A digital electrocardio collection method based on signal pattern recognition

Abstract

The application discloses a digital electrocardio collection method based on signal pattern recognition, relates to the technical field of electrocardio signal intelligent collection, and comprises the following steps: collecting a human electrocardio analog signal, amplifying, filtering and analog-digital converting the human electrocardio analog signal to generate an original digital electrocardio data frame; performing time alignment, lead coding, amplitude normalization and electrode state coding on the original digital electrocardio data frame to obtain to-be-recognized electrocardio data and extract electrocardio collection state features; and performing signal pattern recognition on the electrocardio collection state features to obtain a ternary signal pattern state composed of an electrocardio state, a noise state and an electrode state; and the application realizes dynamic adjustment of collection parameters and feature output states, reduces the influence of noise interference and lead falling on electrocardio results, and improves the reliability of digital electrocardio collection information.

Description

Technical Field

[0001] This invention relates to the field of intelligent electrocardiogram (ECG) signal acquisition technology, and in particular to a digital ECG acquisition method based on signal pattern recognition. Background Technology

[0002] With the development of biomedical sensors, low-power analog front-ends, embedded processors, and physiological signal processing algorithms, digital ECG acquisition technology has been widely applied in bedside monitoring, Holter monitoring, wearable ECG detection, and telemedicine. Existing digital ECG acquisition systems typically acquire weak analog ECG signals from the human body surface via ECG electrodes. After pre-amplification, filtering, power frequency suppression, and analog-to-digital conversion, digital ECG data is generated and further encapsulated by combining lead identification, sampling frequency, and timestamp information. In subsequent processing, the system generally performs noise reduction, feature extraction, and anomaly detection on the ECG waveform based on fixed filtering parameters, preset threshold rules, or single signal characteristics. In recent years, with the application of pattern recognition technology in physiological signal analysis, some ECG acquisition systems have begun to explore using signal pattern recognition methods to analyze ECG waveform morphology, noise characteristics, or acquisition status to improve the ability to identify ECG signal quality and waveform status in complex acquisition environments.

[0003] In actual ECG acquisition, the acquisition quality is affected not only by the waveform state of the ECG itself, but also by a combination of factors such as power frequency interference, baseline drift, electromyographic noise, poor electrode contact, lead detachment, and motion artifacts. Current technologies, particularly signal pattern recognition, primarily focus on individually identifying ECG waveform abnormalities or noise types, failing to adequately utilize the correlation between ECG state, noise state, and electrode state. Therefore, how to comprehensively determine the ECG state, noise state, and electrode state based on signal pattern recognition during digital ECG acquisition, and how to coordinate the acquisition parameters and output state of the next acquisition window based on the recognition results, is a crucial technical challenge for improving the stability of ECG acquisition and the reliability of digital ECG data. Summary of the Invention

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

[0005] Therefore, this invention provides a digital electrocardiogram (ECG) acquisition method based on signal pattern recognition to solve the problems of insufficient accuracy in the acquisition process, such as the lack of unified identification and linkage control of ECG state, noise state, and electrode state, as well as the lack of accuracy in the adjustment of acquisition parameters and the control of ECG feature output.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: This invention provides a digital electrocardiogram (ECG) acquisition method based on signal pattern recognition, comprising: The system acquires simulated human electrocardiogram (ECG) signals, amplifies, filters, and performs analog-to-digital conversion to generate raw digital ECG data frames. These raw digital ECG data frames are then time-aligned, lead-coded, amplitude-normalized, and electrode-state-coded to obtain the ECG data to be identified. ECG acquisition state features are extracted, and signal pattern recognition is performed on these features to obtain a three-dimensional signal pattern state composed of ECG state, noise state, and electrode state. Acquisition control parameters are generated based on this three-dimensional signal pattern state. When the ECG state indicates a suspected ECG abnormality, and the noise state indicates the presence of noise interference, deterministic ECG abnormalities are suppressed. The system outputs normal conclusions and simultaneously outputs abnormal confirmation markers under noise interference, while retaining the original ECG data and acquisition window identifier. It also generates enhanced acquisition control parameters. When the noise status indicates the presence of power frequency interference and baseline drift, it generates filter adjustment parameters. When the electrode status indicates lead detachment, it generates control parameters to pause ECG feature output. Based on the enhanced acquisition control parameters, filter adjustment parameters, and control parameters to pause ECG feature output, it generates ECG acquisition execution control parameters for the next acquisition window and executes ECG acquisition according to the ECG acquisition execution control parameters in the next acquisition window, generating digital ECG acquisition information.

[0007] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the steps of acquiring human ECG analog signals and amplifying, filtering, and performing analog-to-digital conversion on the human ECG analog signals to generate raw digital ECG data frames are as follows: Impedance matching and differential amplification are performed on the ECG analog signal to obtain a pre-amplified signal. The pre-amplified signal is then combined with the quality index for adaptive filtering and baseline correction to obtain a clean analog sequence. Quantization and frame structure encapsulation are performed on the pure analog sequence to obtain the original digital ECG data frame.

[0008] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps of performing time alignment, lead encoding, amplitude normalization, and electrode state encoding on the original digital ECG data frames to obtain the ECG data to be identified and extracting ECG acquisition state features are as follows: Extract ECG acquisition frame parsing information from the original digital ECG data frames, and construct a sample matrix to be aligned based on the frame number and sampling point number in the ECG acquisition frame parsing information; A unified time reference is determined by calculating the slope energy centroid position of the lead sample values ​​in the ECG acquisition frame parsing information; based on the unified time reference, the sample matrix to be aligned is time-aligned, and the aligned sample values ​​are subjected to lead encoding and amplitude normalization to obtain the lead data sequence; Electrode status codes are generated based on the electrode impedance values, saturation sampling markers, and horizontal sampling markers in the ECG acquisition frame parsing information; the lead data sequence and electrode status codes are combined to form ECG data to be identified, and ECG acquisition status features are extracted from the ECG data to be identified.

[0009] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps for performing signal pattern recognition on the ECG acquisition state characteristics to obtain a three-element signal pattern state composed of ECG state, noise state, and electrode state are as follows: The ECG acquisition status features are aligned and combined and encoded according to the feature arrangement rules to construct a multimodal ECG feature vector; Based on multimodal ECG feature vectors, the confidence scores for ECG state, noise interference, and electrode contact state are calculated using a set of state discrimination rules. The ECG state confidence scores are compared with the ECG abnormality judgment threshold range to determine whether the ECG state is normal or suspected abnormal. The noise interference confidence scores are compared with the interference intensity threshold to classify the noise state into no interference, power frequency interference, baseline drift, and composite interference. The confidence level of electrode contact status is compared with the threshold range of electrode contact status to determine the electrode status as good fit, high impedance, or lead detachment. The determined ECG status, noise status, and electrode status are combined and mapped according to the status coding protocol to generate a ternary signal pattern status.

[0010] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps for generating acquisition control parameters according to the three-element signal pattern state are as follows: Based on the combination of confidence levels of ECG status, noise interference, and electrode contact status, a preset control strength rule table is queried to form the acquisition control strength. When any of the following conditions is met: suspected ECG abnormality, enhanced noise interference, or enhanced abnormal electrode contact, the acquisition control strength is increased. Configure power frequency notch adjustment parameters when the noise condition includes power frequency interference, and configure baseline correction parameters and high-pass filter adjustment parameters when the noise condition includes baseline drift; perform output gating judgment on ECG status, noise status and electrode status; When the ECG status is suspected to be abnormal and the noise status indicates the presence of noise interference, the ECG abnormality result is set to the suppressed output state. When the electrode status is lead detachment, the ECG feature is set to the paused output state. The acquisition control parameters are generated by combining the acquisition control strength, power frequency notch adjustment parameters, baseline correction parameters, and high-pass filter adjustment parameters.

[0011] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, when the ECG state is suspected to be abnormal and the noise state indicates the presence of noise interference, the output of a definitive ECG abnormality conclusion is suppressed, while an abnormality pending confirmation marker under noise interference is output, the original ECG data and acquisition window identifier are retained, and enhanced acquisition control parameters are generated. The specific steps are as follows: When the ECG status is suspected to be abnormal and the noise status is one of power frequency interference, baseline drift, and combined interference, the confidence level of the ECG status, the confidence level of the noise interference, and the acquisition control strength are obtained. The confidence level of the ECG status, the confidence level of the noise interference, and the acquisition control strength are divided into intervals to obtain the ECG abnormality confidence level, the noise interference confidence level, and the acquisition control strength level. The ECG abnormality confidence level, the noise interference confidence level, and the acquisition control strength level are then combined to form an abnormal output gating index. Based on the abnormal output gating index, the abnormal output gating rule table is queried to determine the abnormal output suppression level, and an abnormal output suppression flag is generated according to the abnormal output suppression level; the noise state, noise interference confidence level and acquisition control strength level are combined to form a sampling rate adjustment index, and the preset sampling rate adjustment rule table is queried based on the sampling rate adjustment index to obtain the sampling rate enhancement level and generate the sampling rate enhancement parameters for the next acquisition window; The amplification gain correction direction is determined based on the lead sampling value, saturation sampling mark, and horizontal sampling mark, and the amplification gain correction parameters are generated based on the amplification gain correction direction and the acquisition control intensity level. The filter compensation type is determined based on the noise status, and filter compensation parameters are generated based on the filter compensation type, noise interference confidence level, and acquisition control strength level. The abnormal output suppression flag, sampling rate boosting parameter, amplification gain correction parameter, and filter compensation parameter are combined into enhanced acquisition control parameters.

[0012] As a preferred embodiment of the digital ECG acquisition method based on signal pattern recognition described in this invention, the step of generating filter adjustment parameters when the noise state indicates the presence of power frequency interference and baseline drift includes the following specific steps: The power frequency interference state and the baseline drift state are used as periodic noise evidence anchor points and low frequency drift evidence anchor points, respectively, and connected to the noise interference feature node and the filter parameter control node in the signal mode relationship diagram to form a noise filtering relationship sub-graph. Noise interference characterization parameters are extracted from the noise filtering relationship subgraph, and power frequency interference control weights and baseline drift control weights are generated. Based on the power frequency interference control weights, notch center frequency, notch bandwidth and notch suppression depth are generated. Based on the baseline drift control weights, high-pass cutoff frequency and baseline correction step size are generated and combined into filter adjustment parameters.

[0013] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps for generating control parameters to pause ECG feature output when the electrode state indicates lead detachment are as follows: The lead detachment state is used as the anchor point for detachment evidence and connected to the electrode contact feature node, lead validity node and ECG feature output control node in the signal pattern relationship diagram to form a lead detachment control sub-diagram. Extract lead detachment judgment parameters from the lead detachment control subgraph within the current acquisition window, generate lead detachment control weights, and mark leads whose lead detachment control weights reach the detachment pause threshold as invalid leads. The output path of the ECG waveform characteristics corresponding to the invalid lead is blocked, the output state of the ECG characteristics is set to paused, and the invalid lead identifier, paused output marker and recovery detection condition are combined into control parameters for pausing the output of the ECG characteristics.

[0014] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps for generating ECG acquisition execution control parameters for the next acquisition window based on enhanced acquisition control parameters, filtering adjustment parameters, and control parameters for pausing ECG feature output are as follows: ECG acquisition and control parameters are extracted from the acquisition control parameters, and control parameters are fused in the order of pause output marker, filter adjustment parameters, and enhanced acquisition control parameters in the ECG acquisition and control parameters. For control parameters that have not been generated, they will not participate in control parameter fusion, and the corresponding sampling rate, amplification gain, filtering parameters and ECG feature output state will remain unchanged, and the ECG acquisition execution control parameters for the next acquisition window will be generated. Write the ECG acquisition execution control parameters into the acquisition configuration item of the next acquisition window, and control the ECG feature output according to the pause output flag.

[0015] As a preferred embodiment of the digital electrocardiogram (ECG) acquisition method based on signal pattern recognition described in this invention, the specific steps for performing ECG acquisition according to ECG acquisition execution control parameters in the next acquisition window to generate digital ECG acquisition information are as follows: Based on the acquisition configuration items of the next acquisition window, the driving sampling clock, amplification gain structure and filtering processing structure operate according to the updated sampling rate, amplification gain and filtering parameters, and open, suppress and pause the ECG feature output channel according to the ECG feature output status. Acquire lead data within the adjusted next acquisition window, and record the actual sampling rate, actual amplification gain, filtering parameters, lead status, and ECG characteristic output status to form an ECG data frame with control identifiers. Encapsulate the ECG data frame with control identifiers, the three-element signal mode status, acquisition control parameters, and acquisition timestamps to generate digital ECG acquisition information.

[0016] The beneficial effects of this invention are as follows: By performing time alignment, lead encoding, amplitude normalization, and electrode state encoding on the original digital ECG data frames, and by performing signal pattern recognition based on ECG acquisition state characteristics, the collaborative discrimination of ECG state, noise state, and electrode state is achieved, which improves the accuracy of acquisition state recognition in complex acquisition environments; based on the three-element signal pattern state, enhanced acquisition control parameters, filtering adjustment parameters, and pause output control parameters are generated and fused to form the ECG acquisition execution control parameters for the next acquisition window, realizing the dynamic adjustment of acquisition parameters and characteristic output state, thereby reducing the impact of noise interference and lead dropout on ECG results and improving the reliability of digital ECG acquisition information. Attached Figure Description

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

[0018] Figure 1 This is a flowchart of a digital electrocardiogram (ECG) acquisition method based on signal pattern recognition.

[0019] Figure 2 A flowchart for generating the ternary signal mode state.

[0020] Figure 3 A flowchart for generating enhanced acquisition and control parameters.

[0021] Figure 4 A flowchart for generating control parameters for ECG acquisition. Detailed Implementation

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

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

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

[0025] Reference Figures 1-4 This is one embodiment of the present invention, which provides a digital electrocardiogram (ECG) acquisition method based on signal pattern recognition, comprising the following steps: S1. Acquire human electrocardiogram (ECG) analog signals, and amplify, filter, and perform analog-to-digital conversion on the human ECG analog signals to generate raw digital ECG data frames.

[0026] Impedance matching and differential amplification are performed on the ECG analog signal to obtain a pre-amplified signal. The pre-amplified signal is then combined with the quality index for adaptive filtering and baseline correction to obtain a clean analog sequence.

[0027] Furthermore, at least one of the following features is extracted from the pre-amplification signal: electrode contact stability, amplitude stability, saturation distortion, flat line anomaly, power frequency interference, and baseline drift. A quality index is generated to characterize the usability and interference level of the signal. Based on the quality index, the parameters of power frequency notch filtering, low-pass filtering, high-pass filtering, or baseline correction are adjusted, and the low-frequency trend component is extracted from the filtered signal for baseline compensation to obtain a clean analog sequence for quantization and frame structure encapsulation.

[0028] Quantization and frame structure encapsulation are performed on the pure analog sequence to obtain the original digital ECG data frame.

[0029] Furthermore, quantization processing is performed on the pure analog sequence, and lead sampling values ​​are organized according to lead number. When the lead sampling value reaches the quantization boundary or a clipping state occurs, a saturation sampling marker is generated. In a continuous preset number of sampling points, when the amplitude difference between adjacent sampling points is less than the preset horizontal amplitude threshold, and the rate of change of the sampling value is lower than the preset lower limit of the rate of change, a horizontal sampling marker is generated. At the same time, the electrode impedance value is bound to the lead number. The frame number, sampling frequency, lead number, lead sampling value, electrode impedance value, saturation sampling marker, horizontal sampling marker, and sampling point number are encapsulated to obtain the original digital ECG data frame.

[0030] S2. Perform time alignment, lead encoding, amplitude normalization, and electrode state encoding on the original digital ECG data frame to obtain the ECG data to be identified and extract the ECG acquisition state features. Perform signal pattern recognition on the ECG acquisition state features to obtain a three-element signal pattern state composed of ECG state, noise state, and electrode state.

[0031] Extract ECG acquisition frame parsing information from the original digital ECG data frames, and construct a sample matrix to be aligned based on the frame number and sampling point number in the ECG acquisition frame parsing information.

[0032] Furthermore, the original digital ECG data frames are parsed to obtain ECG acquisition frame parsing information, including frame number, sampling frequency, lead number, lead sample value, electrode impedance value, saturation sampling marker, horizontal sampling marker, and sampling point number. A global sampling index is formed based on the frame number and sampling point number, and the lead sample values ​​are grouped according to the lead number to construct a sample matrix to be aligned, with the lead number as the row index and the global sampling index as the column index. The electrode impedance value, saturation sampling marker, and horizontal sampling marker are used as auxiliary information of the acquisition status and are associated with the sample matrix to be aligned according to the lead number and the global sampling index. Missing sampling markers or empty markers are configured for positions where the frame number is missing, the sampling point number is discontinuous, or the lead sample value is invalid.

[0033] It should be noted that the ECG acquisition frame parsing information includes frame number, sampling frequency, lead number, lead sampling value, electrode impedance value, saturation sampling marker, horizontal sampling marker, and sampling point number.

[0034] A unified time reference is determined by calculating the slope energy centroid position of the lead sample values ​​in the ECG acquisition frame parsing information; based on the unified time reference, the sample matrix to be aligned is time-aligned, and the aligned sample values ​​are subjected to lead encoding and amplitude normalization to obtain the lead data sequence.

[0035] Furthermore, based on the lead number and sampling point number, the sampled values ​​of each lead are arranged in the sampling order to form a sampled value sequence, and a slope energy sequence is generated according to the amplitude change of adjacent sampling points. Using the sampling point number as the position index and the slope energy as the weight, the slope energy centroid position of each lead is obtained. Combined with the electrode impedance value, saturation sampling mark, and horizontal sampling mark, invalid leads and abnormal sampling points are eliminated and weighted, and the slope energy centroid positions of the valid leads are merged to obtain a unified time reference. When the slope energy is insufficient, the centroid positions of other valid leads and the time positions corresponding to the frame number and sampling point number are used as compensation references.

[0036] The time offset of each lead's sampled value is determined based on a unified time reference. The lead sampled value sequence is then processed by at least one of the following methods: translation, padding, and resampling, to align the sampled values ​​corresponding to the same ECG waveform phase in different leads. The aligned sampled values ​​are then encoded using lead encoding. The amplitude scaling factor and offset are determined by combining the amplitude range, mean, and preset reference amplitude. The sampled values ​​are then mapped to a unified amplitude range and arranged according to lead number and time order to obtain the lead data sequence.

[0037] Electrode status codes are generated based on the electrode impedance values, saturation sampling markers, and horizontal sampling markers in the ECG acquisition frame parsing information; the lead data sequence and electrode status codes are combined to form ECG data to be identified, and ECG acquisition status features are extracted from the ECG data to be identified.

[0038] Furthermore, the electrode impedance value, saturation sampling mark, and horizontal sampling mark are extracted according to the lead number and associated with the corresponding lead sampling value. The electrode impedance value is compared with the preset impedance state range to obtain the initial electrode contact state. Then, it is corrected by combining the continuous saturation sampling mark and the horizontal sampling mark, and the electrode state code is generated according to the preset coding rules.

[0039] The lead data sequence is arranged according to lead number and time order to form the ECG waveform data corresponding to each lead. The electrode status code of the same lead is bound to the ECG waveform data to obtain the ECG data to be identified. ECG acquisition status features are extracted from the ECG data to be identified. The ECG acquisition status features are used to characterize the ECG waveform quality, noise interference and electrode contact status within the current acquisition window.

[0040] It should be noted that ECG acquisition status features refer to the set of features extracted from the ECG data to be identified, used to characterize the quality of the ECG waveform, noise interference, and electrode contact status within the current acquisition window.

[0041] The ECG acquisition status features are aligned and combined and encoded according to the feature arrangement rules to construct a multimodal ECG feature vector.

[0042] Furthermore, based on the acquisition window, lead number, and sampling time position, a correspondence is established between ECG waveform features, noise interference features, and electrode contact features, so that different types of features corresponding to the same lead or the same time period within the same acquisition window can be grouped into the same feature combination. For features with inconsistent sampling point numbers, feature dimensions, or time positions, interpolation, padding, truncation, or window summarization methods are used to unify them to the same length and the same time scale, resulting in an aligned feature set. The aligned feature set is arranged according to the feature type order, and the arranged feature values, feature type identifiers, and lead identifiers are combined and encoded to construct a multimodal ECG feature vector.

[0043] It should be noted that the feature arrangement rule refers to the rule of organizing features in a fixed order, such as arranging them in sequence according to lead number, sampling time, and feature type, so that the multimodal ECG feature vectors generated by different acquisition windows have a consistent data order.

[0044] Based on multimodal ECG feature vectors, the confidence scores of ECG state, noise interference, and electrode contact state are calculated using a set of state discrimination rules. The ECG state confidence scores are compared with the ECG abnormality judgment threshold range to determine whether the ECG state is normal or suspected abnormal. The noise interference confidence scores are compared with the interference intensity threshold to classify the noise state into no interference, power frequency interference, baseline drift, and composite interference.

[0045] Furthermore, based on multimodal ECG feature vectors, ECG waveform features, noise interference features, and electrode contact features are formed respectively, and these features are input into a preset state discrimination rule set. The preset state discrimination rule set judges the degree of ECG waveform abnormality, noise interference intensity, and electrode contact effectiveness, respectively, and outputs the original ECG score, original noise score, and original electrode contact score. For each type of original score, the portion below the lower limit is corrected to the lower limit, and the portion above the upper limit is corrected to the upper limit. Then, the lower limit is subtracted from the corrected original score, and the result is divided by the difference between the upper and lower limits to obtain the ECG state confidence score, noise interference confidence score, and electrode contact state confidence score.

[0046] The confidence level of the electrocardiogram (ECG) status is compared with the ECG abnormality judgment threshold range. If it is below the lower limit, it is determined to be a normal ECG; if it is within the threshold range, it is determined to be a suspected ECG abnormality; if it is above the upper limit, it is still determined to be a suspected ECG abnormality, and a high-confidence abnormality label is generated as the basis for generating subsequent acquisition control parameters. After limiting the confidence level of noise interference to a preset confidence level range, it is compared with the interference intensity threshold. If it does not reach the interference intensity threshold, it is determined to be no interference. If it reaches the interference intensity threshold, it is further combined with the power frequency band energy ratio, baseline drift amplitude, and baseline drift duration to determine power frequency interference, baseline drift, and composite interference, respectively.

[0047] It should be noted that the state discrimination rule set includes ECG state discrimination rules, noise state discrimination rules, and electrode contact state discrimination rules; In the multimodal ECG feature vector, ECG waveform features include amplitude deviation, slope energy anomaly, rhythm fluctuation, and main wave group morphology deviation. Noise interference characteristics include the power frequency band energy ratio, baseline drift amplitude, low-frequency drift duration, and high-frequency noise intensity; Electrode contact characteristics include electrode impedance state, saturation sampling state, horizontal sampling state, and lead validity; The lower and upper limits of the rating are determined based on the rating range, which is 0 to 100. The lower limit of the rating is 0 and the upper limit of the rating is 100.

[0048] ECG status confidence is a numerical value calculated based on ECG waveform characteristics, used to characterize the degree of confidence that the ECG signal in the current acquisition window belongs to a suspected abnormal state. The ECG abnormality judgment threshold refers to the judgment boundary used to distinguish between normal ECG and suspected ECG abnormality. The ECG abnormality judgment threshold range can be set to [0.60, 0.85]. When the ECG state confidence is less than 0.60, it is judged as normal ECG. When the ECG state confidence is greater than or equal to 0.60, it is judged as suspected ECG abnormality. When the ECG state confidence is greater than 0.85, a high-confidence abnormality label is generated.

[0049] Noise interference confidence level refers to the confidence level value generated based on the noise-related characteristics within the current acquisition window, which is used to represent the probability and strength of the ECG signal being affected by noise; The interference intensity threshold is a boundary set for the effective determination of noise. It is used to determine whether the current noise has reached a level that needs to be identified and processed. The interference intensity threshold can be set to 0.50 to 0.70, for example, 0.60.

[0050] The confidence level of electrode contact status is compared with the threshold range of electrode contact status to determine the electrode status as good fit, high impedance, or lead detachment. The determined ECG status, noise status, and electrode status are combined and mapped according to the status coding protocol to generate a ternary signal pattern status.

[0051] Furthermore, the confidence level of the electrode contact status is limited to a preset confidence level range and compared with the electrode contact status threshold range to determine whether the electrode status is good fit, high impedance, or lead detachment. When the confidence level of the electrode contact status is at the boundary of an adjacent range, it is corrected by combining the electrode impedance value, the horizontal line sampling mark, and the saturation sampling mark. If the electrode impedance value is high and not continuous horizontal, it is corrected to high impedance. If the electrode impedance value meets the detachment judgment condition and is accompanied by continuous horizontal lines, it is corrected to lead detachment. If the electrode impedance value is normal and not continuous saturation but continuous horizontal lines, it is corrected to good fit.

[0052] The determined ECG state, noise state, and electrode state are converted into corresponding coded values ​​according to the state coding protocol, and a three-segment state code is formed in the order of ECG state code, noise state code, and electrode state code. The three-segment state code is matched with the preset state relationship table to obtain the comprehensive state name and control attributes corresponding to the current acquisition window. Among them, normal ECG, no interference, and good fit correspond to normal acquisition state, suspected ECG abnormality with noise interference corresponds to abnormality pending confirmation state under noise interference, lead detachment corresponds to lead detachment control state, and when suspected ECG abnormality, compound interference, and high impedance coexist, it corresponds to compound abnormality control state.

[0053] It should be noted that the confidence level of electrode contact status refers to the confidence level value generated based on electrode impedance value, saturation sampling mark, horizontal sampling mark, and lead validity. The electrode contact state threshold range refers to the boundary for judging the electrode state. When the confidence level of the electrode contact state is normalized to 0 to 1, the good fit range can be set to 0 to 0.40, the high impedance range can be set to 0.40 to 0.70, and the lead detachment range can be set to 0.70 to 1.00.

[0054] A state coding protocol refers to the defined state codes, their order, and mapping relationships. The three-element signal mode state refers to the comprehensive acquisition state determined by the ECG state, noise state, and electrode state, which is used to reflect the combined relationship between the ECG waveform, noise interference, and electrode contact in the current acquisition window.

[0055] S3. Generate acquisition control parameters based on the three-element signal mode status. When the ECG status is suspected ECG abnormality and the noise status indicates the presence of noise interference, suppress the output of the definitive ECG abnormality conclusion, output the abnormality pending confirmation mark under noise interference, retain the original ECG data and acquisition window identifier, and generate enhanced acquisition control parameters. When the noise status indicates the presence of power frequency interference and baseline drift, generate filter adjustment parameters. When the electrode status indicates lead detachment, generate control parameters to pause the output of ECG features.

[0056] Based on the combination of confidence levels of ECG status, noise interference, and electrode contact status, a preset control strength rule table is queried to form the acquisition control strength. When any of the following conditions is met: suspected ECG abnormality, enhanced noise interference, or enhanced abnormal electrode contact, the acquisition control strength is increased.

[0057] Furthermore, based on the ECG state, noise state, electrode state, and corresponding confidence levels in the three-element signal mode, the three states are converted into ECG control factors, noise control factors, and electrode control factors, respectively. These factors are then combined with the confidence level to form ECG control contributions, noise control contributions, and electrode control contributions. When there is suspected ECG abnormality, noise state changes from no interference to power frequency interference, baseline drift or compound interference, or electrode state changes from good fit to high impedance or lead detachment, the corresponding control contribution increases with increasing confidence level.

[0058] The three types of control contributions are synthesized according to the preset control strength rule table to obtain the acquisition control strength. When any control contribution reaches a high level, it is upgraded to the enhanced acquisition level. When the electrode status is lead detachment, it is upgraded to the pause output control level. When suspected ECG abnormalities and noise interference coexist, it is upgraded to the abnormality suppression and enhanced acquisition control level. The acquisition control strength is used to subsequently determine the amplitude of sampling rate enhancement, amplification gain correction, filtering adjustment, and ECG feature output control.

[0059] The expression for the acquisition control strength is: ; in, Indicates the first The intensity of acquisition control for each acquisition window; Indicates the first The total length of the state evolution path within each acquisition window; This represents the state evolution path formed by the lead sampling points, ECG state, noise state, and electrode state. Indicates the positional parameters on the state evolution path; Indicates path location Lead validity weighting at the location; Indicates path location The overall state driving quantity at the location; Indicates path location The data collection reliability margin at the location; Indicates stability margin; Indicates path location The noise state at the point is coupled with the electrode state, resulting in an enhanced term. This represents the coupling enhancement weight.

[0060] It should be noted that the control strength rule table can be set based on historical ECG acquisition samples, noise interference samples, abnormal electrode contact samples, and equipment calibration results. It is used to establish the correspondence between ECG status, noise status, electrode status, and corresponding confidence level and acquisition control strength.

[0061] Configure power frequency notch adjustment parameters when the noise state includes power frequency interference, and configure baseline correction parameters and high-pass filter adjustment parameters when the noise state includes baseline drift; perform output gating judgment on ECG state, noise state and electrode state.

[0062] Furthermore, by combining the noise status, power frequency band energy ratio, power frequency interference confidence level, baseline drift amplitude, and baseline drift duration of the current acquisition window, the filtering adjustment method is determined; when the noise status includes power frequency interference, power frequency notch adjustment parameters including notch center frequency, notch bandwidth, and notch suppression depth are generated; when the noise status includes baseline drift, baseline correction parameters and high-pass filter adjustment parameters are generated; when the noise status is composite interference, the above-mentioned filter adjustment parameters are generated simultaneously.

[0063] Output gating conditions are formed based on the ECG status, noise status, and electrode status of the current acquisition window. When the ECG is normal, interference-free, and well-fitted, an output permission flag is generated. When an ECG abnormality is suspected and accompanied by one of the following: power frequency interference, baseline drift, or composite interference, an abnormal output suppression flag is generated, and the process is switched to enhanced acquisition and filter adjustment. When the electrode status is high impedance, an output risk flag is generated. When the electrode status is lead detachment, a pause output flag is generated, and the ECG characteristic output path of the corresponding lead is blocked.

[0064] When the ECG status is suspected to be abnormal and the noise status indicates the presence of noise interference, the ECG abnormality result is set to the suppressed output state. When the electrode status is lead detachment, the ECG feature is set to the paused output state. The acquisition control parameters are generated by combining the acquisition control strength, power frequency notch adjustment parameters, baseline correction parameters, and high-pass filter adjustment parameters.

[0065] Furthermore, when the ECG status is suspected to be abnormal, it is determined whether the noise status is power frequency interference, baseline drift, or a combination of interference. If noise interference exists, the current ECG abnormality result is marked as an abnormality pending confirmation under noise interference, and an abnormal output suppression flag is configured to suppress the output of definitive ECG abnormality conclusions. At the same time, the original ECG data, ECG status confidence, noise interference confidence, and acquisition window identifier are retained as the basis for enhanced acquisition and filtering adjustment in the next acquisition window. If the noise status is without interference, the ECG abnormality result output channel remains open.

[0066] When the electrode status is lead detachment, an invalid lead identifier is generated based on the lead number, and the ECG feature output path of the corresponding lead is set to paused output. When the electrode status is high impedance, a contact abnormality warning mark is generated, and the ECG feature output result includes electrode contact risk information. When the electrode status is good fit, the ECG feature output path remains open. The acquisition control strength, abnormal output suppression mark, paused output mark, invalid lead identifier, power frequency notch adjustment parameter, baseline correction parameter, and high-pass filter adjustment parameter are encapsulated according to parameter type to form acquisition control parameters for generating ECG acquisition execution control parameters in the next acquisition window.

[0067] When the ECG status is suspected to be abnormal and the noise status is one of power frequency interference, baseline drift, and combined interference, the confidence level of the ECG status, the confidence level of the noise interference, and the acquisition control strength are obtained. The confidence level of the ECG status, the confidence level of the noise interference, and the acquisition control strength are divided into intervals to obtain the confidence level of the ECG abnormality, the confidence level of the noise interference, and the acquisition control strength level. The confidence level of the ECG abnormality, the confidence level of the noise interference, and the acquisition control strength level are combined to form an abnormal output gating index.

[0068] Furthermore, based on the three-dimensional signal mode status of the current acquisition window, the ECG status and noise status are jointly judged. When the ECG status is suspected to be abnormal and the noise status is one of power frequency interference, baseline drift, or combined interference, the current acquisition window is marked as a noise-accompanied abnormality window. For the noise-accompanied abnormality window, the ECG status confidence score, noise interference confidence score, and acquisition control strength are extracted. The ECG status confidence score is used to characterize the credibility of the suspected ECG abnormality, the noise interference confidence score is used to characterize the degree of influence of noise on the ECG abnormality judgment, and the acquisition control strength is used to determine the enhancement acquisition and filtering adjustment amplitude of the next acquisition window.

[0069] The confidence levels of ECG status, noise interference, and acquisition control strength are limited to preset ranges and divided into ECG abnormality confidence levels, noise interference confidence levels, and acquisition control strength levels according to the corresponding intervals. Among them, the confidence levels of ECG status and noise interference are divided into low, medium, and high levels, respectively, and the acquisition control strength is divided into normal control level, enhanced control level, and strong enhanced control level. The three level fields are arranged in order to form an abnormal output gating index, which is used to query the abnormal output gating rule table and determine the abnormal output suppression level.

[0070] It should be noted that the control intensity range refers to a preset range used to divide the acquisition control intensity into different control levels, which is used to determine the adjustment range of the next acquisition window.

[0071] The abnormal output suppression level is determined by querying the abnormal output gating rule table based on the abnormal output gating index, and an abnormal output suppression flag is generated based on the abnormal output suppression level. The sampling rate adjustment index is composed of the noise status, noise interference confidence level and acquisition control strength level. The sampling rate adjustment rule table is queried based on the sampling rate adjustment index to obtain the sampling rate enhancement level and generate the sampling rate enhancement parameters for the next acquisition window.

[0072] Furthermore, an abnormal output gating index is composed of the ECG abnormality confidence level, noise interference confidence level, and acquisition control strength level, and is matched with the rule items in the abnormal output gating rule table to obtain the abnormal output suppression level. An abnormal output suppression flag is generated based on the abnormal output suppression level. The abnormal output suppression flag includes a suppression status bit, a suppression level field, a trigger reason field, and a recovery condition field, which are used to control the output of deterministic ECG abnormality conclusions. In the next acquisition window, the recovery condition field is used to determine whether to release the suppression.

[0073] The sampling rate adjustment index is composed of noise status, noise interference confidence level, and acquisition control strength level, and is matched with the rule items in the preset sampling rate adjustment rule table to obtain the sampling rate boost level. Combining the current sampling rate, sampling rate boost level, and maximum sampling rate limit, sampling rate boost parameters are generated. The sampling rate boost parameters include target sampling rate, boost level, effective acquisition window, and sampling rate limit flag, which are used to control the sampling clock of the next acquisition window.

[0074] It should be noted that the abnormal output gating rule table stores the correspondence between multiple index keys and output processing methods. Each rule item includes an index key, abnormal output suppression level, suppression status bit, trigger reason field, and recovery condition field.

[0075] The sampling rate adjustment rule table includes noise type conditions, interference confidence level conditions, control strength level conditions, sampling rate increase level, target sampling rate generation method, maximum sampling rate limit, and effective window identifier.

[0076] The direction of amplification gain correction is determined based on the lead sampling value, saturation sampling mark, and horizontal sampling mark, and amplification gain correction parameters are generated based on the direction of amplification gain correction and the acquisition control intensity level.

[0077] Furthermore, based on the amplitude distribution, saturation sampling marker, and flat sampling marker of the sampled values ​​of each lead within the current acquisition window, the direction of amplification gain correction is determined; when the lead sampled values ​​are continuously close to the upper or lower limit of quantization and saturation sampling markers appear consecutively, it is determined to reduce the gain; when the lead sampled values ​​are in the low amplitude range for a long time and no effective waveform change or accompanying flat sampling marker appears, it is determined to increase the gain; when the lead sampled values ​​are in the effective amplitude range and no saturation sampling markers or flat sampling markers appear consecutively, it is determined to maintain the gain.

[0078] Amplification gain correction parameters are generated by combining the amplification gain correction direction and the acquisition control strength level. The higher the acquisition control strength level, the greater the gain adjustment range, and it is subject to the maximum allowable gain and the minimum allowable gain limit. The amplification gain correction parameters include the gain correction direction, gain adjustment level, target amplification gain, effective lead number, effective acquisition window, and gain limit marker, which are used to control the adjustment method of the amplification gain structure in the next acquisition window.

[0079] It should be noted that the effective amplitude range is 10% to 90% of the full quantization range. When the lead sampling value falls within the effective amplitude range, it means that the current amplification gain meets the requirements for ECG waveform acquisition. Amplification gain correction parameters refer to a set of parameters used to control the amplification gain structure adjustment method in the next acquisition window, including gain correction direction, gain adjustment level, target amplification gain, effective lead number, effective acquisition window, and gain limit flag.

[0080] The filter compensation type is determined based on the noise status, and filter compensation parameters are generated based on the filter compensation type, noise interference confidence level, and acquisition control strength level. The abnormal output suppression flag, sampling rate boosting parameter, amplification gain correction parameter, and filter compensation parameter are combined into enhanced acquisition control parameters.

[0081] Furthermore, the filter compensation type is determined based on the noise state. When the noise state is power frequency interference, power frequency notch compensation parameters are generated, including the notch center frequency, notch bandwidth, and notch suppression depth. When the noise state is baseline drift, baseline drift compensation parameters are generated, including the baseline correction strength, high-pass cutoff frequency, and baseline correction step size. When the noise state is composite interference, both power frequency notch compensation parameters and baseline drift compensation parameters are generated. When the noise state is no interference, no filter compensation parameters are generated or the current filter state is maintained.

[0082] The higher the confidence level of noise interference, the higher the filtering compensation strength, the higher the acquisition control strength level, and the greater the adjustment range of filtering parameters. The filtering strength is limited by the allowable range of notch bandwidth, high-pass cutoff frequency, baseline correction step size, and suppression depth. After the filtering compensation parameters are generated, the abnormal output suppression flag, sampling rate boosting parameters, amplification gain correction parameters, and filtering compensation parameters are encapsulated according to parameter type to form enhanced acquisition control parameters.

[0083] It should be noted that the power frequency notch compensation parameters include the notch center frequency, notch bandwidth, and notch suppression depth. The higher the noise interference confidence level, the greater the notch suppression depth; the higher the acquisition control strength level, the greater the notch bandwidth adjustment range. The enhanced acquisition control parameters include an abnormal output control field, a sampling rate adjustment field, a gain correction field, a filter compensation field, and an effective acquisition window field, which are used to control the abnormal result output, sampling clock, amplification gain, and filtering processing mode in the next acquisition window.

[0084] The power frequency interference state and the baseline drift state are used as periodic noise evidence anchor points and low-frequency drift evidence anchor points, respectively, and connected to the noise interference feature node and the filter parameter control node in the signal mode relationship diagram to form a noise filtering relationship sub-graph.

[0085] Furthermore, the anchor points for periodic noise evidence include power frequency interference identifiers, power frequency band energy percentages, noise interference confidence levels, affected lead numbers, and acquisition window identifiers; the anchor points for low-frequency drift evidence include baseline drift identifiers, baseline drift amplitudes, baseline drift durations, noise interference confidence levels, affected lead numbers, and acquisition window identifiers.

[0086] In the signal pattern diagram, the periodic noise evidence anchor point is connected to the power frequency band energy node, noise interference characteristic node, and notch filter parameter control node. The low-frequency drift evidence anchor point is connected to the low-frequency trend component node, baseline drift characteristic node, baseline correction parameter node, and high-pass filter parameter node. The corresponding evidence anchor point and associated nodes are retained according to the noise state. When the noise state is composite interference, both types of evidence anchor points and associated nodes are included in the noise filtering relationship subgraph to determine the power frequency interference control weight, baseline drift control weight, and filter adjustment parameters.

[0087] It should be noted that each node in the signal pattern diagram is configured with a node number, node type, source acquisition window, effective lead, evidence parameter field, state encoding field, and control parameter field. The evidence parameter field stores the feature values ​​used by the node for judgment, the state encoding field records the encoding results of ECG state, noise state, and electrode state, and the control parameter field records the parameter adjustment objects triggered by the node. Connection edges are configured with a starting node number, target node number, association weight, weight source field, and effective acquisition window. The association weight is generated from the corresponding evidence parameter, corresponding confidence level, and effective lead range, and is used to characterize the strength of the influence of the evidence node on the control node. The signal pattern diagram consists of feature nodes, state nodes, and control nodes. Feature nodes include ECG waveform feature nodes, noise interference feature nodes, electrode contact feature nodes, power frequency band energy nodes, baseline drift feature nodes, low-frequency trend component nodes, and lead validity nodes. The control nodes include a sampling rate control node, an amplification gain control node, a notch filter parameter control node, a baseline correction parameter node, a high-pass filter parameter node, and an ECG feature output control node.

[0088] Noise interference characterization parameters are extracted from the noise filtering relationship subgraph, and power frequency interference control weights and baseline drift control weights are generated. Based on the power frequency interference control weights, notch center frequency, notch bandwidth and notch suppression depth are generated. Based on the baseline drift control weights, high-pass cutoff frequency and baseline correction step size are generated and combined into filter adjustment parameters.

[0089] Furthermore, the noise filtering relationship subgraph includes periodic noise evidence anchor points, low-frequency drift evidence anchor points, noise interference feature nodes, and filter parameter control nodes. The periodic noise evidence anchor points provide the power frequency band energy proportion, the number of power frequency interference duration windows, and the number of affected leads. The low-frequency drift evidence anchor points provide the baseline drift amplitude, baseline drift duration, and low-frequency trend change amplitude. These parameters, combined with the noise interference confidence level, form the noise interference characterization parameters. Higher power frequency band energy proportions, more power frequency interference duration windows, and more affected leads result in higher power frequency interference control weights. Similarly, larger baseline drift amplitudes, longer baseline drift durations, and more pronounced low-frequency trend changes result in higher baseline drift control weights.

[0090] The notch center frequency, notch bandwidth, and notch suppression depth are generated based on the power frequency interference control weight, and the high-pass cutoff frequency and baseline correction step size are generated based on the baseline drift control weight. When the power frequency interference control weight is high, the notch suppression depth is increased and the notch bandwidth is appropriately widened; when the baseline drift control weight is high, the high-pass cutoff frequency and baseline correction step size are increased. The notch center frequency, notch bandwidth, notch suppression depth, high-pass cutoff frequency, baseline correction step size, effective lead number, effective acquisition window, and parameter limit markers are encapsulated to form filter adjustment parameters, which are used for power frequency notch, baseline correction, and high-pass filtering processing in the next acquisition window.

[0091] The calculation expression for the power frequency interference control weight is as follows: ; in, Indicates the power frequency interference control weight; This represents a normalized mapping, obtained by applying upper and lower bounds to the values ​​within the parentheses. The correlation weight between power frequency noise evidence and notch filter control is obtained through equipment calibration. This indicates the proportion of energy in the power frequency band, which is obtained by comparing the power frequency band energy with the total frequency band energy. The duration of power frequency interference is indicated by the number of consecutive hits in the power frequency interference detection sampling window; The range of leads affected by power frequency interference is indicated by the number of leads whose power frequency band energy exceeds the judgment threshold. The adjustment coefficient, which indicates the duration of power frequency interference, is obtained through equipment calibration; The adjustment coefficient, representing the range of affected leads, is determined by the number of leads and the lead layout. This represents the reliability margin of the power frequency filter, which is obtained by the margin between the allowable filtering range of the equipment and the current required notch adjustment amount. This represents the stability margin of the power frequency filter, which is obtained by continuously sampling the stability of the power frequency center frequency offset within the window. This represents the confidence enhancement coefficient for power frequency noise, obtained through noise interference confidence calibration. The confidence level of noise interference is represented by the noise interference features determined by the state discrimination rule set. The formula for calculating the baseline drift control weights is: ; in, Indicates the baseline drift control weights; The correlation weight between low-frequency drift evidence and baseline correction control is obtained through equipment calibration. This indicates the baseline drift magnitude, obtained by the offset of the baseline estimate relative to the reference baseline; The duration of baseline drift is indicated by the number of consecutive acquisition windows that hit the baseline drift detection criteria. This indicates the magnitude of low-frequency trend changes, obtained by measuring the magnitude of changes in the low-frequency trend curve within the statistical window. The adjustment factor, representing the duration of baseline drift, is obtained through equipment calibration; The adjustment coefficient, representing the amplitude of low-frequency trend changes, is calibrated using the sampling frequency and low-frequency drift identification results; This represents the baseline correction confidence margin, obtained by the margin between the allowed baseline correction step size and the currently required correction step size; This represents the baseline correction stability margin, which is obtained by measuring the stability of baseline drift changes within a continuously acquired window. This represents the baseline drift confidence enhancement coefficient, which is obtained through noise interference confidence calibration.

[0092] It should be noted that the noise interference characterization parameters are used to describe the noise type, noise intensity, duration and range of influence within the current acquisition window, including power frequency interference characterization parameters and baseline drift characterization parameters.

[0093] The center frequency of the notch filter is determined by the frequency band where the power frequency interference is located, and the bandwidth and notch suppression depth are jointly determined by the power frequency interference control weight and the allowable filtering range of the equipment. The weights for power frequency interference control and baseline drift control are both between 0 and 1. A weight less than 0.3 is classified as low weight, a weight greater than or equal to 0.3 and less than 0.7 is classified as medium weight, and a weight greater than or equal to 0.7 is classified as high weight.

[0094] The lead detachment state is used as the anchor point for detachment evidence and connected to the electrode contact feature node, lead validity node, and ECG feature output control node in the signal pattern diagram to form a lead detachment control sub-graph.

[0095] Furthermore, the lead detachment evidence anchor points corresponding to the lead detachment state include lead detachment identifier, detached lead number, electrode impedance status, continuous horizontal sampling marker, continuous detachment window number, and acquisition window identifier, used to characterize the channel where lead detachment occurred and the judgment basis; in the signal pattern relationship diagram, the detachment evidence anchor points are connected to the electrode contact feature node, lead validity node, and ECG feature output control node to represent the relationship between lead detachment and electrode impedance abnormality, horizontal sampling abnormality, or no valid input in the lead, and to control whether the ECG waveform features, rhythm features, and abnormal judgment results of the corresponding lead enter the output path; when multiple leads detach simultaneously, detachment evidence anchor points are generated separately and together form a lead detachment control subgraph, used for subsequent identification of invalid leads, generation of pause output markers, and configuration of recovery detection conditions.

[0096] It should be noted that configuring recovery detection conditions refers to the detection rules set to determine whether the lead can resume output after the lead has been determined to be detached and the ECG characteristic output has been suspended.

[0097] Extract lead detachment judgment parameters from the lead detachment control subgraph within the current acquisition window, generate lead detachment control weights, and mark leads whose lead detachment control weights reach the detachment pause threshold as invalid leads.

[0098] Furthermore, the detached lead number is determined by the detachment evidence anchor point in the lead detachment control subgraph, the electrode impedance state and horizontal sampling state are determined by the electrode contact feature node, and the valid waveform change state is determined by the lead validity node. The number of consecutive detachment windows is obtained by combining the detachment records of adjacent acquisition windows. The detached lead number, electrode impedance state, consecutive horizontal sampling mark, valid waveform change state, number of consecutive detachment windows, and electrode contact state confidence are used as lead detachment judgment parameters, and lead detachment control weights are generated according to preset weight rules.

[0099] When the electrode impedance reaches the detachment judgment interval, continuous horizontal sampling markers appear, or the number of consecutive detachment windows increases, the lead detachment control weight is increased; when valid waveform changes still exist, the lead detachment control weight is decreased. When the lead detachment control weight reaches the detachment pause threshold, the corresponding lead is marked as an invalid lead; when the detachment pause threshold is not reached, it remains in a pending confirmation state or a valid state, and the judgment continues in conjunction with subsequent acquisition windows. The invalid lead mark is used to subsequently block the ECG characteristic output path of the corresponding lead.

[0100] It should be noted that the parameters for determining lead detachment include the detached lead number, electrode impedance status, continuous horizontal line sampling marker, effective waveform change status, number of consecutive detachment windows, and confidence level of electrode contact status. The detachment pause threshold serves as the boundary for determining whether the lead should pause output, and can be set based on electrode type, device calibration data, and historical detachment samples. The weighting rules are set based on electrode type, equipment calibration data, historical lead detachment samples, normal fit samples, and short-term contact fluctuation samples.

[0101] The output path of the ECG waveform characteristics corresponding to the invalid lead is blocked, the output state of the ECG characteristics is set to paused, and the invalid lead identifier, paused output marker and recovery detection condition are combined into control parameters for pausing the output of the ECG characteristics.

[0102] Furthermore, the output path corresponding to the invalid lead is found in the ECG feature output control node, and the open state is switched to the blocked state, so that the corresponding ECG waveform features, rhythm features and abnormal judgment results are not output to the external results; at the same time, the ECG feature output status is updated to paused output and a paused output mark is generated. The paused output mark includes the paused status bit, invalid lead number, pause trigger reason and effective acquisition window.

[0103] The recovery detection conditions and pause output markers are saved together, including electrode impedance returning to the normal contact range, disappearance of the continuous horizontal sampling marker, reappearance of valid waveform changes in lead sampling values, and no reappearance of lead detachment judgment results within the continuous acquisition window; the invalid lead identifier, pause output marker, and recovery detection conditions are encapsulated as control parameters for pausing ECG characteristic output, and the pause output state of the corresponding lead is determined based on the recovery detection conditions in the next acquisition window.

[0104] It should be noted that the recovery detection conditions are a set of conditions used to determine whether a lead can resume its characteristic ECG output after it has been identified as detached and its output has been suspended.

[0105] S4. Based on the enhanced acquisition control parameters, filter adjustment parameters, and control parameters for pausing ECG feature output, generate ECG acquisition execution control parameters for the next acquisition window, and perform ECG acquisition according to the ECG acquisition execution control parameters in the next acquisition window to generate digital ECG acquisition information.

[0106] ECG acquisition and control parameters are extracted from the acquisition control parameters, and then the control parameters are fused in the order of pause output marker, filter adjustment parameters, and enhanced acquisition control parameters in the ECG acquisition and control parameters.

[0107] Furthermore, the ECG acquisition control parameters are formed by fields directly related to the next acquisition window, including ECG characteristic output status, invalid lead number, notch center frequency, notch bandwidth, notch suppression depth, high-pass cutoff frequency, baseline correction step size, target sampling rate, amplification gain correction direction, and target amplification gain. When fusing control parameters, the pause output flag has the highest priority and is used to set the ECG characteristic output status of the corresponding lead to pause output while retaining the invalid lead identifier. The filter adjustment parameters are used to determine the notch parameters, baseline correction parameters, and high-pass filter parameters, and are given priority when they point to the same filtering object as the filter compensation field. The enhancement acquisition control parameters are used to supplement the sampling rate enhancement parameters, amplification gain correction parameters, and abnormal output suppression flags. If the enhancement acquisition control parameters conflict with the pause output flag, the pause output result is retained, and the non-conflicting parameters are written into the fusion result.

[0108] It should be noted that the acquisition control parameters include pause output flag, invalid lead flag, filter adjustment parameters, enhanced acquisition control parameters, and effective acquisition window field; The fused control parameters include the ECG feature output status of the next acquisition window, the target sampling rate, the target amplification gain, the filtering parameters, the abnormal output suppression status, and the effective acquisition window identifier.

[0109] For control parameters that have not been generated, they will not participate in control parameter fusion, and the corresponding sampling rate, amplification gain, filtering parameters and ECG feature output state will remain unchanged, and the ECG acquisition execution control parameters for the next acquisition window will be generated.

[0110] Furthermore, ungenerated control parameters refer to parameter items that do not exist because the corresponding control conditions are not triggered in the current acquisition window. For example, a pause output marker is not generated when no lead detachment occurs, filter adjustment parameters are not generated when there is no power frequency interference or baseline drift, and enhanced acquisition control parameters are not generated when there is noise accompanied by suspected ECG abnormalities.

[0111] For the control parameters that have already been generated, they are written into the execution configuration according to the priority of pause output marker, filter adjustment parameters, and enhanced acquisition control parameters. For parameters that have not been generated, they are marked as untriggered, do not participate in fusion, do not overwrite the current valid configuration, and keep the original sampling rate, original amplification gain, original filter parameters, and original ECG feature output status unchanged. After fusion is completed, the ECG acquisition execution control parameters for the next acquisition window are generated. The ECG acquisition execution control parameters include target sampling rate, target amplification gain, filter parameters, ECG feature output status, abnormal output suppression status, and effective acquisition window identifier.

[0112] Write the ECG acquisition execution control parameters into the acquisition configuration item of the next acquisition window, and control the ECG feature output according to the pause output flag.

[0113] Furthermore, the acquisition configuration items include sampling clock configuration items, amplification gain configuration items, filtering configuration items, ECG feature output configuration items, and effective window identifiers. The target sampling rate is written to the sampling clock configuration item, the target amplification gain is written to the amplification gain configuration item, the notch filter parameter, high-pass filter parameter, and baseline correction parameter are written to the filtering configuration item, and the abnormal output suppression status, ECG feature output status, and invalid lead number are written to the ECG feature output configuration item. When the pause output flag is active, the ECG feature output channel of the corresponding lead is closed according to the invalid lead number, so that the ECG waveform features, rhythm features, and abnormal judgment results of the invalid lead are not included in the output results. Leads not pointed to by the pause output flag maintain their original output status, forming the effective acquisition configuration for the next acquisition window.

[0114] Based on the acquisition configuration items of the next acquisition window, the driving sampling clock, amplification gain structure and filtering processing structure operate according to the updated sampling rate, amplification gain and filtering parameters, and the ECG feature output channel is opened, suppressed and paused according to the ECG feature output status.

[0115] Furthermore, the target sampling rate in the acquisition configuration is loaded into the sampling clock control structure, generating a sampling trigger signal according to the target sampling rate; the target amplification gain is loaded into the amplification gain structure, enabling it to process the ECG analog signal according to the target amplification factor; the notch center frequency, notch bandwidth, notch suppression depth, high-pass cutoff frequency, and baseline correction step size are loaded into the filtering processing structure, enabling it to perform power frequency suppression, low-frequency drift reduction, and baseline correction. The ECG feature output status is used to control the output channels of each lead: when output is enabled, ECG waveform features, rhythm features, and abnormal judgment results enter the output channel; when output is suppressed, ECG abnormal results are temporarily not output, but sampling data and status information are retained; when output is paused, the ECG feature output channel of the corresponding lead is closed according to the invalid lead number.

[0116] Acquire lead data within the adjusted next acquisition window, and record the actual sampling rate, actual amplification gain, filtering parameters, lead status, and ECG characteristic output status to form an ECG data frame with control identifiers. Encapsulate the ECG data frame with control identifiers, the three-element signal mode status, acquisition control parameters, and acquisition timestamps to generate digital ECG acquisition information.

[0117] Furthermore, the sampling clock triggers sampling for each lead according to the target sampling rate, the amplification gain structure processes the ECG analog signal according to the target amplification gain, and the filtering processing structure processes the sampled signal according to the updated notch filter parameters, high-pass filter parameters, and baseline correction parameters to obtain the lead data in the next acquisition window. The lead data, actual sampling rate, actual amplification gain, actual effective filter parameters, lead status, ECG characteristic output status, abnormal output suppression flag, paused output flag, and effective acquisition window identifier are encapsulated into an ECG data frame with control identifiers.

[0118] After an ECG data frame with a control identifier is formed, the current acquisition window's time position in the continuous acquisition process is determined according to the acquisition timestamp, and the acquisition timestamp is bound to the frame sequence number, lead number, and effective acquisition window identifier. The ECG data frame with the control identifier is used as the data body, the three-element signal mode status is used as the status description field, the acquisition control parameters are used as the control description field, and the acquisition timestamp is used as the time tracking field to form digital ECG acquisition information.

[0119] It should be noted that the digital ECG acquisition information includes lead sampling data, control identifiers, three-dimensional signal mode status, acquisition control parameters, acquisition timestamps, and frame integrity verification information, which are used for subsequent ECG data storage, abnormal result verification, acquisition process tracking, and closed-loop control analysis. The three-element signal mode status is used to identify the ECG status, noise status, and electrode status corresponding to the current acquisition window. The acquisition control parameters are used to identify the sampling rate adjustment, amplification gain correction, filtering adjustment, and output control mode used in the current acquisition window or the next acquisition window.

[0120] In summary, this invention achieves collaborative discrimination of ECG state, noise state, and electrode state by performing time alignment, lead encoding, amplitude normalization, and electrode state encoding on the original digital ECG data frames, and by performing signal pattern recognition based on ECG acquisition state characteristics. This improves the accuracy of acquisition state recognition in complex acquisition environments. Furthermore, it generates and fuses enhanced acquisition control parameters, filtering adjustment parameters, and pause output control parameters based on the three signal pattern states to form the ECG acquisition execution control parameters for the next acquisition window. This enables dynamic adjustment of acquisition parameters and characteristic output states, thereby reducing the impact of noise interference and lead dropout on ECG results and improving the reliability of digital ECG acquisition information.

[0121] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A digital electrocardiogram (ECG) acquisition method based on signal pattern recognition, characterized in that, include: The system acquires human electrocardiogram (ECG) analog signals, amplifies, filters, and performs analog-to-digital conversion on these signals to generate raw digital ECG data frames. The original digital ECG data frames are time-aligned, lead-coded, amplitude-normalized, and electrode-state-coded to obtain the ECG data to be identified and extract ECG acquisition state features. Signal pattern recognition is performed on the ECG acquisition state features to obtain a three-element signal pattern state composed of ECG state, noise state, and electrode state. Based on the three-element signal mode status, acquisition control parameters are generated. When the ECG status is suspected ECG abnormality and the noise status indicates the presence of noise interference, the output of the definitive ECG abnormality conclusion is suppressed. At the same time, an abnormality pending confirmation mark under noise interference is output, while the original ECG data and acquisition window identifier are retained, and enhanced acquisition control parameters are generated. When the noise status indicates the presence of power frequency interference and baseline drift, filter adjustment parameters are generated. When the electrode status indicates lead detachment, control parameters for pausing ECG feature output are generated. Based on the enhanced acquisition control parameters, filter adjustment parameters, and control parameters for pausing ECG feature output, the ECG acquisition execution control parameters for the next acquisition window are generated. ECG acquisition is then performed in the next acquisition window according to the ECG acquisition execution control parameters to generate digital ECG acquisition information.

2. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 1, characterized in that, The process of acquiring simulated human electrocardiogram (ECG) signals, amplifying, filtering, and performing analog-to-digital conversion on these signals to generate raw digital ECG data frames involves the following steps: Impedance matching and differential amplification are performed on the ECG analog signal to obtain a pre-amplified signal. The pre-amplified signal is then combined with the quality index for adaptive filtering and baseline correction to obtain a clean analog sequence. Quantization and frame structure encapsulation are performed on the pure analog sequence to obtain the original digital ECG data frame.

3. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 1 or 2, characterized in that, The process of performing time alignment, lead encoding, amplitude normalization, and electrode state encoding on the original digital electrocardiogram (ECG) data frames to obtain the ECG data to be identified and extracting ECG acquisition state features is as follows: Extract ECG acquisition frame parsing information from the original digital ECG data frames, and construct a sample matrix to be aligned based on the frame number and sampling point number in the ECG acquisition frame parsing information; A unified time reference is determined by calculating the slope energy centroid position of the lead sample values ​​in the ECG acquisition frame parsing information; based on the unified time reference, the sample matrix to be aligned is time-aligned, and the aligned sample values ​​are subjected to lead encoding and amplitude normalization to obtain the lead data sequence; Electrode status codes are generated based on the electrode impedance values, saturation sampling markers, and horizontal sampling markers in the ECG acquisition frame parsing information; the lead data sequence and electrode status codes are combined to form ECG data to be identified, and ECG acquisition status features are extracted from the ECG data to be identified.

4. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 3, characterized in that, The specific steps for performing signal pattern recognition on the ECG acquisition status characteristics to obtain a three-element signal pattern state composed of ECG status, noise status, and electrode status are as follows: The ECG acquisition status features are aligned and combined and encoded according to the feature arrangement rules to construct a multimodal ECG feature vector; Based on multimodal ECG feature vectors, the confidence scores of ECG state, noise interference, and electrode contact state are calculated using a set of state discrimination rules. The ECG state confidence scores are then compared with the ECG abnormality judgment threshold range to determine whether the ECG state is normal or suspected abnormal. The noise interference confidence level is compared with the interference intensity threshold to classify the noise state into no interference, power frequency interference, baseline drift and combined interference. The confidence level of electrode contact status is compared with the threshold range of electrode contact status to determine the electrode status as good fit, high impedance, or lead detachment. The determined ECG status, noise status, and electrode status are combined and mapped according to the status coding protocol to generate a ternary signal pattern status.

5. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 4, characterized in that, The specific steps for generating acquisition control parameters based on the three-element signal mode state are as follows: Based on the combination of confidence levels of ECG status, noise interference, and electrode contact status, a preset control strength rule table is queried to form the acquisition control strength. When any of the following conditions is met: suspected ECG abnormality, enhanced noise interference, or enhanced abnormal electrode contact, the acquisition control strength is increased. Configure power frequency notch adjustment parameters when the noise condition includes power frequency interference, and configure baseline correction parameters and high-pass filter adjustment parameters when the noise condition includes baseline drift; perform output gating judgment on ECG status, noise status and electrode status; When the ECG status is suspected to be abnormal and the noise status indicates the presence of noise interference, the ECG abnormality result is set to the suppressed output state. When the electrode status is lead detachment, the ECG feature is set to the paused output state. The acquisition control parameters are generated by combining the acquisition control strength, power frequency notch adjustment parameters, baseline correction parameters, and high-pass filter adjustment parameters.

6. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 5, characterized in that, When the ECG status is suspected to be abnormal and the noise status indicates the presence of noise interference, the output of the definitive ECG abnormality conclusion is suppressed, while an abnormality pending confirmation flag under noise interference is output. The original ECG data and acquisition window identifier are retained, and enhanced acquisition control parameters are generated. The specific steps are as follows: When the ECG status is suspected to be abnormal and the noise status is one of power frequency interference, baseline drift, and combined interference, obtain the confidence level of the ECG status, the confidence level of the noise interference, and the acquisition control strength. The confidence levels of ECG status, noise interference, and acquisition control strength are divided into intervals to obtain the ECG abnormality confidence level, noise interference confidence level, and acquisition control strength level. The ECG abnormality confidence level, noise interference confidence level, and acquisition control strength level are then combined to form an abnormal output gating index. The abnormal output suppression level is determined by querying the abnormal output gating rule table based on the abnormal output gating index, and an abnormal output suppression flag is generated based on the abnormal output suppression level. The sampling rate adjustment index is composed of noise status, noise interference confidence level, and acquisition control strength level. The sampling rate adjustment rule table is queried based on the sampling rate adjustment index to obtain the sampling rate boost level and generate the sampling rate boost parameters for the next acquisition window. The amplification gain correction direction is determined based on the lead sampling value, saturation sampling flag, and horizontal sampling flag, and the amplification gain correction parameters are generated based on the amplification gain correction direction and the acquisition control strength level. The filter compensation type is determined based on the noise status, and filter compensation parameters are generated based on the filter compensation type, noise interference confidence level, and acquisition control strength level. The abnormal output suppression flag, sampling rate boosting parameter, amplification gain correction parameter, and filter compensation parameter are combined into enhanced acquisition control parameters.

7. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 6, characterized in that, When the noise status indicates the presence of power frequency interference and baseline drift, filter adjustment parameters are generated. The specific steps are as follows: The power frequency interference state and the baseline drift state are used as periodic noise evidence anchor points and low frequency drift evidence anchor points, respectively, and connected to the noise interference feature node and the filter parameter control node in the signal mode relationship diagram to form a noise filtering relationship sub-graph. Noise interference characterization parameters are extracted from the noise filtering relation subgraph, and power frequency interference control weights and baseline drift control weights are generated. The notch center frequency, notch bandwidth, and notch suppression depth are generated based on the power frequency interference control weights. The high-pass cutoff frequency and baseline correction step size are generated based on the baseline drift control weights and combined into filter adjustment parameters.

8. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 6, characterized in that, When the electrode status indicates lead detachment, control parameters are generated to pause the ECG characteristic output. The specific steps are as follows: The lead detachment state is used as the anchor point for detachment evidence and connected to the electrode contact feature node, lead validity node and ECG feature output control node in the signal pattern relationship diagram to form a lead detachment control sub-diagram. Extract lead detachment judgment parameters from the lead detachment control subgraph within the current acquisition window, generate lead detachment control weights, and mark leads whose lead detachment control weights reach the detachment pause threshold as invalid leads; The ECG waveform characteristic output path corresponding to the invalid lead is blocked, the ECG characteristic output state is set to pause output, and the invalid lead identifier, pause output flag and recovery detection condition are combined into control parameters for pausing ECG characteristic output.

9. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 6 or 8, characterized in that, Based on the enhanced acquisition control parameters, filter adjustment parameters, and control parameters for pausing ECG feature output, the ECG acquisition execution control parameters for the next acquisition window are generated. The specific steps are as follows: ECG acquisition and control parameters are extracted from the acquisition control parameters, and control parameters are fused in the order of pause output marker, filter adjustment parameters, and enhanced acquisition control parameters in the ECG acquisition and control parameters. For control parameters that have not been generated, they will not participate in control parameter fusion, and the corresponding sampling rate, amplification gain, filtering parameters and ECG feature output state will remain unchanged, and the ECG acquisition execution control parameters for the next acquisition window will be generated. Write the ECG acquisition execution control parameters into the acquisition configuration item of the next acquisition window, and control the ECG feature output according to the pause output flag.

10. The digital electrocardiogram acquisition method based on signal pattern recognition as described in claim 9, characterized in that, The step of performing ECG acquisition according to the ECG acquisition execution control parameters in the next acquisition window to generate digital ECG acquisition information is as follows: Based on the acquisition configuration items of the next acquisition window, the driving sampling clock, amplification gain structure and filtering processing structure operate according to the updated sampling rate, amplification gain and filtering parameters, and open, suppress and pause the ECG feature output channel according to the ECG feature output status. Acquire lead data within the adjusted next acquisition window, and record the actual sampling rate, actual amplification gain, filtering parameters, lead status, and ECG characteristic output status to form an ECG data frame with control identifiers. Encapsulate the ECG data frame with control identifiers, the three-element signal mode status, acquisition control parameters, and acquisition timestamps to generate digital ECG acquisition information.

Images