Remote electrocardio telemetering method and system, terminal and medium

Abstract

The invention relates to a remote electrocardio telemetering method and system, a terminal and a medium, and belongs to the technical field of electrocardio monitoring. The remote electrocardio telemetering method comprises the steps that a cloud end distributes monitoring equipment and initial parameters according to patient information; the edge end receives real-time electrocardiogram data, obtains environment data in combination with the position of a patient, collects physiological data through a wearable sensor, constructs a spatio-temporal context sensing map, fuses the data to generate situation enhanced electrocardiogram signal representation, inputs a comprehensive risk prediction model to obtain a dynamically updated prediction risk value and uploads the prediction risk value; the cloud judges the predicted risk value, if the predicted risk value is larger than a threshold value, a matched disease type is called, and differential prompts are generated and pushed to the patient and the doctor in combination with the co-disease relation network; if the electrocardiogram data does not exceed the threshold value, carrying out deep analysis on the electrocardiogram data, identifying abnormity, determining an abnormity level, generating a personalized report in combination with treatment history, pushing the personalized report, and triggering a clinical response protocol. The method has the beneficial effect of feeding back the abnormal electrocardiogram condition of the patient in time.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of electrocardio monitoring, and in particular to a remote electrocardio telemetry method, system, terminal and medium. BACKGROUND

[0002] As one of the major diseases that seriously threaten human health worldwide, the incidence and mortality of cardiovascular diseases are increasing year by year. Timely and accurate monitoring of electrocardio signals is of great significance for early diagnosis, treatment and prevention of cardiovascular diseases.

[0003] At present, the conventional electrocardio monitoring methods mainly include routine electrocardio graph examination and dynamic electrocardio graph monitoring. The routine electrocardio graph examination refers to recording the electrocardio activity of a patient in a lying state by an electrocardio graph, usually recording the heart cycle for about 10-20 seconds, about 6-20 times; it can observe the activity of the heart from multiple angles and make positioning diagnosis for diseases such as myocardial infarction and premature beat. The dynamic electrocardio graph monitoring is to let the patient wear a recording device to record the electrocardio signals continuously for 24 hours or more, and then the doctor analyzes the recorded data.

[0004] However, the routine electrocardio graph examination can only obtain little information about the heart condition, and the probability of recording arrhythmia in a limited time is very low, especially for some paroxysmal arrhythmia. Even if the patient has subjective symptoms, it is often difficult to capture in the routine electrocardio graph examination. Although the dynamic electrocardio graph monitoring can record electrocardio signals for a long time, it needs to be analyzed after the examination is completed, which cannot provide real-time feedback on the electrocardio abnormality of the patient, cannot provide risk prompt for the patient and the doctor in time, and is not conducive to early intervention and treatment of diseases. SUMMARY

[0005] In order to timely feedback the electrocardio abnormality of the patient, the present application provides a remote electrocardio telemetry method, system, terminal and medium.

[0006] In the first aspect, the present application provides a remote electrocardio telemetry method, which adopts the following technical scheme: A remote electrocardio telemetry method, comprising: The cloud end allocates corresponding monitoring equipment and initial monitoring parameters for the patient according to the patient information; The edge end receives real-time electrocardio data sent by the monitoring equipment; The edge end acquires environmental data of the location where the patient is located and multi-modal physiological data collected by a wearable sensor based on the location information of the patient; The edge end constructs a spatio-temporal context perception graph, aligns the real-time electrocardio data, environmental data and physiological data on a time axis, and generates a situation-enhanced electrocardio signal representation through attention mechanism weighted fusion; inputting the context-enhanced electrocardiosignal representation into a comprehensive risk prediction model with online incremental learning capability, obtaining a dynamically updated prediction risk value, and uploading to the cloud; judging whether the prediction risk value is greater than a dynamically adjusted risk threshold value by the cloud; if yes, according to the prediction risk value and its change rate, retrieving a matching predicted disease type from a multi-dimensional disease diagnosis knowledge graph, and combining the comorbidity relationship network in the patient's electronic medical record to generate differentiated risk prompt content, and pushing to the patient end and the doctor platform; if no, the cloud analyzes the real-time electrocardio data to determine whether it is abnormal; if yes, determining the abnormality level and generating a personalized risk assessment report combined with the treatment response history; pushing the risk assessment report to the patient end and the doctor platform, and triggering the corresponding level of clinical response protocol.

[0007] By adopting the above technical solution, the cloud precisely allocates monitoring equipment and initial parameters to the patient, ensuring the pertinence of monitoring. The edge end constructs a spatio-temporal context perception graph combined with multi-source data, generates a context-enhanced electrocardiosignal representation, making the data more comprehensive and in line with the actual situation of the patient. Inputting it into the comprehensive risk prediction model with online incremental learning can dynamically update the prediction risk value, improving the timeliness and accuracy of risk prediction. The cloud judges according to the risk threshold value, if the risk value exceeds the standard, retrieves the matching disease type and generates differentiated prompt combined with the comorbidity relationship network, providing precise early warning for the patient and the doctor; if it does not exceed the standard, deeply analyzes the electrocardio data, generates a personalized risk assessment report when abnormal, and triggers the clinical response protocol, which helps to early detect potential health problems, realizes personalized medical intervention, improves medical efficiency and quality, and improves the effect of patient health management.

[0008] Optionally, the step of allocating corresponding monitoring equipment and initial monitoring parameters to the patient according to the patient information includes: According to the patient information, retrieving the patient's historical physical examination report containing electrocardio data, past hospitalization records and genetic susceptibility markers; According to the patient's basic archives, generating an initial monitoring prescription, which includes sampling frequency, lead number, and filter parameter suggestion; Combining the initial monitoring prescription and the historical physical examination report, evaluating the prior probability of cardiovascular event occurrence of the patient through the Bayesian inference engine to generate a revised monitoring prescription; the revised monitoring prescription also includes individualized noise suppression strategy, wireless transmission power consumption control mode and emergency wake-up mechanism configuration parameters.

[0009] By adopting the technical scheme, the health history and potential risk factors of the patient can be comprehensively understood, the setting of the monitoring device is more targeted and scientific, and more accurate, efficient and reliable cardiovascular health monitoring is achieved.

[0010] Optionally, the step of analyzing the real-time electrocardio data to determine whether it is abnormal comprises: extracting key electrocardio feature parameters in the real-time electrocardio data; calling a patient historical data set, and constructing an individualized electrocardio baseline dynamic envelope by using a sliding window method, and introducing a variational mode decomposition technique to remove motion artifacts and electromyographic interference, and improving the signal-to-noise ratio; combining a preset general medical electrocardio standard threshold value and patient living habit data to generate a context-sensitive personalized baseline matching the current physiological state of the patient; comparing the real-time key electrocardio feature parameters with the corresponding context-sensitive personalized baseline, and calculating parameter offset values and statistical significance thereof; when the parameter offset value exceeds an offset threshold value and the duration exceeds a preset time window, determining that it is abnormal.

[0011] By adopting the technical scheme, the key electrocardio feature parameters are extracted, which can focus on the information with the most diagnostic value in the electrocardio data, and provide a core basis for subsequent analysis. The patient historical data set is called, the individualized electrocardio baseline dynamic envelope is constructed by using the sliding window method, the variational mode decomposition technique is combined to remove motion artifacts and electromyographic interference, the signal-to-noise ratio of the data is effectively improved, the electrocardio data is more pure and reliable, and a foundation is laid for accurate analysis. The context-sensitive personalized baseline is generated by combining the general medical electrocardio standard threshold value and the patient living habit data, the individual differences and the current physiological state of the patient are fully considered, the judgment standard is more in line with the actual situation. The real-time key electrocardio feature parameters are compared with the personalized baseline, and the parameter offset values and the statistical significance thereof are calculated, which can accurately capture the change of the electrocardio data. This comprehensive judgment method improves the accuracy and reliability of abnormal judgment, and helps to timely find the electrocardio abnormality of the patient.

[0012] Optionally, before the step of determining that it is abnormal when the parameter offset value exceeds the offset threshold value and the duration exceeds the preset time window, the step comprises: if the patient does not have a sufficient length of historical data, calling a historical data set of the same type of patient of the same age group and the same gender, and screening a subset with a similar basic disease spectrum; performing standardized feature extraction on each sample in the subset to establish a group-level electrocardio feature distribution model; fitting an age-gender-pathology joint response surface of each key electrocardio parameter in the group-level electrocardio feature distribution model by using a Gaussian process regression to generate an initial personalized baseline function; Further, K-means clustering is used to identify the three typical physiological mode clusters closest to the target patient, and the weighted average of the corresponding baseline is taken as the virtual individualized baseline; The real-time key ECG feature parameters are compared with the corresponding virtual individualized baseline to obtain parameter offset values, and a confidence score is calculated, and when the confidence score is lower than the set lower limit, an artificial review process is triggered.

[0013] By using the above technical solutions, the reference data volume is expanded by retrieving the same range of historical data sets, the population rule is grasped by standardizing the data subset and modeling, the personalized baseline is generated by using Gaussian process regression and K-means clustering to fit the actual patient, the offset value is obtained by comparing the real-time parameters with the baseline, and the confidence score is calculated, and when the score is low, the artificial review is triggered. Overall, it provides more rich information, standard comparability and personalized baseline for ECG data analysis, ensures the accuracy and reliability of ECG abnormality judgment, and effectively supports patient health monitoring.

[0014] Optionally, the step of determining the abnormality level comprises: According to the abnormal type of real-time ECG data, the current medication scheme and historical diagnosis and treatment records of the patient are retrieved; Based on the embedded drug-ECG correlation rule base, the causal possibility between abnormal ECG features and medication is analyzed, and a medication risk index is generated; According to the patient's recent operation history, complications and treatment stage, a clinical weight factor coefficient is generated; The heart rate variability and autonomic nervous balance index are introduced as physiological elasticity indicators to modify the risk total score calculation formula; the final risk total score = parameter offset value x clinical weight factor coefficient x (1+ medication risk index / n) x (1- heart rate variability improvement coefficient); According to the final risk total score, the abnormality level is divided.

[0015] By using the above technical solutions, the final risk total score calculation formula integrates the parameter offset value, the clinical weight factor coefficient, the medication risk index and the heart rate variability improvement coefficient, and generates a comprehensive risk assessment result, which provides a clear risk level division for the patient, and helps to take appropriate medical measures in time, improves the scientificity and effectiveness of patient management.

[0016] Optionally, the remote ECG telemetry method further comprises: A two-way encrypted patient-doctor real-time communication mechanism is established to support text, voice, picture upload and video consultation access; The patient is allowed to initiate an emergency consultation request through the patient terminal one key, and the system automatically adds the latest m-minute ECG segment and risk assessment summary; The doctor is allowed to initiate a structured follow-up reminder through the doctor platform; After pushing the risk assessment report, the system automatically generates a structured follow-up task sheet, and sets follow-up priority, response time limit and responsible physician role according to the abnormality level.

[0017] By adopting the technical scheme, the communication, emergency response and follow-up functions are integrated to form a closed-loop management, improve the reliability of remote electrocardiogram telemetry (such as reducing the misdiagnosis rate), and at the same time, reduce the work burden of doctors.

[0018] Optionally, the remote electrocardiogram telemetry method further comprises: In the case that two consecutive high-risk abnormal events are detected and the patient has no active response, a non-invasive physiological intervention plan is started: the edge end regulates the wearable monitoring device to output a specific frequency of transcutaneous vagus nerve stimulation pulse signals to adjust the autonomic nervous tension and suppress sympathetic storm; voice soothing instructions and breathing guidance animations are pushed to the patient end at the same time to induce vagus nerve activation; During the intervention process, the change of heart rate variability index is continuously monitored, and if the LF / HF ratio decreases by ≥ a first threshold value or the RMSSD increases by ≥ a second threshold value, it is determined that the intervention is effective, the stimulation is paused, and the efficacy data is recorded; If not, automatically upgrade to the remote emergency dispatch process, link the 120 emergency center and send the patient's location, vital sign snapshot and electronic health record summary.

[0019] By adopting the technical scheme, in the case that two consecutive high-risk abnormal events are detected and the patient has no active response, the plan is started, the wearable monitoring device is regulated to output a specific frequency of transcutaneous vagus nerve stimulation pulse signals to adjust the autonomic nervous tension and suppress sympathetic storm, which helps to quickly stabilize the patient's heart rhythm. Voice soothing instructions and breathing guidance animations are pushed to the patient end at the same time to induce vagus nerve activation, further enhancing the intervention effect. During the intervention process, the change of heart rate variability index is continuously monitored, and if the LF / HF ratio decreases by ≥ a first threshold value or the RMSSD increases by ≥ a second threshold value, it is determined that the intervention is effective, the stimulation is paused, and the efficacy data is recorded, providing a basis for subsequent treatment. If not, automatically upgrade to the remote emergency dispatch process, link the 120 emergency center and send the patient's location, vital sign snapshot and electronic health record summary, to ensure that the patient can receive professional treatment in a timely manner.

[0020] In a second aspect, the present application provides a remote electrocardiogram telemetry system, which adopts the following technical scheme: A remote electrocardiogram telemetry system comprises: The cloud end is configured to assign a corresponding monitoring device and initial monitoring parameters to a patient according to patient information, and to determine whether the predicted risk value is greater than a dynamically adjusted risk threshold value, and if so, to retrieve a matching predicted disease type from a multi-dimensional disease diagnosis knowledge graph according to the predicted risk value and its change rate, generate differentiated risk prompt content in combination with a comorbidity relationship network in the patient's electronic medical record, and push the risk prompt content; if not, the cloud end analyzes the real-time electrocardiogram data to determine whether it is abnormal; if abnormal, determines an abnormality level based on the spatial distribution characteristics and duration of the abnormal waveform, generates a personalized risk assessment report in combination with treatment response history, and pushes the risk assessment report to trigger a clinical response protocol of a corresponding level; The edge end is configured to receive real-time electrocardiogram data sent by the monitoring device, acquire environmental data of a location where the patient is located based on the location information of the patient, and acquire multi-modal physiological data collected by a wearable sensor; and to construct a spatio-temporal context perception graph, align the real-time electrocardiogram data, the environmental data and the physiological data on a time axis, and generate a context-enhanced electrocardiogram signal representation through attention mechanism weighted fusion, and input the context-enhanced electrocardiogram signal representation into a comprehensive risk prediction model with online incremental learning capability to obtain a dynamically updated predicted risk value and upload the predicted risk value to the cloud end. The patient end is configured to receive the differentiated risk prompt content and the risk assessment report. The doctor platform is configured to receive the differentiated risk prompt content and the risk assessment report.

[0021] In a third aspect, the present application provides a terminal, which adopts the following technical solution: A terminal comprises: A memory storing a remote electrocardiogram telemetry program; A processor configured to execute the program stored on the memory to implement the steps of the remote electrocardiogram telemetry method.

[0022] In a fourth aspect, the present application provides a computer readable storage medium, which adopts the following technical solution: A computer readable storage medium storing a computer program capable of being loaded and executed by a processor to implement the remote electrocardiogram telemetry method.

[0023] In summary, the present application has at least the following beneficial effects: The cloud accurately allocates monitoring equipment and initial parameters for the patient, ensuring the pertinence of monitoring. The edge end constructs a spatio-temporal context perception graph combining multi-source data, generates a context-enhanced electrocardiogram signal representation, and makes the data more comprehensive and in line with the actual condition of the patient. The comprehensive risk prediction model based on online incremental learning is inputted, which can dynamically update the predicted risk value and improve the timeliness and accuracy of risk prediction. The cloud makes a judgment according to the risk threshold. If the risk value exceeds the standard, the matching disease type is retrieved and a differential prompt is generated in combination with the comorbidity relationship network, so as to provide accurate early warning for the patient and the doctor. If the risk value does not exceed the standard, the electrocardiogram data is deeply analyzed, an individualized risk assessment report is generated when an abnormality occurs, and a clinical response protocol is triggered, which is helpful for early detection of potential health problems, realization of individualized medical intervention, improvement of medical efficiency and quality, and improvement of the effect of patient health management. BRIEF DESCRIPTION OF DRAWINGS

[0024] Figure 1 is a first flowchart of the method embodiment of the present application; Figure 2 is a second flowchart of the method embodiment of the present application; Figure 3 is a third flowchart of the method embodiment of the present application; Figure 4 is a fourth flowchart of the method embodiment of the present application; Figure 5 is a fifth flowchart of the method embodiment of the present application; Figure 6 is a sixth flowchart of the method embodiment of the present application; Figure 7 is a seventh flowchart of the method embodiment of the present application; Figure 8 is a structural block diagram of the system embodiment of the present application.

[0025] Marked for explanation: 1, electrocardiogram monitoring equipment; 2, wearable sensor; 3, edge end; 4, cloud; 5, patient end; 6, doctor platform. DETAILED DESCRIPTION

[0026] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the following will combine the drawings in the embodiments of the present application to clearly and completely describe the technical scheme in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application. Figure 1 -attached Figure 8 , the technical scheme in the embodiments of the present application is clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0027] The first embodiment of the present application discloses a remote electrocardiogram telemetry method. The implementation is based on the modern medical information technology framework, integrating cloud computing, edge computing, machine learning algorithms and medical database systems to ensure technical feasibility and clinical practicability. The whole system adopts modular design, uses RESTful API for data exchange, MySQL or MongoDB stores patient data, TensorFlow or PyTorch implements machine learning model, and follows HL7 / FHIR medical data standard to ensure interoperability.

[0028] Referring to Figure 1 and Figure 2 The remote electrocardiogram telemetry method can include S110-S200: S110, the cloud end allocates corresponding monitoring equipment and initial monitoring parameters for the patient according to the patient information; S120, the edge end receives the real-time electrocardiogram data sent by the monitoring equipment; S130, the edge end obtains environmental data of the location where the patient is located based on the location information of the patient, and multi-modal physiological data collected by wearable sensors; S140, the edge end constructs a spatio-temporal context perception graph, aligns the real-time electrocardiogram data, environmental data and physiological data on the time axis, and generates a context-enhanced electrocardiogram signal representation through attention mechanism weighted fusion; S150, input the context-enhanced electrocardiogram signal representation into a comprehensive risk prediction model with online incremental learning ability, obtain a dynamically updated prediction risk value, and upload to the cloud end; S160, the cloud end judges whether the prediction risk value is greater than the dynamically adjusted risk threshold; S170, if yes, according to the prediction risk value and its change rate, the matching prediction disease type is retrieved from the multi-dimensional disease diagnosis knowledge graph, and the differential risk prompt content is generated combined with the comorbidity relationship network in the patient's electronic medical record, and pushed to the patient end and the doctor platform; S180, if not, the cloud end analyzes the real-time electrocardiogram data to determine whether it is abnormal; S190, if yes, determine the abnormality level, and generate a personalized risk assessment report combined with the treatment response history; S200, push the risk assessment report to the patient end and the doctor platform, and trigger the corresponding level of clinical response protocol.

[0029] Specifically, in step S120, the edge computing node (such as Raspberry Pi or a dedicated gateway device) receives the raw electrocardiogram data stream sent by the electrocardiogram monitoring device (such as Holter or wearable ECG patch) in real time through Bluetooth 5.0 or MQTT protocol, and the data is marked with a 10ms timestamp and buffered to the local memory queue.

[0030] In step S130, the edge end obtains location information using the GPS module or Wi-Fi positioning service of the patient's smartphone, integrates real-time environmental data (such as temperature, humidity, and altitude) through the environmental sensor API (such as Weather Underground), and collects multi-modal physiological data (such as blood oxygen saturation, skin conductance, and accelerometer data) through BLE connection with wearable sensors (such as Fitbit or Apple Watch). All data is serialized using the Protobuf protocol to reduce transmission delay.

[0031] In step S140, the edge end constructs a spatio-temporal context-aware graph: first, the real-time electrocardiogram data, environmental data, and physiological data are aligned on the time axis using the dynamic time warping (DTW) algorithm to compensate for clock drift between devices; then, the attention mechanism (a lightweight model based on the Transformer architecture) is applied to weight the fused data, for example, giving higher weight to high heart rate variability (HRV) data, generating a context-enhanced electrocardiogram signal representation, which is stored as an HDF5 format file for subsequent analysis. This representation is input into a comprehensive risk prediction model (such as an online incremental learning LSTM neural network), which outputs a dynamic risk value (range 0-100) and uploads it to the cloud server via HTTPS protocol; the model is initially trained based on historical data sets and updated in real time using stochastic gradient descent (SGD), with the model retrained every 5 minutes to adapt to new data streams. The cloud server determines whether the predicted risk value is greater than the dynamically adjusted risk threshold, which is based on the patient's historical risk mean value calculated by a sliding window plus 2 times the standard deviation.

[0032] If the predicted risk value exceeds the threshold, step S170 is executed: the system calls a multi-dimensional disease diagnosis knowledge graph (built on a Neo4j graph database, with nodes including symptoms, diseases, and treatment methods), matches the predicted disease type (such as atrial fibrillation or myocardial ischemia) based on the risk value change rate (such as derivative calculation), queries the comorbidity relationship network in the patient's electronic medical record (using SPARQL to query RDF data), generates differentiated risk prompt content (such as "high-risk atrial fibrillation, recommend immediate rest"), and pushes it to the patient's mobile App and doctor's Web platform through Firebase Cloud Messaging. The prompt content can be customized, such as considering the patient's language preference, etc.

[0033] Refer toFigure 3 , S110, the step of assigning a patient with a corresponding monitoring device and initial monitoring parameters according to patient information includes S310-S330: S310, according to the patient information, the historical physical examination report containing the electrocardiogram data, the past hospitalization record and the genetic susceptibility marker of the patient are called; S320, according to the patient's basic file, an initial monitoring prescription is generated, which includes sampling frequency, lead number, filter parameter suggestion; S330, combined with the initial monitoring prescription and the historical physical examination report, the prior probability of the occurrence of cardiovascular events of the patient is evaluated by the Bayesian inference engine to generate a revised monitoring prescription; The revised monitoring prescription also includes individualized noise suppression strategy, wireless transmission power consumption control mode and emergency wake-up mechanism configuration parameters.

[0034] Specifically, the cloud server first calls the standardized interface (such as HL7 FHIR) through the patient's unique identifier (such as medical insurance ID or electronic health record number) to automatically call its physical examination report (including 12-lead electrocardiogram data) within the past 5 years, cardiovascular event records (such as myocardial infarction, atrial fibrillation history) in hospitalization records and genetic test results (such as KCNH2 gene mutation and other arrhythmia susceptibility markers), and stores them in an encrypted distributed database; The standardized interface interfaces with the hospital information system (HIS), the laboratory information system (LIS) and the genetic database.

[0035] The clinical decision support module built-in cloud server matches the initial monitoring prescription from the preset clinical guideline template library according to the patient's age, underlying diseases (such as hypertension classification, diabetes duration) and medication history (such as whether to take beta blockers), for example, setting 2000Hz sampling frequency, proton multi-lead dynamic electrocardiogram recorder synchronous acquisition and 0.05-150Hz band-pass filter parameters for heart failure patients.

[0036] Subsequently, through the Bayesian inference engine (such as using the PyMC3 library to build a probabilistic graph model), the abnormal event rate (such as the frequency of past ventricular tachycardia episodes) in the historical data is taken as the prior probability, and the posterior probability is calculated by combining the dynamic risk factors (such as recent blood pressure fluctuations), and the prescription is revised accordingly, such as increasing the motion artifact suppression algorithm weight for high-intensity exercise patients, enabling low-power Bluetooth transmission mode (wake up every 30 seconds) for elderly patients, and configuring emergency wake-up mechanism when heart rate >100 times / min. Related parameters in the revised monitoring prescription can be issued to the edge device through the cloud.

[0037] Reference Figure 4 , S180, the step of analyzing real-time electrocardiogram data to determine whether it is abnormal includes S410-S450: S410, extract key electrocardiogram feature parameters from real-time electrocardiogram data; S420, retrieve patient historical dataset, and construct individualized ECG baseline dynamic envelope using sliding window method, and introduce variational mode decomposition technique to remove motion artifact and electromyographic interference, and improve signal-to-noise ratio; S430, combine the preset general medical ECG standard threshold and patient lifestyle data to generate a context-sensitive personalized baseline matching the current physiological state of the patient; S440, compare the real-time key ECG feature parameters with the corresponding context-sensitive personalized baseline, and calculate the parameter offset value and its statistical significance; S450, when the parameter offset value exceeds the offset threshold and the duration exceeds the preset time window, determine it as an abnormality.

[0038] Specifically, if the risk value does not exceed the standard, S180 is performed: in S410, the cloud server uses a QRS detection algorithm (such as Pan-Tompkins algorithm) to extract key ECG feature parameters (such as RR interval, QT interval and ST segment offset) from real-time ECG data. The parameters are stored in the form of floating point array.

[0039] In S420, the system retrieves the patient historical dataset, applies a sliding window method (window size 30 minutes, step 5 minutes) to construct an individualized ECG baseline dynamic envelope (upper envelope is the maximum value, lower envelope is the minimum value), and introduces variational mode decomposition (VMD) technique to decompose signal components, remove motion artifact and electromyographic interference (set modal number K=5, penalty parameter α=2000), and improve signal-to-noise ratio to more than 30dB.

[0040] In S430, the preset general medical ECG standard threshold (such as QTc<440ms in AHA guidelines) and patient lifestyle data (sleep and diet patterns extracted from mobile App logs) are combined to generate a context-sensitive personalized baseline (such as night baseline adjusted to accommodate sleep period heart rate decrease) through a linear regression model.

[0041] In S440, the system compares the real-time key ECG feature parameters with the corresponding context-sensitive personalized baseline, calculates the parameter offset value using Z-score, and evaluates the statistical significance (p<0.05 is considered significant) through t-test. The offset value is stored as a time series.

[0042] Reference Figure 5 The steps before determining the abnormality when the parameter offset value exceeds the offset threshold and the duration exceeds the preset time window include S510-S550: S510, if the patient does not have sufficient length of historical data, retrieve the historical dataset of similar patients of the same age group and gender, and select a subset with similar basic disease spectrum; S520, performing standardized feature extraction on each sample in the subset to establish a population-level electrocardiogram feature distribution model; S530, fitting an age-gender-pathology joint response surface of each key electrocardiogram parameter in the population-level electrocardiogram feature distribution model using Gaussian process regression to generate an initial individualized baseline function; S540, further using K-means clustering to identify three typical physiological mode clusters closest to the target patient, and taking the weighted average of the corresponding baseline as a virtual individualized baseline; S550, comparing real-time key electrocardiogram feature parameters with the corresponding virtual individualized baseline to obtain parameter offset values and calculate a confidence score, and triggering an artificial review process when the confidence score is below a set lower limit.

[0043] Specifically, if the patient lacks sufficient historical data (such as a new patient or less than 30 days of data), the historical data set of the same type of patient in the same age group and the same gender is retrieved (from the cloud database, screening conditions such as age ± 5 years, same gender, and similar spectrum of underlying diseases).

[0044] The system performs standardized feature extraction on each individual in the subset of the same type of patients selected above, covering key electrocardiogram parameters such as QRS complex width, QTc interval, P-wave amplitude variability, heart rate variability (HRV) indicators (such as RMSSD, SDNN), T-wave morphology asymmetry index, and unifying the sampling frequency to 500 Hz to ensure time domain alignment; then eliminating the bias caused by device differences through Z-score normalization, and constructing a population-level electrocardiogram feature multi-dimensional probability distribution model based on the non-parametric kernel density estimation method. This model not only captures the marginal distribution characteristics of each parameter, but also uses Copula function to model the non-linear dependence structure between parameters, thus forming a statistical benchmark framework that can reflect real physiological heterogeneity.

[0045] S530 uses Gaussian process regression (GPR) technology to fit the response surface of each electrocardiogram parameter with respect to demographic and pathological factors, with age as the main covariate, gender as the classification covariate, and key comorbidity load (such as the presence of hypertension, atrial fibrillation, and renal dysfunction) coded as a binary covariate matrix; for example, the QTc interval is modeled as a non-monotonic function of age, which tends to be stable in middle age but shows an accelerated extension trend in the elderly group, with an additional 50-60 ms effect of women being generally longer than men and an additional 15-20 ms tailing effect caused by diabetes; the initial individualized baseline function generated from this can not only extrapolate the "expected normal value" of the target patient, but also provide a prediction uncertainty interval, laying the foundation for subsequent dynamic adjustment.

[0046] S540 is to further approach the individual true physiological state, the system applies K-means clustering algorithm (set k=3) on the established population model, and identifies three typical electrocardio mode clusters based on the first five-dimensional feature space after dimensionality reduction based on principal component analysis (cumulative explained variance>85%): for example, "low voltage with slow conduction type", "increased sympathetic tension type" and "increased repolarization dispersion type"; then the Mahalanobis distance between the target patient static feature vector and each cluster center is calculated to obtain three similarity weights (normalized by softmax), and the GPR predicted baseline of the corresponding cluster sample is fused according to the weight to form a virtual individualized baseline trajectory with population representativeness and individual closeness.

[0047] Finally, in the S550 step, the system continuously receives the real-time electrocardio parameter stream, compares it with the aforementioned virtual baseline point by point, calculates the standardized deviation of each parameter (such as ΔQTc =|measured value-predicted mean| / predicted standard deviation), and combines the trend slope, mutation amplitude and multi-parameter collaborative deviation in the sliding window to output a comprehensive confidence score (range 0-1) using a Bayesian belief network to fuse multiple evidence sources; when the score is lower than the preset threshold (such as 0.3) and lasts for more than a preset time window (such as 15 minutes), the system triggers the manual review process immediately, pushes the warning to the on-duty doctor terminal and attaches the visual comparison atlas and potential cause prompt.

[0048] Referring to Figure 6 , the step of determining the abnormality level includes S610-S650: S610, according to the abnormal type of real-time electrocardio data, the current medication scheme and historical diagnosis and treatment records of the patient are called; S620, based on the embedded drug-electrocardio correlation rule base, the causal possibility between abnormal electrocardio features and medication is analyzed to generate a medication risk index; S630, according to the patient's recent operation history, complications and treatment stage, a clinical weight factor coefficient is generated; S640, the heart rate variability and autonomic nervous balance index are introduced as physiological elasticity indicators to modify the risk total score calculation formula; the final risk total score=parameter deviation×clinical weight factor coefficient×(1+medication risk index / n)×(1-heart rate variability improvement coefficient); S650, according to the final risk total score, the abnormality level is divided.

[0049] Specifically, in the S610 step, the system calls the current medication scheme (obtained from the pharmacy database) and historical diagnosis and treatment records (including allergy history) of the patient according to the abnormal type (such as ST segment elevation), and queries the associated data through SQL.

[0050] S620, the system activates the embedded "drug-ECG correlation rule knowledge base", which is trained by evidence-based medical guidelines (such as ACC / AHA), FDA adverse drug reaction database (FAERS), and real-world research literature, and uses ontology modeling to establish a causal reasoning graph between drug categories (such as class III antiarrhythmic drugs, mental drugs with QT interval prolongation risk) and specific ECG abnormal patterns. Combined with Bayesian network, it evaluates whether the current abnormality is likely to be induced by existing medication, and quantitatively generates a "medication risk index" - for example, when a patient is taking sotalol and T wave alternans occurs, the system calculates a risk index of 7.2 (full score 10) based on prior probability and likelihood ratio, reflecting its high potential arrhythmogenic effect.

[0051] In step S630, the system further extracts the patient's surgical intervention records in the past 30 days (such as coronary stent implantation), comorbidity status (such as diabetic nephropathy, chronic obstructive pulmonary disease), and current treatment stage (acute phase, rehabilitation phase, or hospice care), and uses analytic hierarchy process (AHP) to build a clinical weight factor system: for example, cardiac autonomic nervous function instability within 72 hours after surgery is assigned a weight of 1.8, while stable chronic heart failure patients are assigned a weight of 1.2. This coefficient reflects the different risk levels of the same ECG abnormality in different clinical backgrounds.

[0052] In step S640, the system introduces more detailed physiological elasticity evaluation indicators, namely RMSSD and SDNN parameters in heart rate variability (HRV), and combines LF / HF power ratio as an autonomic nervous balance index to measure the body's ability to adjust to external disturbances. If the 24-hour HRV shows an upward trend, the "heart rate variability improvement coefficient" is defined as a positive number (such as 0.15), forming a negative correction term in the final risk total score formula, reflecting the buffering effect of the patient's intrinsic recovery potential on the overall risk.

[0053] In step S650, the system classifies the abnormality level according to the final risk total score (such as low risk <20, medium risk 20-50, high risk >50), and generates a personalized risk assessment report (PDF format, including charts and recommended actions) combined with treatment response history (extracted from EHR past intervention effects).

[0054] Finally, the cloud server pushes the risk assessment report to the patient end (mobile App notification) and the doctor platform (EHR integration) through the message queue (such as RabbitMQ), and triggers the clinical response protocol (such as automatically booking emergency or sending ambulance dispatch instructions when the risk is high), and the protocol level is determined by the final risk total score, ensuring end-to-end closed-loop management.

[0055] In addition, the remote ECG telemetry method can further include the following steps: A two-way encrypted patient-doctor real-time communication mechanism is established to support text, voice, picture upload and video consultation access; patients are allowed to initiate emergency consultation requests through the patient terminal with one key, and the system automatically attaches the latest m-minute ECG segment and risk assessment summary; doctors are allowed to initiate structured follow-up reminders through the doctor platform; after pushing the risk assessment report, the system automatically generates a structured follow-up task sheet, and sets the follow-up priority, response time limit and responsible doctor role according to the abnormal level.

[0056] Specifically, the two-way encrypted communication adopts end-to-end AES-256 encryption, supports patients to upload chest tightness symptom description (text) and sublingual nitroglycerin ECG (picture) through the APP, and the doctor platform integrates the Zoom video consultation interface with a delay control within 200ms. When the patient initiates an emergency consultation, the system automatically intercepts the latest 30-second-10-minute ECG segment (including abnormal waveform markers) and risk assessment summary (such as "3 episodes of ventricular tachycardia occurred 2 minutes ago, risk value 92 points") and pushes it to the top of the doctor terminal. The structured follow-up function generates a task sheet based on ICD-10 coding (such as "atrial fibrillation patients need to record INR value every week"), and the priority is set according to the abnormal level (3-level abnormality requires follow-up within 24 hours).

[0057] Further, the remote electrocardiogram telemetry method further comprises S710-S730: S710, in the case of detecting two consecutive high-risk abnormal events and no active response from the patient, starting a non-invasive physiological intervention plan: the edge end controls the monitoring device to output a specific frequency of transcutaneous vagus nerve stimulation pulse signal to adjust the autonomic nervous tension and suppress sympathetic storm; synchronously push voice soothing instructions and breathing guidance animation to the patient terminal to induce vagus nerve activation; S720, continuously monitoring the heart rate variability index change during the intervention process, if the LF / HF ratio decreases by ≥ the first threshold value or the RMSSD increases by ≥ the second threshold value, it is determined that the intervention is effective, the stimulation is suspended and the efficacy data is recorded; S730, if not effective, automatically upgrade to a remote emergency dispatch process, link the 120 emergency center and send the patient's location, vital signs snapshot and electronic health record summary.

[0058] Specifically, after detecting two consecutive high-risk abnormalities (such as ventricular tachycardia for more than 30 seconds) and the patient not responding, the edge end outputs a 20-50Hz, 0.5mA pulse signal through the built-in transcutaneous vagus nerve stimulation module of the electrocardiogram monitoring device, synchronously pushes voice soothing instructions and animations to the patient end, and guides the patient to perform the "4-7-8 breathing method" (inhale for 4 seconds, hold breath for 7 seconds, and exhale for 8 seconds). The intervention effect is evaluated by monitoring the LF / HF ratio (the first threshold is set to 20%) and the RMSSD value (the second threshold is set to 15ms), and if the threshold is met, the stimulation is suspended and the data is recorded (such as "50% reduction in the number of premature ventricular contractions after stimulation"); if it is invalid, the system automatically dials 120, sends the patient's accurate location through Beidou positioning, and uploads the summary of the patient's vital signs (heart rate, blood pressure, blood oxygen) and electronic health records (allergies, medication history) to the emergency center dispatch system in the past 1 hour.

[0059] Reference Figure 8 For ease of understanding, the following examples are provided: a remote electrocardiogram telemetry system, comprising: an electrocardiogram monitoring device 1, a wearable sensor 2, an edge end 3, a cloud end 4, a patient end 5 and a doctor platform 6. The electrocardiogram monitoring device 1 and the wearable sensor 2 both interact with the edge end 3, the edge end 3 and the cloud end 4 interact with each other, and the cloud end 4 interacts with the patient end 5 and the doctor platform 6. The cloud end 4 can transmit electrocardiogram data to the doctor platform 6 in real time for the doctor to review at any time; and the doctor platform 6 can also retrieve real-time data, historical data, etc. stored in the cloud end 4.

[0060] The cloud end 4 is used to allocate corresponding electrocardiogram monitoring devices 1 and initial monitoring parameters to the patient according to the patient information, and to judge whether the predicted risk value is greater than the dynamically adjusted risk threshold. If yes, the matching predicted disease type is retrieved from the multi-dimensional disease diagnosis knowledge graph according to the predicted risk value and its change rate, and the differential risk prompt content is generated in combination with the comorbidity relationship network in the patient's electronic medical record, and is pushed to the patient end 5 and the doctor platform 6; if not, the cloud end 4 analyzes the real-time electrocardiogram data to determine whether it is abnormal; if it is abnormal, the abnormal level is determined based on the spatial distribution characteristics and duration of the abnormal waveform, and the personalized risk assessment report is generated in combination with the treatment response history, and the risk assessment report is pushed to the patient end 5 and the doctor platform 6 to trigger the corresponding level of clinical response protocol.

[0061] The edge end 3 is used for receiving real-time electrocardiogram data sent by the electrocardiogram monitoring device 1, obtaining environmental data of a position where the patient is located based on position information of the patient, and collecting multi-modal physiological data through the wearable sensor 2; and for constructing a spatio-temporal context perception graph, aligning the real-time electrocardiogram data, the environmental data and the physiological data on a time axis, and generating a context-enhanced electrocardiogram signal representation through attention mechanism weighted fusion, and inputting the context-enhanced electrocardiogram signal representation into a comprehensive risk prediction model with online incremental learning capability to obtain a dynamically updated prediction risk value and upload the prediction risk value to the cloud 4.

[0062] In addition, in order to realize efficient access and low-delay data interaction of the electrocardiogram monitoring device 1, the system adopts an adapter mode at the device access layer to support plug-and-play of multiple types of monitoring devices through dynamic plug-in design. Specifically, when a new electrocardiogram monitoring device 1 is added, only the corresponding data format conversion plug-in needs to be developed and registered to the plug-in management module of the edge node, without modifying the core access program to complete protocol adaptation and data analysis, which significantly reduces the development cost of device compatibility. At the data transmission layer, in view of the problems of frequent handshake of the traditional HTTP protocol and high resource occupation of the WebSocket long connection, the system adopts a custom reliable transmission protocol based on UDP, through built-in retransmission mechanism, sliding window flow control and data fragmentation reorganization strategy, while maintaining the low delay characteristic of UDP, the transmission reliability is guaranteed. The protocol supports the server to actively push configuration instructions or warning information to the terminal, realizing real two-way real-time communication. In addition, the real-time electrocardiogram data is compressed by using the LZW lossless compression algorithm, and the differential coding is performed combined with the periodic characteristics of the electrocardiogram signal, so that the data amount is compressed by 40%-60% without losing key waveform information, which effectively relieves the bandwidth pressure of the edge end 3 to the cloud 3, reduces the risk of data buffer overflow, and improves the transmission security through compressed package encryption.

[0063] The third embodiment of the present application provides a terminal, which can include a memory and a processor as an implementation manner of the terminal, and the terminal can further include: The memory is configured to store a remote electrocardiogram telemetry program. The processor is configured to execute the program stored on the memory to implement the steps of the remote electrocardiogram telemetry method.

[0064] The memory can be in communication connection with the processor through a communication bus, and the communication bus can be an address bus, a data bus, a control bus, etc.

[0065] In addition, the memory can include a random access memory (RAM) and can also include a non-volatile memory (NVM), such as at least one disk memory.

[0066] The processor can be a general purpose processor, including a central processing unit (CPU), a network processor (NP), or the like; and can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the like.

[0067] The fourth embodiment of the present application provides a computer readable storage medium, which stores a computer program capable of being loaded by a processor and executing the remote electrocardiogram telemetry method.

[0068] The computer readable storage medium can be any available medium or a data storage device, such as a server, data center, or the like, including one or more available media sets. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid state disk), or the like.

[0069] The above are preferred embodiments of the present application, and are not sequentially limiting the protection scope of the present application. Any feature disclosed in the specification (including the abstract and the drawings) can be replaced by other equivalent or similar purpose alternative features, unless specifically described. That is, each feature is only an example of a series of equivalent or similar features, unless specifically described.

Claims

1. A method of remote electrocardiographic telemetry, characterized in that, The method comprises the following steps: The cloud assigns a corresponding monitoring device and initial monitoring parameters to the patient according to the patient information; The edge receives real-time electrocardiogram data sent by the monitoring device; The edge obtains environmental data of the location where the patient is located based on the patient's location information, and multi-modal physiological data collected by wearable sensors; The edge constructs a spatio-temporal context perception graph, aligns the real-time electrocardiogram data, environmental data and physiological data on the time axis, and generates a context-enhanced electrocardiogram signal representation through attention mechanism weighted fusion; The context-enhanced electrocardiogram signal representation is input into a comprehensive risk prediction model with online incremental learning capability to obtain a dynamically updated prediction risk value, which is uploaded to the cloud; The cloud judges whether the prediction risk value is greater than the dynamically adjusted risk threshold; If yes, the matching predicted disease type is retrieved from the multi-dimensional disease diagnosis knowledge graph according to the prediction risk value and its change rate, and the differential risk prompt content is generated by combining the comorbidity relationship network in the patient's electronic medical record, and pushed to the patient end and the doctor platform; If no, the real-time electrocardiogram data is analyzed to determine whether it is abnormal; If yes, the abnormality level is determined, and a personalized risk assessment report is generated by combining the treatment response history; The risk assessment report is pushed to the patient end and the doctor platform, and the corresponding level of clinical response protocol is triggered.

2. The remote electrocardiogram telemetry method of claim 1, wherein, The step of assigning a corresponding monitoring device and initial monitoring parameters to the patient according to the patient information comprises: According to the patient information, the patient's historical physical examination report containing electrocardiogram data, past hospitalization records and genetic susceptibility markers are retrieved; According to the patient's basic archives, an initial monitoring prescription is generated, which includes sampling frequency, lead number, and filter parameter suggestion; Combined with the initial monitoring prescription and the historical physical examination report, the patient's prior probability of cardiovascular event occurrence is evaluated by a Bayesian inference engine to generate a revised monitoring prescription; the revised monitoring prescription also includes individualized noise suppression strategy, wireless transmission power consumption control mode and emergency wake-up mechanism configuration parameters.

3. The method of claim 1, wherein, The step of analyzing the real-time electrocardiogram data to determine whether it is abnormal comprises: Extracting key electrocardiogram feature parameters from the real-time electrocardiogram data; Retrieving the patient's historical data set and constructing an individualized electrocardiogram baseline dynamic envelope using the sliding window method, and introducing variational mode decomposition technology to remove motion artifacts and electromyographic interference to improve signal-to-noise ratio; Combined with the pre-set universal medical electrocardiogram standard threshold and the patient's living habit data, a context-sensitive personalized baseline matching the patient's current physiological state is generated; Compare the real-time key electrocardiogram feature parameters with the corresponding context-sensitive personalized baseline, and calculate the parameter offset value and its statistical significance; When the parameter offset value exceeds the offset threshold and the duration exceeds the pre-set time window, it is determined to be abnormal.

4. The remote electrocardiogram telemetry method of claim 3, wherein, The steps before determining to be abnormal when the parameter offset value exceeds the offset threshold and the duration exceeds the pre-set time window comprise: If the patient does not have enough length of historical data, the historical data set of the same type of patient with similar basic disease spectrum is selected from the same age group and the same gender population. Performing standardized feature extraction on each sample in the subset, establishing a population-level electrocardiogram feature distribution model; Fitting the age-gender-pathology joint response surface of each key electrocardiogram parameter in the population-level electrocardiogram feature distribution model using Gaussian process regression to generate an initial individualized baseline function; Further using K-means clustering to identify the three closest physiological mode clusters to the target patient, and taking the weighted average of the corresponding baseline as the virtual individualized baseline; Comparing the real-time key electrocardiogram feature parameters with the corresponding virtual individualized baseline to obtain parameter offset values and calculate a confidence score, and triggering an artificial review process when the confidence score is below a set lower limit.

5. The method of claim 1, wherein, The step of determining the abnormality level includes: According to the type of abnormality of the real-time electrocardiogram data, the current medication regimen and historical diagnosis and treatment records of the patient are retrieved; Based on the embedded drug-electrocardiogram association rule base, the causal possibility between abnormal electrocardiogram features and medication is analyzed to generate a medication risk index; According to the patient's recent operation history, complications and treatment stage, a clinical weight factor coefficient is generated; The heart rate variability and autonomic nervous balance index are introduced as physiological elasticity indicators to modify the risk total score calculation formula; the final risk total score = parameter offset value × clinical weight factor coefficient × (1 + medication risk index / n) × (1 - heart rate variability improvement coefficient); According to the final risk total score, the abnormality level is divided.

6. The method of claim 1, wherein, The remote electrocardiogram telemetry method further includes: Establishing a two-way encrypted patient-doctor real-time communication mechanism to support text, voice, picture upload and video consultation access; Allowing the patient to initiate an emergency consultation request through the patient terminal with one key, and the system automatically attaching the latest m-minute electrocardiogram segment and risk assessment summary; Supporting the doctor to initiate a structured follow-up reminder through the doctor platform; After pushing the risk assessment report, the system automatically generates a structured follow-up task sheet, and sets the follow-up priority, response time limit and responsible doctor role according to the abnormality level.

7. The method of claim 1, wherein, The remote electrocardiogram telemetry method further includes: In the case of detecting two consecutive high-risk abnormal events and the patient has no active response, starting a non-invasive physiological intervention plan: the edge end controls the monitoring device to output a specific frequency of transcutaneous vagus nerve stimulation pulse signal to adjust the autonomic nervous tension and suppress sympathetic storm; synchronously pushing voice soothing instructions and breathing guidance animation to the patient terminal to induce vagus nerve activation; During the intervention, the heart rate variability index is continuously monitored, and if the LF / HF ratio decreases by ≥ a first threshold value or the RMSSD increases by ≥ a second threshold value, the intervention is determined to be effective, the stimulation is suspended, and the efficacy data is recorded; If it is invalid, automatically upgrade to a remote emergency dispatch process, link the 120 emergency center and send the patient's location, vital signs snapshot and electronic health record summary.

8. A remote electrocardiographic telemetry system characterized by, Performing the remote electrocardiogram telemetry method as claimed in any one of claims 1-7, including: The cloud (4) is configured to assign a corresponding electrocardiogram monitoring device (1) and initial monitoring parameters to a patient according to patient information, and to determine whether the predicted risk value is greater than a dynamically adjusted risk threshold. If yes, the matching predicted disease type is retrieved from a multi-dimensional disease diagnosis knowledge graph according to the predicted risk value and its change rate, and the differential risk prompt content is generated by combining the comorbidity relationship network in the patient's electronic medical record and is pushed. If no, the cloud (4) analyzes the real-time electrocardiogram data to determine whether it is abnormal. If yes, the abnormal level is determined based on the spatial distribution characteristics and duration of the abnormal waveform, and the personalized risk assessment report is generated by combining the treatment response history, and the risk assessment report is pushed to trigger the corresponding level of clinical response protocol. The edge (3) is configured to receive real-time electrocardiogram data sent by the electrocardiogram monitoring device (1), and to obtain environmental data of the location where the patient is located based on the location information of the patient, and to collect multi-modal physiological data through the wearable sensor (2). The time-space context perception graph is constructed, the real-time electrocardiogram data, environmental data and physiological data are aligned on the time axis, and the situation-enhanced electrocardiogram signal representation is generated through attention mechanism weighted fusion, and the situation-enhanced electrocardiogram signal representation is input into the comprehensive risk prediction model with online incremental learning ability to obtain a dynamically updated predicted risk value, and uploaded to the cloud (4). The patient end (5) is configured to receive the differential risk prompt content and the risk assessment report. The doctor platform (6) is configured to receive the differential risk prompt content and the risk assessment report.

9. A terminal, characterized by comprising: It comprises: a memory storing a remote electrocardiogram telemetry program; a processor configured to execute the program stored on the memory to implement the steps of the remote electrocardiogram telemetry method of any one of claims 1-7.

10. A computer-readable storage medium, characterized in that, a computer program capable of being loaded and executed by the processor to implement the remote electrocardiogram telemetry method of any one of claims 1-7.

Images