratio between the power value at this power point and the power value of the dominant frequency is used as the elastic harmonic ratio. A normal pulse wave has a relatively fixed morphology, producing a stable harmonic structure. In atrial fibrillation, due to the irregular pulse wave (RR) intervals and the different hemodynamic states in each cardiac cycle, the pulse wave morphology varies, and the harmonic structure is disrupted. The elastic search mechanism adapts to the harmonic frequency changes during normal heart rate fluctuations.

[0160] 3) Standardized spectral entropy The calculation involves first smoothing the first power spectrum using Laplace (by adding a constant of 1e-9) to avoid zero-value problems. Then, the information entropy, or spectral entropy, is calculated using the Shannon entropy formula for the smoothed power spectrum. This spectral entropy is then standardized by dividing it by the theoretical maximum value log2(K) to obtain the standardized spectral entropy. Spectral entropy quantifies the regularity of heart rhythm. A low spectral entropy in a normal heart rhythm indicates relatively regular heartbeat intervals; an increased spectral entropy in atrial fibrillation reflects increased randomness in heartbeat intervals. Spectral entropy is a direct mathematical expression of the "absolutely irregular" characteristic of atrial fibrillation.

[0161] 4) Multi-scale frequency instability The calculation involves dividing each frame of sub-signal into multiple sub-windows according to a preset window. For example, if each frame of sub-signal is 30 seconds long, the preset window can be 5 seconds, and there can be 75% overlap when dividing according to the preset window. The dominant frequency of each sub-window is determined, and the standard deviation of the dominant frequency is calculated to obtain the multi-scale dominant frequency instability. The ventricular response in atrial fibrillation is not only irregular, but the degree of irregularity itself also changes over time. Multiscale analysis can capture this dynamic characteristic, improving the detection sensitivity of paroxysmal atrial fibrillation.

[0162] If the target feature extraction method is the second feature extraction method, then the process of using the second feature extraction method to extract features from each sub-signal in multiple frames of sub-signals following a consecutive preset number of frames, and obtaining the corresponding feature information, includes:

[0163] (1) The envelope is extracted by sampling moving root mean square method to obtain the second envelope signal.

[0164] The sliding window can be 400ms. The second envelope signal can be represented as: Where M is the total number of sampling points within the sliding window. The sliding root mean square (RMS) method for envelope extraction achieves a good balance between computational complexity and envelope quality, and has a natural suppression effect on moderate levels of noise.

[0165] (2) The second envelope signal is resampled to obtain the second sampled signal.

[0166] The second envelope signal has a sampling rate of 125Hz. Resampling (downsampling) the second envelope signal reduces the sampling rate to 5Hz, yielding the second sampled signal. The 5Hz sampling rate is a user-verified standard sampling rate, achieving an optimal balance between frequency resolution and computational complexity.

[0167] In an optional embodiment, an anti-aliasing filter with a cutoff frequency of 2.2 Hz can be applied before resampling the second envelope signal to ensure that the heart rate information is fully preserved.

[0168] (3) The second power spectrum is calculated by using the Goertzel algorithm with a second preset number of specific frequency detections to obtain the second power spectrum.

[0169] The specific process of the Goertzel algorithm is the same as described above, except for the selection data and method for specific frequencies. Specifically, 24 frequency points are evenly distributed within the range of 0.5Hz-3.5Hz, with a frequency resolution of approximately 0.125Hz.

[0170] After obtaining the second power spectrum, post-processing is performed: a simplified 3-point moving average smoothing is used: [0.25, 0.5, 0.25]. Background noise estimation is obtained by calculating the minimum value across the entire frequency band plus a standard deviation; this method is simple yet effective.

[0171] (4) Perform second frequency domain characteristic calculation on the second power spectrum to obtain characteristic information. The characteristic information includes the dominant frequency energy ratio, fundamental harmonic ratio, fundamental spectral entropy, and multi-scale dominant frequency instability.

[0172] 1) Dominant frequency energy spectrum proportion Calculation: Determine the ratio of the power value at the maximum power frequency point in the second power spectrum to the total power, and obtain the proportion of the main frequency energy spectrum. .

[0173] 2) Basic harmonic ratio Calculation: For the second power spectrum, the exact... The ratio between the power value at a given frequency and the power value at the main frequency is determined as the basic harmonic ratio.

[0174] 3) Fundamental spectral entropy The calculation is as follows: After normalizing the second power spectrum, the information entropy, i.e. the basic spectral entropy, is calculated using the Shannon entropy formula.

[0175] 4) Multi-scale frequency instability The calculation method is the same as that in the first feature extraction method, except that the window can overlap by 50%.

[0176] If the target feature extraction method is the third feature extraction method, then the process of using the third feature extraction method to extract features from each sub-signal in multiple frames of sub-signals following a consecutive preset number of frames, and obtaining the corresponding feature information, includes:

[0177] (1) The envelope is extracted using the peak detection and connection method to obtain the third envelope signal.

[0178] First, all peaks in the sub-signal are identified, and then linear interpolation is performed between adjacent peaks to obtain the third envelope signal. This envelope extraction method has low computational cost and is the most robust under extreme noise conditions.

[0179] (2) The third envelope signal is resampled to obtain the third sampled signal.

[0180] The sampling rate of the third envelope signal is 125Hz. Resampling (downsampling) the third envelope signal reduces the sampling rate to 4Hz, resulting in the second sampled signal. Although this reduces the frequency resolution, it improves the signal-to-noise ratio by decreasing the sampling rate.

[0181] (3) The third power spectrum is calculated by using the Goertzel algorithm with a third preset number of specific frequency detections to obtain the third power spectrum.

[0182] The specific process of the Goertzel algorithm is the same as described above, except for the selection data and method for specific frequencies. Furthermore, no further post-processing steps are required after obtaining the third power spectrum. Specifically, 16 frequency points are evenly distributed within the range of 0.5Hz-3.5Hz, with a frequency resolution of approximately 0.2Hz.

[0183] (4) Perform third frequency domain characteristic calculation on the third power spectrum to obtain characteristic information. The characteristic information includes the main frequency energy ratio, the fundamental harmonic ratio, the fundamental spectral entropy, and the main frequency instability.

[0184] For the calculation of feature information, refer to the specific process of calculating the second frequency domain feature in the above embodiments. In an optional embodiment, only the main frequency energy ratio and main frequency instability can be calculated.

[0185] In one embodiment, such as Figure 7 As shown, this invention relates to an implementation method for determining anomalies in a target photoplethysmography signal based on the quality assessment results of all frame sub-signals and multiple feature information. The steps of this implementation method include:

[0186] Step 700: Determine the target feature value based on the feature information of each frame sub-signal.

[0187] After acquiring the feature information of each frame of sub-signal, the terminal can determine the target feature value of the sub-signal based on the feature information. If the feature information of each frame of sub-signal includes any one of the following features: main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability, then that feature information is directly determined as the target feature value; if the feature information of each frame of sub-signal includes multiple features of the following features: main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability, then the target feature value is obtained by weighted averaging of the multiple features.

[0188] In an optional embodiment, if the feature information of each frame of sub-signal includes the main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability, then the target feature value is obtained by weighted averaging of these four features. The weight of each feature is related to the quality assessment result of that sub-signal.

[0189] Step 710: Based on the quality assessment results of all frame sub-signals and the target feature values, determine the abnormality of the target photoplethysmography signal.

[0190] After obtaining the target feature value of each frame of sub-signal, the terminal determines the abnormality of the target photoplethysmography signal based on the quality assessment results of multiple frames of sub-signal and the target feature value. In other words, the specific methods for determining the abnormality of the target photoplethysmography signal based on the target feature value are different depending on the quality assessment results of multiple frames of sub-signal.

[0191] In one embodiment, such as Figure 8 As shown, this invention relates to an implementation method for determining anomalies in the target photoplethysmography signal based on the quality assessment results of all frame sub-signals and target feature values. The steps of this implementation method include:

[0192] Step 800: For each frame sub-signal, if the sub-signal is a high-quality signal, compare the target feature value of the sub-signal with the first feature threshold to obtain the first comparison result.

[0193] The first feature threshold is preset by the user and stored in the terminal. Optionally, the first feature threshold is 0.75.

[0194] For each frame of sub-signal, if the terminal determines that the sub-signal is a high-quality signal based on the sub-signal quality assessment result, it compares the target feature value of the sub-signal with a first feature threshold to obtain a first comparison result. The first comparison result includes either the target feature value being less than the first feature threshold, or the target feature value being greater than or equal to the first feature threshold. Using the same method, the terminal can obtain the first comparison result corresponding to the high-quality signals in all frame sub-signals.

[0195] Step 810: When the sub-signal is a general signal, compare the target feature value of the sub-signal with the second feature threshold to obtain a second comparison result; the first feature threshold is less than the second feature threshold.

[0196] The second feature threshold is preset by the user and stored in the terminal. Optionally, the second feature threshold is 0.8.

[0197] For each frame of sub-signal, if the terminal determines that the sub-signal is a general signal based on the sub-signal quality assessment result, it compares the target feature value of the sub-signal with a second feature threshold to obtain a second comparison result. The second comparison result includes either the target feature value being less than the second feature threshold, or the target feature value being greater than or equal to the second feature threshold. Using the same method, the terminal can obtain the second comparison result corresponding to general signals in all frame sub-signals.

[0198] Step 820: When the sub-signal is a poor-quality signal, compare the target feature value of the sub-signal with the third feature threshold to obtain the third comparison result; the second feature threshold is less than the third feature threshold.

[0199] The third feature threshold is preset by the user and stored in the terminal. Optionally, the third feature threshold is 0.85.

[0200] For each frame of sub-signal, if the terminal determines that a sub-signal is a poor-quality signal based on its quality assessment results, it compares the target feature value of the sub-signal with a third feature threshold to obtain a third comparison result. The third comparison result includes cases where the target feature value is less than the third feature threshold, or where the target feature value is greater than or equal to the third feature threshold. Using the same method, the terminal can obtain the third comparison result corresponding to poor-quality signals in all frames of sub-signals.

[0201] Step 830: Determine the abnormality of the target photoplethysmography signal based on any one of the first comparison result, second comparison result, and third comparison result corresponding to all frame sub-signals.

[0202] The comparison results for all frame sub-signals include any one of the first comparison result, the second comparison result, and the third comparison result. Based on the comparison results for all frame sub-signals, the terminal can determine any anomalies in the target photoplethysmography (PPG) signal. Specifically, the terminal can determine the anomalies in the target PPG signal by determining the comparison results of consecutive target frame sub-signals. The comparison results of consecutive target frame sub-signals are related to the quality assessment results of those consecutive target frame sub-signals, and the target frame number is also related to the quality assessment results of the sub-signals. In other words, if all consecutive target frame sub-signals are high-quality signals, and the target frame number is the first frame number, then the comparison result is the first comparison result. If the first comparison results for all consecutive first frame sub-signals are that the target feature value is greater than or equal to the first feature threshold, then the target PPG signal is determined to be abnormal, indicating that the object under test has an atrial fibrillation event. If all consecutive target frame sub-signals in all frame sub-signals are normal signals, and the target frame number is the second frame number, then the comparison result is the second comparison result. If the second comparison results for all consecutive second frame sub-signals are that the target feature value is greater than or equal to the second feature threshold, then the target photoplethysmography (PPG) signal is determined to be abnormal, indicating that the object under test has atrial fibrillation. If all consecutive target frame sub-signals in all frame sub-signals are poor-quality signals, and the target frame number is the third frame number, then the comparison result is the third comparison result. If the third comparison results for all consecutive third frame sub-signals are that the target feature value is greater than or equal to the third feature threshold, then the target PPG signal is determined to be abnormal, indicating that the object under test has atrial fibrillation. Here, the first frame number is less than the second frame number, and the second frame number is less than the third frame number. Optionally, the first frame number is 2, the second frame number is 3, and the third frame number is 4.

[0203] In this embodiment, for each frame sub-signal, a first comparison result, a second comparison result, and a third comparison result can be determined based on its quality assessment result, target feature information, and corresponding feature thresholds. Anomalies in the target photoplethysmography (PPG) signal are determined based on the comparison results of sub-signals of consecutive target frames across all frame sub-signals. Thus, different methods exist for determining anomalies in the target PPG signal for sub-signals of different qualities, improving the accuracy and robustness of determining anomalies in the target PPG signal, thereby enhancing the accuracy and robustness of atrial fibrillation event detection.

[0204] In an optional embodiment, if the sub-signal is a high-quality signal, the characteristic information includes the proportion of main frequency energy. Elastic harmonic ratio Standardized spectral entropy and multi-scale frequency instability The target feature value corresponding to the sub-signal can be expressed as: ,in, , , and These are the main frequency energy percentages. Elastic harmonic ratio Standardized spectral entropy and multi-scale frequency instability The weighting. For example, in a high-quality signal mode, an aggressive judgment strategy is employed, with weighting biased towards features sensitive to atrial fibrillation. It is 0.15. It is 0.3. It is 0.35. The value is 0.2. The terminal will... The first comparison result is obtained by comparing the signal with the first feature threshold (0.75). If it is determined that the sub-signals of two consecutive frames are high-quality signals and the target feature value is greater than or equal to the first feature threshold, it is determined that there is an anomaly in the target photoplethysmography signal, indicating that the object under test has a transient atrial fibrillation event.

[0205] If the sub-signal is a general signal, the characteristic information includes the proportion of the main frequency energy. Elastic harmonic ratio Standardized spectral entropy and multi-scale frequency instability The target feature value corresponding to the sub-signal can be expressed as: ,in, , , and These are the main frequency energy percentages. Elastic harmonic ratio Standardized spectral entropy and multi-scale frequency instability The weights. For example, in general signal patterns, a linear interpolation mechanism using a quality assessment structure is employed to define the quality factor. Where a=0.4 and b=0.3. The SQI is mapped to the range [0,1], and the weights corresponding to each feature are calculated by interpolation. =0.25 (keep constant) =0.25+0.05q, =0.25+0.1q, =0.25-0.15q. When the difference between SQI and the third preset threshold is within the first range, it indicates that the sub-signal corresponding to SQI is close to a high-quality signal, with a weight of [0.25, 0.28, 0.32, 0.15]. When the difference between SQI and the fourth preset threshold is within the first range, it indicates that the sub-signal corresponding to SQI is close to a low-quality signal, with a balanced weight of [0.25, 0.25, 0.25, 0.25]. This ensures that the better the quality of the sub-signal, the more likely it is to use features sensitive to atrial fibrillation. The terminal will... The result is compared with the second feature threshold (0.8) to obtain the second comparison result. If it is determined that the sub-signals of three consecutive frames are all normal signals and the target feature value is greater than or equal to the second feature threshold, it is determined that there is an anomaly in the target photoplethysmography signal, indicating that the object under test has a transient atrial fibrillation event.

[0206] If the sub-signal is a poor-quality signal, the characteristic information can include only the two most robust features: the proportion of main frequency energy. and frequency instability The target feature value corresponding to the sub-signal can be expressed as: ,in, , , and These are the main frequency energy percentages. Elastic harmonic ratio Standardized spectral entropy and multi-scale frequency instability The weight, = =0. The weights of the features corresponding to poor-quality signals are conservatively configured. =0.2, =0.15, =0.15, With a weighting of 0.5, clinical observation has shown that, in noisy environments, irregular heart rhythm (dominant frequency instability) is the most reliable indicator of atrial fibrillation. The terminal will... The result is compared with the third feature threshold (0.85) to obtain the third comparison result. If it is determined that four or more consecutive sub-signals are normal signals and the target feature value is greater than or equal to the third feature threshold, it is determined that the target photoplethysmography signal is abnormal, indicating that the object under test has a transient atrial fibrillation event.

[0207] In one embodiment, such as Figure 9 As shown, one implementation method involves acquiring the photoplethysmography signal of an object to be detected. The steps of this implementation method include:

[0208] Step 900: Obtain the initial photoplethysmography signal of the object to be detected.

[0209] The initial photoplethysmography (PPG) signal can be a signal acquired using a PPG sensor. For a description of the PPG sensor acquiring the PPG signal, please refer to the specific description in the above embodiments; it will not be repeated here. The terminal acquires the initial PPG signal of the object to be detected.

[0210] Step 910: Preprocess the initial photoplethysmography signal to obtain the target photoplethysmography signal; the preprocessing includes filtering and framing.

[0211] After obtaining the initial photoplethysmography (PPG) signal, the terminal preprocesses it to obtain the target PPG signal. In other words, the terminal first performs smoothing filtering on the initial PPG signal, and then performs FIR bandpass filtering to remove high-frequency noise such as baseline interference and power frequency interference from the initial PPG signal. After completing the filtering of the initial PPG signal, the terminal performs frame processing on the filtered signal according to a preset time window (30s). The preset window sliding step size is 10 seconds, so each frame sub-signal has a two-thirds overlap.

[0212] In this embodiment, after acquiring the signal collected by the PPG sensor, filtering is performed to remove high-frequency noise, thereby improving the accuracy of subsequent processing based on the target photoplethysmography signal. Furthermore, by segmenting the filtered signal into frames, it is easier to judge each frame individually, improving the accuracy of identifying anomalies in the target photoplethysmography signal.

[0213] In one embodiment, such as Figure 10 As shown, a signal detection method is provided, the steps of which include:

[0214] Step 1001: Obtain the initial photoplethysmography signal of the object to be detected;

[0215] Step 1002: Filter and frame the initial photoplethysmography signal to obtain the target photoplethysmography signal; the target photoplethysmography signal includes multiple frames of sub-signals;

[0216] Step 1003: For each frame of sub-signal, determine the median of all pulse intervals in the sub-signal; determine the first evaluation index based on each pulse interval and the median of the sub-signal; obtain the second evaluation index based on the sample entropy of the sub-signal.

[0217] Step 1004: Determine the peak information of the sub-signal; determine the variance of the maximum slope sequence based on the peak information to obtain the first variance; determine the variance of the pulse width sequence based on the peak information to obtain the second variance; and determine the variance of the peak amplitude sequence based on the peak information to obtain the third variance.

[0218] Step 1005: Calculate the weighted average of the first variance, the second variance, and the third variance to obtain the third evaluation index;

[0219] Step 1006: Calculate the weighted average of the first evaluation indicator, the second evaluation indicator, and the third evaluation indicator to obtain the quality evaluation result;

[0220] Step 1007: If it is determined that the sub-signals of a consecutive preset number of frames in the multi-frame sub-signals are high-quality signals, the first feature extraction method is used to extract features from the multi-frame sub-signals after the consecutive preset number of frames to obtain the first feature information.

[0221] Step 1008: Determine the first target feature value based on the first feature information, compare the first target feature value with the first feature threshold, and obtain the first comparison result;

[0222] Step 1009: If it is determined that the sub-signals of a consecutive preset number of frames in the multi-frame sub-signals are general signals, the second feature extraction method is used to extract features from the multi-frame sub-signals after the consecutive preset number of frames to obtain the second feature information.

[0223] Step 1010: Determine the second target feature value based on the second feature information, compare the second target feature value with the second feature threshold, and obtain the second comparison result;

[0224] Step 1011: If it is determined that the sub-signals of a consecutive preset number of frames in the multi-frame sub-signals are inferior signals, the second feature extraction method is used to extract features from the multi-frame sub-signals after the consecutive preset number of frames to obtain the third feature information.

[0225] Step 1012: Determine the third target feature value based on the third feature information, compare the third target feature value with the third feature threshold, and obtain the third comparison result;

[0226] Step 1013: Determine the abnormality of the target photoplethysmography signal based on any one of the first comparison result, second comparison result, and third comparison result corresponding to all frame sub-signals; the abnormality of the target photoplethysmography signal is used to characterize the presence of atrial fibrillation event in the object to be detected.

[0227] In an optional embodiment, after signal detection of all frame sub-signals in the target photoplethysmography signal and determining any abnormalities in the target photoplethysmography signal, it can be determined whether the object under test has an atrial fibrillation event and the frequency of the atrial fibrillation event, thereby generating a visual report for the user to view.

[0228] In this embodiment, the feature thresholds and the weights corresponding to each feature can be set by the user according to the physiological differences of the subject's age, gender, underlying diseases, and different usage scenarios such as exercise, rest, and sleep. This makes the signal detection method more practical for individual differences and scenario differences of the subject.

[0229] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0230] Based on the same inventive concept, this application also provides a signal detection device for implementing the signal detection method described above. The solution provided by this device is similar to the implementation described in the above method; therefore, the specific limitations in one or more signal detection device embodiments provided below can be found in the limitations of the signal detection method described above, and will not be repeated here.

[0231] In one exemplary embodiment, such as Figure 11 As shown, a signal detection device is provided, comprising: an acquisition module 10, an evaluation module 11, an extraction module 12, and a determination module 13, wherein:

[0232] The acquisition module 10 is used to acquire the target photoplethysmography signal of the object to be detected; the target photoplethysmography signal includes multiple frames of sub-signals;

[0233] Evaluation module 11 is used to perform quality evaluation on each frame of sub-signal and obtain quality evaluation results; the quality evaluation results include any one of the following: the sub-signal is a high-quality signal, a normal signal, and a poor-quality signal;

[0234] Extraction module 12 is used to extract features from multiple sub-frame signals based on the quality assessment results of the multiple sub-frame signals to obtain multiple feature information; each sub-frame signal corresponds to one feature information;

[0235] The determination module 13 is used to determine the abnormality of the target photoplethysmography signal based on the quality assessment results of all frame sub-signals and multiple feature information.

[0236] In one embodiment, the evaluation module 11 includes a first determining unit. The first determining unit is configured to, for each frame of sub-signal, determine the median of all pulse intervals in the sub-signal; determine a first evaluation index based on each pulse interval and the median in the sub-signal; determine the sample entropy of the sub-signal to obtain a second evaluation index; determine the peak information of the sub-signal and determine a third evaluation index based on the peak information; and determine a quality evaluation result based on at least one of the first, second, and third evaluation indices.

[0237] In one embodiment, the first determining unit is specifically configured to: determine the maximum slope sequence corresponding to the sub-signal based on the peak information, and calculate the variance of the maximum slope sequence to obtain a first variance; determine the pulse width sequence corresponding to the sub-signal based on the peak information, and calculate the variance of the pulse width sequence to obtain a second variance; determine the peak amplitude sequence corresponding to the sub-signal based on the peak information, and calculate the variance of the peak amplitude sequence to obtain a third variance; and determine a third evaluation index based on at least one of the first variance, the second variance, and the third variance.

[0238] In one embodiment, the extraction module 12 includes a second determining unit and an extraction unit. The second determining unit is used to determine a target feature extraction method based on the quality assessment results of the multi-frame sub-signals; the target feature extraction method includes envelope extraction, downsampling processing, power spectrum calculation, and frequency domain feature calculation; the extraction unit is used to extract features from the multi-frame sub-signals according to the target feature extraction method to obtain multiple feature information; each feature information includes at least one of the following: dominant frequency energy ratio, harmonic ratio, spectral entropy, and dominant frequency instability.

[0239] In one embodiment, the second determining unit is specifically used to determine the first feature extraction method as the target feature extraction method when determining that a sub-signal with a consecutive preset number of frames in a multi-frame sub-signal is a high-quality signal; wherein, the first feature extraction method includes a first envelope extraction, a first downsampling process, a first power spectrum calculation, and a first frequency domain feature calculation; the first envelope extraction adopts the Hilbert transform method, and the first power spectrum calculation adopts the Gosser algorithm for a first preset number of specific frequency detections.

[0240] In one embodiment, the second determining unit is further configured to determine the second feature extraction method as the target feature extraction method when determining that a sub-signal with a consecutive preset number of frames in the multi-frame sub-signal is a general signal; wherein the second feature extraction method includes a second envelope extraction, a second downsampling process, a second power spectrum calculation, and a second frequency domain feature calculation; the second envelope extraction uses the moving root mean square method, the second power spectrum calculation uses the Gossel algorithm with a second preset number of specific frequency detections, the sampling frequency of the second downsampling process is less than the sampling frequency of the first downsampling process, and the second preset number is less than the first preset number.

[0241] In one embodiment, the second determining unit is further configured to determine the third feature extraction method as the target feature extraction method when determining that a sub-signal with a consecutive preset number of frames in the multi-frame sub-signal is a poor-quality signal; wherein the third feature extraction method includes a third envelope extraction, a third downsampling process, a third power spectrum calculation, and a third frequency domain feature calculation; the third envelope extraction adopts the peak detection and connection method, the third power spectrum calculation adopts the Gossel algorithm with a third preset number of specific frequency detections, the sampling frequency of the third downsampling process is less than the sampling frequency of the second downsampling process, and the third preset number is less than the second preset number.

[0242] In one embodiment, the extraction unit is specifically used to extract features from multiple frames of sub-signals following a consecutive preset number of sub-signals using a target feature extraction method, thereby obtaining multiple feature information.

[0243] In one embodiment, the determining module 13 includes a third determining unit. The third determining unit is used to determine the target feature value based on the feature information of each frame sub-signal; and to determine the abnormality of the target photoplethysmography signal based on the quality assessment results of all frame sub-signals and the target feature value.

[0244] In one embodiment, the third determining unit is specifically configured to, for each frame sub-signal, compare the target feature value of the sub-signal with a first feature threshold to obtain a first comparison result when the sub-signal is a high-quality signal; compare the target feature value of the sub-signal with a second feature threshold to obtain a second comparison result when the sub-signal is a normal signal; the first feature threshold is less than the second feature threshold; and compare the target feature value of the sub-signal with a third feature threshold to obtain a third comparison result when the sub-signal is a low-quality signal; the second feature threshold is less than the third feature threshold; and determine the abnormality of the target photoplethysmography signal based on any one of the first comparison result, the second comparison result, and the third comparison result corresponding to all frame sub-signals.

[0245] In one embodiment, the acquisition module 10 is specifically used to acquire the initial photoplethysmography signal of the object to be detected; preprocess the initial photoplethysmography signal to obtain the target photoplethysmography signal; the preprocessing includes filtering and framing.

[0246] Each module in the aforementioned signal detection device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0247] In one embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 12 As shown, the computer device includes a processor, memory, input / output interfaces, a communication interface, a display unit, and an input device. The processor, memory, and input / output interfaces are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interfaces. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interfaces are used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When executed by the processor, the computer program implements a signal detection method. The display unit is used to form a visually visible image and can be a display screen, a projection device, or a virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0248] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0249] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0250] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0251] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0252] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0253] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0254] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A signal detection method, characterized in that, The method includes: Acquire the target photoplethysmography signal of the object to be detected; the target photoplethysmography signal includes multiple frames of sub-signals; A quality assessment is performed on the sub-signal of each frame to obtain a quality assessment result; the quality assessment result includes any one of the sub-signal being a high-quality signal, a normal signal, and a poor-quality signal. Based on the quality assessment results of the sub-signals in multiple frames, feature extraction is performed on the sub-signals in multiple frames to obtain multiple feature information; each sub-signal in a frame corresponds to one feature information; the complexity of the feature extraction method corresponding to the high-quality signal is higher than the complexity of the feature extraction method corresponding to the general signal, and the complexity of the feature extraction method corresponding to the general signal is higher than the complexity of the feature extraction method corresponding to the poor-quality signal. Based on the quality assessment results of the sub-signals in all frames and the multiple feature information, the abnormality of the target photoplethysmography signal is determined; the abnormality of the target photoplethysmography signal is determined based on the abnormality of the sub-signals in multiple consecutive frames, and the abnormality of the sub-signal in each frame is determined by comparing the feature information and feature threshold of the sub-signal, with different quality assessment results corresponding to different feature thresholds.

2. The method according to claim 1, characterized in that, The quality of the sub-signals in multiple frames is evaluated to obtain the quality evaluation results, including: For each frame of the sub-signal, the median of all pulse intervals in the sub-signal is determined, and a first evaluation index is determined based on each pulse interval in the sub-signal and the median. The sample entropy of the sub-signal is determined to obtain the second evaluation index; Determine the peak information of the sub-signal, and determine the third evaluation index based on the peak information; The quality assessment result is determined based on at least one of the first assessment indicator, the second assessment indicator, and the third assessment indicator.

3. The method according to claim 2, characterized in that, The step of determining the third evaluation index based on the peak information includes: Based on the peak information, determine the maximum slope sequence corresponding to the sub-signal, and calculate the variance of the maximum slope sequence to obtain the first variance; Based on the peak information, determine the pulse width sequence corresponding to the sub-signal, and calculate the variance of the pulse width sequence to obtain the second variance; Based on the peak information, determine the peak amplitude sequence corresponding to the sub-signal, and calculate the variance of the peak amplitude sequence to obtain the third variance; The third evaluation index is determined based on at least one of the first variance, the second variance, and the third variance.

4. The method according to claim 1, characterized in that, Based on the quality assessment results of the multiple frames of the sub-signals, feature extraction is performed on the multiple frames of the sub-signals to obtain multiple feature information, including: Based on the quality assessment results of the sub-signals in multiple frames, a target feature extraction method is determined; the target feature extraction method includes envelope extraction, downsampling processing, power spectrum calculation, and frequency domain feature calculation. The target feature extraction method is used to extract features from multiple frames of the sub-signal to obtain multiple feature information; each feature information includes at least one of the following: main frequency energy ratio, harmonic ratio, spectral entropy, and main frequency instability.

5. The method according to claim 4, characterized in that, The step of determining the target feature extraction method based on the quality assessment results of multiple frames of the sub-signals includes: If it is determined that a sub-signal with a consecutive preset number of frames in the multiple frames is the high-quality signal, the first feature extraction method is determined as the target feature extraction method; The first feature extraction method includes a first envelope extraction, a first downsampling process, a first power spectrum calculation, and a first frequency domain feature calculation; the first envelope extraction uses the Hilbert transform method, and the first power spectrum calculation uses the Gossel algorithm with a first preset number of specific frequency detections.

6. The method according to claim 4, characterized in that, The step of determining the target feature extraction method based on the quality assessment results of multiple frames of the sub-signals includes: If it is determined that a sub-signal with a consecutive preset number of frames in the multiple frames is the general signal, the second feature extraction method is determined as the target feature extraction method; The second feature extraction method includes a second envelope extraction, a second downsampling process, a second power spectrum calculation, and a second frequency domain feature calculation. The second envelope extraction uses the moving root mean square method, the second power spectrum calculation uses the Gossel algorithm with a second preset number of specific frequency detections, the sampling frequency of the second downsampling process is less than the sampling frequency of the first downsampling process, and the second preset number is less than the first preset number.

7. The method according to claim 4, characterized in that, The step of determining the target feature extraction method based on the quality assessment results of multiple frames of the sub-signals includes: If it is determined that a sub-signal with a consecutive preset number of frames in the sub-signals of multiple frames is the inferior signal, the third feature extraction method is determined as the target feature extraction method; The third feature extraction method includes third envelope extraction, third downsampling processing, third power spectrum calculation, and third frequency domain feature calculation. The third envelope extraction adopts the peak detection and connection method, the third power spectrum calculation adopts the Gossel algorithm with a third preset number of specific frequency detections, the sampling frequency of the third downsampling processing is less than the sampling frequency of the second downsampling processing, and the third preset number is less than the second preset number.

8. The method according to any one of claims 5-7, characterized in that, The step of extracting features from multiple frames of the sub-signal using the target feature extraction method yields multiple feature information, including: The target feature extraction method is used to extract features from multiple frames of sub-signals following the consecutive preset number of frames to obtain the multiple feature information.

9. The method according to claim 4, characterized in that, The step of determining the abnormality of the target photoplethysmography signal based on the quality assessment results of the sub-signals of all frames and the multiple feature information includes: The target feature value is determined based on the feature information of the sub-signal in each frame; Based on the quality assessment results of the sub-signals in all frames and the target feature values, anomalies in the target photoplethysmography signal are determined.

10. The method according to claim 9, characterized in that, The step of determining the abnormality of the target photoplethysmography signal based on the quality assessment results of the sub-signals of all frames and the target feature value includes: For each sub-signal in a frame, if the sub-signal is a high-quality signal, the target feature value of the sub-signal is compared with a first feature threshold to obtain a first comparison result; When the sub-signal is a general signal, the target feature value of the sub-signal is compared with a second feature threshold to obtain a second comparison result; the first feature threshold is less than the second feature threshold. In the case where the sub-signal is a poor-quality signal, the target feature value of the sub-signal is compared with a third feature threshold to obtain a third comparison result; the second feature threshold is less than the third feature threshold. Based on any one of the first comparison result, second comparison result, and third comparison result corresponding to the sub-signals of all frames, determine the abnormality of the target photoplethysmography signal.

11. The method according to claim 1, characterized in that, The acquisition of the photoplethysmography signal of the object to be detected includes: Acquire the initial photoplethysmography signal of the object to be detected; The initial photoplethysmography signal is preprocessed to obtain the target photoplethysmography signal; the preprocessing includes filtering and framing.

12. A signal detection device, characterized in that, The device includes: The acquisition module is used to acquire the target photoplethysmography signal of the object to be detected; the target photoplethysmography signal includes multiple frames of sub-signals; An evaluation module is used to perform quality evaluation on the sub-signal of each frame and obtain a quality evaluation result; the quality evaluation result includes the sub-signal being any one of a high-quality signal, a normal signal, and a poor-quality signal. The extraction module is used to extract features from the sub-signals in multiple frames based on the quality assessment results of the sub-signals in multiple frames, and obtain multiple feature information; each frame of the sub-signal corresponds to one feature information; the complexity of the feature extraction method corresponding to the high-quality signal is higher than the complexity of the feature extraction method corresponding to the general signal, and the complexity of the feature extraction method corresponding to the general signal is higher than the complexity of the feature extraction method corresponding to the low-quality signal. The determination module is used to determine the abnormality of the target photoplethysmography signal based on the quality assessment results of the sub-signals in all frames and the multiple feature information; the abnormality of the target photoplethysmography signal is determined based on the abnormality of the sub-signals in multiple consecutive frames, and the abnormality of the sub-signal in each frame is determined by comparing the feature information and feature threshold of the sub-signal, with different quality assessment results corresponding to different feature thresholds.

13. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 11.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 11.

Images