Early intelligent early warning method for hospital adverse cardiovascular event chain of acute myocardial infarction patient

Abstract

The invention relates to the technical field of medical and industrial cross, and discloses an intelligent early warning method for an in-hospital adverse cardiovascular event chain of an acute myocardial infarction patient, and the method comprises the following steps: S1, the patient stores an installation file in a mobile phone in a WeChat transmission manner, searches a downloaded APP installation file in mobile phone file management, operates an installation program, and sends the installation file to a server; the installation process of the APP is completed according to the prompt, and in the installation process, if a request of mobile phone operation permission is involved, 'permission 'is clicked. Important physiological indexes such as the heart rate, the electrocardiogram, the blood pressure and the oxyhemoglobin saturation of a patient with acute myocardial infarction are monitored and analyzed in real time through wearable equipment based on the Internet of Things technology, the early warning capacity of adverse cardiovascular events can be remarkably improved, and an intelligent platform can conduct early recognition on a high-risk patient, so that the early warning efficiency is improved. Alarm information is sent to an on-duty doctor in time, the doctor is helped to quickly check and intervene, and the occurrence rate of clinical adverse events caused by early warning lag is reduced.

Description

Technical Field

[0001] The present invention relates to the field of medical and engineering intersection technology, and specifically to an early intelligent warning method for an in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction. Background Art

[0002] Patients with acute myocardial infarction are at high risk of adverse clinical events (such as tachyarrhythmias, bradyarrhythmias, cardiac arrest, respiratory failure, and hypotension) during the perioperative period and hospitalization. At present, cardiac intensive care units mainly rely on continuous ECG monitors and a 24-hour medical and nursing duty system to monitor the patient's cardiac status. However, traditional monitoring methods have significant shortcomings, mainly manifested in the low early warning ability of adverse events, resulting in relatively delayed medical intervention. In addition, due to the limitations of monitoring instruments and manual monitoring, it is difficult to identify and deal with adverse events of high-risk patients in a timely manner in clinical practice, affecting the patient's prognosis.

[0003] Existing early warning systems usually rely on the assistance of patients themselves or people around them to trigger early warnings. The early warning efficiency is low and the speed is slow. In addition, the existing early warning systems lack early warning capabilities and find it difficult to integrate multimodal data for personalized early warnings. At the same time, they cannot provide data support, which limits long-term scientific research and data analysis.

[0004] Therefore, those skilled in the art provide an intelligent early warning method for in-hospital adverse cardiovascular event chains in patients with acute myocardial infarction to solve the problems raised in the above background technology. Summary of the invention

[0005] In view of the deficiencies in the prior art, the present invention provides an intelligent early warning method for an in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction to solve the problems raised in the above-mentioned background technology.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: an early intelligent warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction, comprising the following steps:

[0007] S1. The patient saves the installation file to the mobile phone through WeChat transmission, then finds the downloaded APP installation file in the mobile phone file management, runs the installation program, and completes the APP installation process according to the prompts. During the installation process, if it involves a request for mobile phone operation permissions, click "Allow";

[0008] S2. Turn on the Bluetooth function of the mobile phone and set the smart watch to Bluetooth pairing status. The watch name or QR code will be displayed on the watch screen. At this time, open the APP on the mobile phone, the APP will automatically connect to the watch, and a prompt box will pop up. Click "Yes" to confirm the connection. At the same time, click "Yes" on the watch to complete the pairing.

[0009] S3. After pairing is completed, the patient starts real-time data synchronization through the APP. The APP will automatically connect to the watch. After the connection status is displayed as "connected", the next step is performed. The "device code" displayed in the APP is the identification code of the watch. This code is associated with the basic attributes of the patient in the early warning platform. The APP will synchronize the watch data to the system database at the set frequency. At this time, click the "Synchronize Data" button, and the data on the watch will be obtained and uploaded to the intelligent early warning platform;

[0010] S4. Medical staff log in to the intelligent early warning platform through the system account, enter the patient management interface, click the "+Case" button in the "Patient Center" to create a new patient file, fill in the patient's relevant information in the pop-up window, and click the "Confirm" button to complete the patient creation. If you need to modify the patient information, click the "Edit" button on the right side of the patient list, modify the information, and click the "Confirm" button to save;

[0011] S5. Medical staff can view detailed information of patients on the platform, and early warning information can be viewed in real time through the APP or system notification bar. In the system settings, medical staff can adjust the threshold of early warning rules;

[0012] S6. If the system detects abnormal health data of the patient, the APP will send an early warning notification. Medical staff can click the notification icon in the APP to view detailed early warning information. To further confirm the patient's condition, medical staff can directly call the patient or his family through the "Call" function in the APP;

[0013] S7. When medical staff confirm that the warning is a false alarm or has been processed, the warning status can be eliminated through the system function;

[0014] S8. Wearable devices can continuously collect health data after the patient is discharged from the hospital, and upload the collected data to the intelligent early warning platform to form a complete health data chain, and the platform aggregates and analyzes the data.

[0015] Preferably, the early warning mechanism in step S5 is used to analyze and evaluate the patient's health data in real time for early warning, and its algorithm includes heart rate variability, heart rate threshold early warning formula, multimodal data fusion and weighted average early warning formula and machine learning algorithm:

[0016] The heart rate variability is a method of evaluating the cardiac autonomic nervous function by measuring the RR interval. The time domain indicators include:

[0017] SDNN(Standard Deviation):

[0018]

[0019] Among them, RR i is the i-th RR interval, is the average value of the RR intervals, and N is the total number of RR intervals.

[0020] Preferably, the heart rate threshold warning formula is:

[0021] Usually, abnormal heart rate is divided into tachycardia and bradycardia, and the calculation can be set as:

[0022] Tachycardia:

[0023] HR>HR max ,

[0024] Among them, HR max is the set maximum heart rate threshold;

[0025] Bradycardia:

[0026] HR <HR min ,

[0027] Among them, HR min is the minimum heart rate threshold that is set.

[0028] Preferably, the multimodal data fusion and weighted average warning formula is:

[0029] The algorithm takes into account multiple physiological parameters such as heart rate, blood oxygen saturation, blood pressure, etc., performs weighted average calculation, and forms a comprehensive warning index:

[0030] Comprehensive Early Warning Index (CP I):

[0031]

[0032] Among them, w1, w2, w3 are the weights of different parameters, HR, SpO2, BP represent the real-time values ​​of heart rate, blood oxygen saturation and blood pressure respectively, ref represents the reference value, max and min represent the corresponding maximum and minimum thresholds.

[0033] Preferably, the machine learning algorithm:

[0034] Identify abnormal patterns by training a support vector machine or neural network model:

[0035] Support Vector Machine Classifier:

[0036]

[0037] Among them, α i is the Lagrange multiplier, y i is the sample label, K(x i ,x) is the kernel function, b is the bias;

[0038] Neural network output layer formula:

[0039] y=softmax(Wx+b),

[0040] Among them, W is the weight matrix, x is the input, b is the bias, and the b function is used to calculate the classification probability.

[0041] Preferably, in step S6, the algorithm for sending the warning notification by the APP includes the risk assessment index and time series anomaly prediction:

[0042] The risk assessment index:

[0043] After detecting an abnormal parameter, the system uses a comprehensive risk assessment model to further determine whether an early warning is needed. The algorithm formula of the risk assessment index is:

[0044]

[0045] Among them, R i (t) is the risk value of the ith physiological parameter, α i is the weight of the corresponding parameter, A(τ) is the anomaly measure at time τ, T is the length of the evaluation window, β is the weight of the historical anomaly, and an early warning is triggered when the risk assessment index exceeds a preset threshold;

[0046] The time series anomaly prediction:

[0047] The system uses a machine learning model, long short-term memory network, to predict future trends in physiological parameters. If the prediction results show that a high-risk event is about to occur, the system will issue an early warning:

[0048] Long short-term memory network prediction formula:

[0049]

[0050] in, is the predicted physiological parameter value after k time steps in the future, X(t) is the input data at time t, and f LSTM is a trained long short-term memory network model. If If the warning threshold is exceeded, a warning notification will be issued in advance.

[0051] Preferably, in step S7, when the medical staff confirms that the warning is a false alarm or has been processed, the system verifies the warning status and decides whether to eliminate the warning through an algorithm, which involves retrospective analysis and pattern recognition of historical data and credibility assessment:

[0052] The historical data retrospective analysis:

[0053] To confirm whether the warning is a false alarm, the system reviews recent data and uses a sliding window for retrospective analysis:

[0054] Sliding window anomaly integral:

[0055]

[0056] Among them, W is the window size, A(ti) is the anomaly measurement at time ti, and by comparing SWAS(t) with the threshold, the system determines whether the warning has continuous anomalies. If SWAS(t) is less than the set false alarm threshold, it is a false alarm;

[0057] The pattern recognition and warning credibility assessment:

[0058] The system uses a pattern recognition algorithm to compare the physiological data pattern with known abnormal patterns. If the physiological data pattern matches the known normal pattern, the system determines it as a false alarm:

[0059] Warning credibility formula:

[0060]

[0061] Among them, λ j is the weight of the jth pattern recognition algorithm, S j (t) is the similarity score between the physiological data pattern and the jth normal pattern. The similarity score is calculated based on the algorithm. If AC exceeds a certain threshold, the warning credibility is considered low and it is treated as a false alarm.

[0062] Preferably, the algorithm in step S8 is used to continuously collect patient health data, upload data, and summarize and analyze it to form a complete health data chain. The algorithm formula is as follows:

[0063] Data smoothing formula:

[0064]

[0065] Among them, X(t) is the original data at time t, k is the window size, and X smoothed (t) is the smoothed data;

[0066] Outlier filtering formula:

[0067]

[0068] Among them, μ X and σ X is the mean and standard deviation of the data, Z threshold It is the Z-score threshold, which is used to determine whether the data is an outlier.

[0069] Preferably, in step S8, the system aggregates the patient's data and mines potential health patterns and risk factors through big data analysis technology;

[0070] Cluster analysis: The system uses the K-means algorithm to perform cluster analysis on large amounts of patient data to identify groups of patients with similar health characteristics:

[0071]

[0072] Among them, J is the objective function, C i is the i-th cluster, μ i is the center point of the cluster, and x is the sample point in the cluster;

[0073] Association rule mining: The system uses the April algorithm to mine association rules in health data:

[0074]

[0075] Among them, A and B are different sets of physiological parameters, and |D| is the total number of samples in the dataset.

[0076] Preferably, in step S8, based on the continuously collected data, the system generates a long-term health assessment report for the patient to provide feedback and suggestions, and the long-term health score formula is:

[0077]

[0078] Where T is the length of the evaluation period, and F(t) is the comprehensive health status index at each moment;

[0079] The system generates personalized health recommendations based on long-term health scores to provide real-time feedback to patients through the APP.

[0080] The present invention provides an intelligent early warning method for in-hospital adverse cardiovascular event chains in patients with acute myocardial infarction.

[0081] It has the following beneficial effects:

[0082] 1. The present invention uses wearable devices based on Internet of Things technology to monitor and analyze important physiological indicators such as heart rate, electrocardiogram, blood pressure, and blood oxygen saturation of patients with acute myocardial infarction in real time, which can significantly improve the early warning ability of adverse cardiovascular events. The intelligent platform can identify high-risk patients at an early stage and send alarm messages to doctors on duty in a timely manner to help doctors quickly check and intervene, thereby reducing the incidence of clinical adverse events caused by delayed warnings.

[0083] 2. The present invention constructs an individualized myocardial infarction prognosis assessment model based on multimodal data, which can combine the patient's biochemical indicators, examination data, and in-hospital and out-of-hospital visit data to predict the patient's medium- and long-term prognosis. Its personalized assessment helps guide clinical diagnosis and treatment and improve the pertinence and effectiveness of treatment.

[0084] 3. The present invention collects and stores multimodal data of patients through an intelligent platform to form a disease-specific database to support scientific research exploration of clinical cohorts. In addition, wearable devices can continuously collect data outside the hospital to form continuous health tracking records, support closed-loop management of scientific research, and improve scientific research efficiency. The intelligent early warning method is aimed at tertiary A-level hospitals, cardiac specialty hospitals, secondary hospitals and community medical institutions, and can provide a comprehensive cardiac monitoring early warning solution. In addition, cooperation with insurance companies to promote wearable devices and monitoring services can help insurance companies reduce the risk of compensation caused by myocardial infarction, and has high market potential and commercial value. BRIEF DESCRIPTION OF THE DRAWINGS

[0085] Figure 1 is a flow chart of the present invention;

[0086] Figure 2 It is a schematic diagram of system deployment of the present invention. DETAILED DESCRIPTION

[0087] In order to make the technical personnel in the technical field understand the scheme of the present invention, the technical scheme in the embodiment of the present invention will be clearly and completely described below in combination with the drawings in the embodiment of the present invention. Obviously, the described embodiment is a partial embodiment of the present invention, not a complete embodiment. Based on the embodiment of the present invention, other embodiments obtained by ordinary technicians in the field without creative work should fall within the scope of protection of the present invention.

[0088] The present invention is described in detail below in conjunction with the accompanying drawings:

[0089] Example:

[0090] Please see attached Figure 1 and attached Figure 2 The embodiment of the present invention provides an intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction, comprising the following steps:

[0091] S1. The patient saves the installation file to the mobile phone through WeChat transmission, then finds the downloaded APP installation file in the mobile phone file management, runs the installation program, and completes the APP installation process according to the prompts. During the installation process, if it involves a request for mobile phone operation permissions, click "Allow";

[0092] S2. Turn on the Bluetooth function of the mobile phone and set the smart watch to Bluetooth pairing status. The watch name or QR code will be displayed on the watch screen. At this time, open the APP on the mobile phone, the APP will automatically connect to the watch, and a prompt box will pop up. Click "Yes" to confirm the connection. At the same time, click "Yes" on the watch to complete the pairing.

[0093] S3. After pairing is completed, the patient starts real-time data synchronization through the APP. The APP will automatically connect to the watch. After the connection status is displayed as "connected", the next step is performed. The "device code" displayed in the APP is the identification code of the watch. This code is associated with the basic attributes of the patient in the early warning platform. The APP will synchronize the watch data to the system database at the set frequency. At this time, click the "Synchronize Data" button, and the data on the watch will be obtained and uploaded to the intelligent early warning platform;

[0094] S4. Medical staff log in to the intelligent early warning platform through the system account, enter the patient management interface, click the "+Case" button in the "Patient Center" to create a new patient file, fill in the patient's relevant information in the pop-up window, and click the "Confirm" button to complete the patient creation. If you need to modify the patient information, click the "Edit" button on the right side of the patient list, modify the information, and click the "Confirm" button to save;

[0095] S5. Medical staff can view detailed information of patients on the platform, and early warning information can be viewed in real time through the APP or system notification bar. In the system settings, medical staff can adjust the threshold of early warning rules;

[0096] S6. If the system detects abnormal health data of the patient, the APP will send an early warning notification. Medical staff can click the notification icon in the APP to view detailed early warning information. To further confirm the patient's condition, medical staff can directly call the patient or his family through the "Call" function in the APP;

[0097] S7. When medical staff confirm that the warning is a false alarm or has been processed, the warning status can be eliminated through the system function;

[0098] S8. Wearable devices can continuously collect health data after the patient is discharged from the hospital, and upload the collected data to the intelligent early warning platform to form a complete health data chain, and the platform aggregates and analyzes the data.

[0099] Benefits of S1 APP installation and permission configuration Reduce the complexity of the installation process, making it easier for patients to start using the early warning system, and handle mobile phone permission requests during the installation process to ensure that the APP can operate normally and access necessary mobile phone functions;

[0100] The benefits of pairing the S2 smartwatch with the mobile phone APP are that it reduces the number of user operation steps and improves the user experience. After the watch and the mobile phone are successfully paired, data synchronization can start immediately to ensure that the patient's health data can be transmitted to the early warning platform in real time.

[0101] Benefits of S3 data synchronization and uploading: Ensure that key health indicators such as heart rate and blood pressure are continuously monitored, which helps to detect potential risks in a timely manner. The data is automatically uploaded to the intelligent early warning platform to ensure data integrity and security, and reduce errors that may be caused by manual operations;

[0102] Benefits of S4 creating and managing patient records Real-time updating of patient information, providing more personalized medical services, and contributing to more accurate health management;

[0103] The benefits of S5 viewing and adjusting warning information are to facilitate quick response and processing, provide more accurate monitoring and warning, and avoid false alarms or missed alarms;

[0104] Benefits of S6 abnormal health data warning and intervention: medical staff can quickly understand the patient's status and take intervention measures, ensuring timely communication and intervention to prevent further development of adverse events;

[0105] Benefits of S7 false alarm handling and warning status elimination Medical staff can quickly eliminate the warning status through system functions, reducing the interference of false alarms on medical staff and patients, which is conducive to the continuous learning and optimization of the system and improving the accuracy of future warnings;

[0106] The benefits of S8 post-discharge data collection and analysis ensure that the patient's long-term health status is monitored, help identify long-term health trends, and provide data support for future medical decisions and research.

[0107] The early warning mechanism in step S5 is used to analyze and evaluate the patient's health data in real time for early warning. Its algorithms include heart rate variability, heart rate threshold early warning formula, multimodal data fusion and weighted average early warning formula, and machine learning algorithm:

[0108] Heart rate variability is a method of evaluating cardiac autonomic nervous function by measuring RR intervals. Time domain indicators include:

[0109] SDNN(Standard Deviation):

[0110]

[0111] Among them, RR i is the i-th RR interval, is the average value of the RR intervals, and N is the total number of RR intervals.

[0112] The heart rate variability algorithm formula works by analyzing the variability of the RR interval. SDNN reflects the functional state of the cardiac autonomic nervous system. A lower SDNN means a decrease in the cardiac autonomic regulation function, indicating that the patient is at a higher risk of cardiovascular events. If the heart rate variability is abnormal, the system identifies potential adverse cardiovascular events and triggers an early warning in time.

[0113] Heart rate threshold warning formula:

[0114] Usually, abnormal heart rate is divided into tachycardia and bradycardia, and the calculation can be set as:

[0115] Tachycardia:

[0116] HR>HR max ,

[0117] Among them, HR max is the set maximum heart rate threshold;

[0118] Bradycardia:

[0119] HR <HR min ,

[0120] Among them, HR min is the minimum heart rate threshold that is set.

[0121] The heart rate threshold warning algorithm formula is used to monitor the patient's heart rate in real time to determine whether it exceeds the preset safety range. Through the set threshold, the system can issue a warning in time when the patient's heart rate is too fast or too slow, and then through simple threshold judgment, it can quickly identify obvious heart rate abnormalities for rapid response to acute situations.

[0122] Multimodal data fusion and weighted average warning formula:

[0123] The algorithm takes into account multiple physiological parameters such as heart rate, blood oxygen saturation, blood pressure, etc., performs weighted average calculation, and forms a comprehensive warning index:

[0124] Comprehensive Early Warning Index (CP I):

[0125]

[0126] Among them, w1, w2, w3 are the weights of different parameters, HR, SpO2, BP represent the real-time values ​​of heart rate, blood oxygen saturation and blood pressure respectively, ref represents the reference value, max and min represent the corresponding maximum and minimum thresholds.

[0127] The multimodal data fusion and weighted average early warning algorithm formula can comprehensively evaluate the patient's overall health status by fusing multiple physiological parameters, and the weighted average formula can be adjusted according to the importance of different parameters, so that the early warning system can more accurately reflect the patient's true health risks.

[0128] Machine Learning Algorithms:

[0129] Identify abnormal patterns by training a support vector machine or neural network model:

[0130] Support Vector Machine Classifier:

[0131]

[0132] Among them, α i is the Lagrange multiplier, y i is the sample label, K(x i ,x) is the kernel function, b is the bias;

[0133] Neural network output layer formula:

[0134] y=softmax(Wx+b),

[0135] Among them, W is the weight matrix, x is the input, b is the bias, and the b function is used to calculate the classification probability.

[0136] The machine learning algorithm formula can capture complex data patterns that are difficult for humans to identify, predict potential cardiovascular event risks in advance, and provide a basis for early warning. The machine learning model can be continuously optimized as data increases and time passes, gradually improving the accuracy and sensitivity of the early warning.

[0137] In step S6, the algorithm used by the APP to send warning notifications includes risk assessment index and time series anomaly prediction:

[0138] Risk Assessment Index:

[0139] After detecting an abnormal parameter, the system uses a comprehensive risk assessment model to further determine whether an early warning is needed. The algorithm formula of the risk assessment index is:

[0140]

[0141] Among them, R i (t) is the risk value of the ith physiological parameter, α i is the weight of the corresponding parameter, A(τ) is the anomaly measure at time τ, T is the length of the evaluation window, β is the weight of the historical anomaly, and an early warning is triggered when the risk assessment index exceeds a preset threshold;

[0142] Time Series Anomaly Prediction:

[0143] The system uses a machine learning model, long short-term memory network, to predict future trends in physiological parameters. If the prediction results show that a high-risk event is about to occur, the system will issue an early warning:

[0144] Long short-term memory network prediction formula:

[0145]

[0146] in, is the predicted physiological parameter value after k time steps in the future, X(t) is the input data at time t, and fLSTM is a trained long short-term memory network model. If If the warning threshold is exceeded, a warning notification will be issued in advance.

[0147] The risk assessment index algorithm formula calculates the real-time risk values ​​of multiple physiological parameters by weighted calculation, and combines historical abnormalities to generate a comprehensive risk assessment index. The index reflects the patient's overall health risk status. When the risk assessment index RAI(t) exceeds the preset threshold, the system will trigger an early warning notification to remind medical staff to intervene in time, so that the early warning system can identify risks and take into account the impact of historical data to provide more accurate and reliable early warnings.

[0148] The time series anomaly prediction formula uses the LSTM model to predict future physiological parameters. If the prediction results show that the physiological parameters at a certain moment in the future exceed the warning threshold, the system will issue a warning notification in advance, and medical staff can take preventive measures before the patient has obvious symptoms to prevent adverse events. The LSTM network is good at processing and predicting complex patterns and trends in time series data, which is necessary for real-time monitoring of patients with acute myocardial infarction.

[0149] In step S7, when the medical staff confirms that the warning is a false alarm or has been processed, the system will use an algorithm to verify the warning status and decide whether to eliminate the warning. The algorithm involves retrospective analysis and pattern recognition of historical data and credibility assessment:

[0150] Historical data retrospective analysis:

[0151] To confirm whether the warning is a false alarm, the system reviews recent data and uses a sliding window for retrospective analysis:

[0152] Sliding window anomaly integral:

[0153]

[0154] Among them, W is the window size, A(ti) is the anomaly measurement at time ti, and by comparing SWAS(t) with the threshold, the system determines whether the warning has continuous anomalies. If SWAS(t) is less than the set false alarm threshold, it is a false alarm;

[0155] Pattern recognition and warning credibility assessment:

[0156] The system uses a pattern recognition algorithm to compare the physiological data pattern with known abnormal patterns. If the physiological data pattern matches the known normal pattern, the system determines it as a false alarm:

[0157] Warning credibility formula:

[0158]

[0159] Among them, λ j is the weight of the jth pattern recognition algorithm, S j (t) is the similarity score between the physiological data pattern and the jth normal pattern. The similarity score is calculated based on the algorithm. If AC exceeds a certain threshold, the warning credibility is considered low and it is treated as a false alarm.

[0160] The role of historical data retrospective analysis is to evaluate the duration and severity of abnormal situations through the sliding window abnormal integration. If SWAS(t) is lower than the set false alarm threshold, it indicates that the abnormality is short-lived or sporadic, and the warning is judged as a false alarm, which helps to distinguish between real emergencies and sporadic physiological abnormalities and reduce unnecessary interventions. The sliding window analysis provides a continuous perspective to evaluate abnormal events, which can more accurately judge the effectiveness of the warning, reduce false alarms, and improve the accuracy of the warning system.

[0161] The system uses pattern recognition algorithms to compare the matching degree between physiological data and known normal and abnormal patterns. If the similarity score S between the data pattern and the normal pattern is j (t) is higher, the warning credibility AC is lower, prompting the system that the warning is a false alarm, effectively identifying false alarms caused by accidental factors, and by calculating the warning credibility, the system can quantify the credibility of the warning. If AC exceeds the set threshold, the system can automatically determine whether the warning is a false alarm and decide whether to eliminate the warning status, thereby improving the intelligence level of the warning system and reducing the necessity of human intervention.

[0162] The algorithm in step S8 is used to continuously collect patient health data, upload data, and summarize and analyze it to form a complete health data chain. The algorithm formula is as follows:

[0163] Data smoothing formula:

[0164]

[0165] Among them, X(t) is the original data at time t, k is the window size, and X smoothed (t) is the smoothed data;

[0166] Outlier filtering formula:

[0167]

[0168] Among them, μ x and σ x is the mean and standard deviation of the data, Z threshold It is the Z-score threshold, which is used to determine whether the data is an outlier.

[0169] The function of the data smoothing formula is to smooth the time series data by moving average, which can effectively reduce the short-term fluctuations and noise in the measured data. This is crucial for accurately monitoring the health status of patients, because noise can cause false alarms or mask real abnormal signals. The smoothed data is more stable and can provide clearer trends and patterns, which helps the system to more accurately analyze the health status of patients and improve the reliability of the early warning system.

[0170] The outlier filtering formula is used to detect and filter outliers in the data to ensure that the system processes accurate and reliable data. Outliers are caused by equipment failure, patient movement or other external interference. Filtering outliers can prevent misleading the overall analysis and early warning results. By excluding outliers, the system can better maintain the consistency and integrity of the data, which is especially important for long-term health monitoring and trend analysis, and can ensure that the generated health data chain accurately reflects the patient's true health status.

[0171] In step S8, the system aggregates the patient’s data and uses big data analysis techniques to mine potential health patterns and risk factors;

[0172] Cluster analysis: The system uses the K-means algorithm to perform cluster analysis on large amounts of patient data to identify groups of patients with similar health characteristics:

[0173]

[0174] Among them, J is the objective function, C i is the i-th cluster, μ i is the center point of the cluster, and x is the sample point in the cluster;

[0175] Association rule mining: The system uses the April algorithm to mine association rules in health data:

[0176]

[0177] Among them, A and B are different sets of physiological parameters, and |D| is the total number of samples in the dataset.

[0178] K-means cluster analysis can help identify patient groups with similar health characteristics by grouping patient data, which helps to detect the commonalities between certain health characteristics and facilitates the system to further analyze the risk characteristics of different patient groups. Through cluster analysis, the system customizes early warning rules and health management strategies for different patient groups, thereby improving the accuracy and effectiveness of the overall early warning. After identifying similar patient groups, medical institutions can allocate resources more reasonably, thereby improving the overall effectiveness of the early warning system.

[0179] April association rule mining can reveal the potential relationship between certain combinations of physiological parameters and the risk of cardiovascular events by mining the association rules between physiological parameters, which helps to identify key risk factors and provide a more scientific basis for the early warning system. By analyzing the association between physiological parameters, the system can set early warning conditions for specific risk combinations, thereby improving the accuracy of the early warning. At the same time, the results of association rule mining can help medical staff understand the complex relationships in patients' health data and provide data support for personalized health management and clinical decision-making.

[0180] In step S8, based on the continuously collected data, the system generates a long-term health assessment report for the patient to provide feedback and suggestions. The long-term health score formula is:

[0181]

[0182] Where T is the length of the evaluation period, and F(t) is the comprehensive health status index at each moment;

[0183] The system generates personalized health recommendations based on long-term health scores to provide real-time feedback to patients through the APP.

[0184] The function of the long-term health score formula is to calculate the long-term health score within the evaluation period, so that the system can quantify the patient's overall health status during that period, and the score can reflect the patient's average health level. If the long-term health score drops significantly, the system can identify potential health problems or disease progression in advance, and provide medical staff with early intervention signals. Based on the long-term health score, the system can generate personalized health suggestions for each patient, including lifestyle adjustments, drug treatment recommendations, and reminders for regular check-ups, etc., to help patients maintain or improve their health status. At the same time, by providing real-time feedback to patients, patients can promptly understand changes in their health status and take corresponding actions based on the recommendations, thereby enhancing their self-health management capabilities and reducing the risk of disease progression. In addition, the long-term health score provides the system with a macro health monitoring perspective, enabling the system to respond to short-term health fluctuations and identify long-term health changes, thereby improving the system's overall early warning and response capabilities, and helping to respond to the long-term health risks of patients with acute myocardial infarction in a timely manner.

[0185] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction, characterized in that: The steps include: S1. The patient saves the installation file to the mobile phone through WeChat transmission, then finds the downloaded APP installation file in the mobile phone file management, runs the installation program, and completes the APP installation process according to the prompts. During the installation process, if it involves a request for mobile phone operation permissions, click "Allow"; S2. Turn on the Bluetooth function of the mobile phone and set the smart watch to Bluetooth pairing status. The watch name or QR code will be displayed on the watch screen. At this time, open the APP on the mobile phone, the APP will automatically connect to the watch, and a prompt box will pop up. Click "Yes" to confirm the connection. At the same time, click "Yes" on the watch to complete the pairing. S3. After pairing is completed, the patient starts real-time data synchronization through the APP. The APP will automatically connect to the watch. After the connection status is displayed as "connected", the next step is performed. The "device code" displayed in the APP is the identification code of the watch. This code is associated with the basic attributes of the patient in the early warning platform. The APP will synchronize the watch data to the system database at the set frequency. At this time, click the "Synchronize Data" button, and the data on the watch will be obtained and uploaded to the intelligent early warning platform; S4. Medical staff log in to the intelligent early warning platform through the system account, enter the patient management interface, click the "+Case" button in the "Patient Center" to create a new patient file, fill in the patient's relevant information in the pop-up window, and click the "OK" button to complete the patient creation. If you need to modify the patient information, click the "Edit" button on the right side of the patient list, modify the information, and click the "OK" button to save; S5. Medical staff can view detailed information of patients on the platform, and early warning information can be viewed in real time through the APP or system notification bar. In the system settings, medical staff can adjust the threshold of early warning rules; S6. If the system detects abnormal health data of the patient, the APP will send an early warning notification. Medical staff can click the notification icon in the APP to view detailed early warning information. To further confirm the patient's condition, medical staff can directly call the patient or his family through the "Call" function in the APP; S7. When medical staff confirm that the warning is a false alarm or has been processed, the warning status can be eliminated through the system function; S8. Wearable devices can continuously collect health data after the patient is discharged from the hospital, and upload the collected data to the intelligent early warning platform to form a complete health data chain, and the platform aggregates and analyzes the data.

2. According to claim 1, an intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction is characterized by: The early warning mechanism in step S5 is used to analyze and evaluate the patient's health data in real time for early warning, and its algorithm includes heart rate variability, heart rate threshold early warning formula, multimodal data fusion and weighted average early warning formula and machine learning algorithm: The heart rate variability is a method of evaluating the cardiac autonomic nervous function by measuring the RR interval. The time domain indicators include: SDNN(Standard Deviation): Among them, RR i is the i-th RR interval, is the average value of the RR intervals, and N is the total number of RR intervals.

3. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 2 is characterized in that: The heart rate threshold warning formula: Usually, abnormal heart rate is divided into tachycardia and bradycardia, and the calculation can be set as: Tachycardia: HR>HR max , Among them, HR max is the set maximum heart rate threshold; Bradycardia: HR<HR min , Among them, HR min is the minimum heart rate threshold that is set.

4. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 2 is characterized in that: The multimodal data fusion and weighted average warning formula: The algorithm takes into account multiple physiological parameters such as heart rate, blood oxygen saturation, blood pressure, etc., performs weighted average calculation, and forms a comprehensive warning index: Comprehensive Early Warning Index (CPI): Among them, w1, w2, w3 are the weights of different parameters, HR, SpO2, BP represent the real-time values ​​of heart rate, blood oxygen saturation and blood pressure respectively, ref represents the reference value, max and min represent the corresponding maximum and minimum thresholds.

5. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 2, characterized in that: The machine learning algorithm: Identify abnormal patterns by training a support vector machine or neural network model: Support Vector Machine Classifier: Among them, α i is the Lagrange multiplier, y i is the sample label, K(x i ,x) is the kernel function, b is the bias; Neural network output layer formula: y=softmax(Wx+b), Among them, W is the weight matrix, x is the input, b is the bias, and the b function is used to calculate the classification probability.

6. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 1, characterized in that: In step S6, the algorithm for sending the warning notification by the APP includes the risk assessment index and time series anomaly prediction: The risk assessment index is: After detecting an abnormal parameter, the system uses a comprehensive risk assessment model to further determine whether an early warning is needed. The algorithm formula of the risk assessment index is: Among them, R i (t) is the risk value of the ith physiological parameter, α i is the weight of the corresponding parameter, A(τ) is the anomaly measure at time τ, T is the length of the evaluation window, β is the weight of the historical anomaly, and an early warning is triggered when the risk assessment index exceeds a preset threshold; The time series anomaly prediction: The system uses a machine learning model, long short-term memory network, to predict future trends in physiological parameters. If the prediction results show that a high-risk event is about to occur, the system will issue an early warning: Long short-term memory network prediction formula: in, is the predicted physiological parameter value after k time steps in the future, X(t) is the input data at time t, and f LSTM is a trained long short-term memory network model. If If the warning threshold is exceeded, a warning notification will be issued in advance.

7. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 1, characterized in that: In step S7, when the medical staff confirms that the warning is a false alarm or has been processed, the system will use an algorithm to verify the warning status and decide whether to eliminate the warning. The algorithm involves retrospective analysis and pattern recognition of historical data and credibility assessment: The historical data retrospective analysis: To confirm whether the warning is a false alarm, the system reviews recent data and uses a sliding window for retrospective analysis: Sliding window anomaly integral: Among them, W is the window size, A(ti) is the anomaly measurement at time ti, and by comparing SWAS(t) with the threshold, the system determines whether the warning has continuous anomalies. If SWAS(t) is less than the set false alarm threshold, it is a false alarm; The pattern recognition and warning credibility assessment: The system uses a pattern recognition algorithm to compare the physiological data pattern with known abnormal patterns. If the physiological data pattern matches the known normal pattern, the system determines it as a false alarm: Warning credibility formula: Among them, λ j is the weight of the jth pattern recognition algorithm, S j (t) is the similarity score between the physiological data pattern and the jth normal pattern. The similarity score is calculated based on the algorithm. If AC exceeds a certain threshold, the warning credibility is considered low and it is treated as a false alarm.

8. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 1, characterized in that: The algorithm in step S8 is used to continuously collect patient health data, upload data, and summarize and analyze it to form a complete health data chain. The algorithm formula is as follows: Data smoothing formula: Among them, X(t) is the original data at time t, k is the window size, and X smoothed (t) is the smoothed data; Outlier filtering formula: Among them, μ X and σ X is the mean and standard deviation of the data, Z threshold It is the Z-score threshold, which is used to determine whether the data is an outlier.

9. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 1, characterized in that: In step S8, the system aggregates the patient's data and uses big data analysis technology to mine potential health patterns and risk factors; Cluster analysis: The system uses the K-means algorithm to perform cluster analysis on large amounts of patient data to identify groups of patients with similar health characteristics: Among them, J is the objective function, C i is the i-th cluster, μ i is the center point of the cluster, and x is the sample point in the cluster; Association rule mining: The system uses the Apriori algorithm to mine association rules in health data: Among them, A and B are different sets of physiological parameters, and |D| is the total number of samples in the dataset.

10. The intelligent early warning method for in-hospital adverse cardiovascular event chain in patients with acute myocardial infarction according to claim 1, characterized in that: In step S8, based on the continuously collected data, the system generates a long-term health assessment report for the patient to provide feedback and suggestions. The long-term health score formula is: Where T is the length of the evaluation period, and F(t) is the comprehensive health status index at each moment; The system generates personalized health recommendations based on long-term health scores to provide real-time feedback to patients through the APP.

Images