Method and system for assisting in screening electrocardiogram abnormity based on patient information

Abstract

The invention belongs to the field of medical health, and relates to a method and system for assisting in screening electrocardiogram abnormity based on patient information. According to the method, multiple parameters related to health are introduced, the sign regression model with high interpretability is constructed by comprehensively analyzing the multiple parameters, and the probability of electrocardiogram abnormity is predicted through the sign regression model. And comparing the electrocardiogram of the detected person with the model to determine whether an abnormal condition exists or not. And if the abnormal possibility is high, the system can remind the patient and the doctor to carry out further heart examination in time. By means of the innovation, the patient can check and know the physical condition of the patient more easily by himself / herself, earlier health warning is provided at the same time, health management can be improved, and the risk of heart diseases can be relieved. According to the method, the feasibility of early diagnosis is improved, the model is more interpretable, and new possibility is provided for application in the fields of medical diagnosis equipment, mobile medical application, health management systems and the like.

Description

Technical Field

[0001] The present invention belongs to the field of medical health, and more specifically, to technologies related to cardiac health assessment and electrocardiogram (ECG) abnormality screening. Background Art

[0002] Heart disease is a major health problem worldwide, and early diagnosis is crucial for preventing and managing heart problems. Traditional ECG examination is a common method that detects heart abnormalities by recording the heart's electrical activity. However, this method has some limitations. First, it usually requires specialized facilities and professional doctors to interpret, which can lead to time delays and high medical costs. Second, traditional ECG mainly relies on electrocardiogram data and ignores other key health-related information, such as the patient's age, gender, liver function, thyroid health, gynecological health, etc. Most importantly, the accuracy of traditional ECG is limited in early diagnosis and may miss potential heart problems. Therefore, it is necessary to introduce a more comprehensive, accurate and efficient method to improve the early diagnosis of heart health problems. This method should be based on multi-parameter comprehensive analysis, consider multiple health-related factors, and have high prediction accuracy. Summary of the invention

[0003] The present invention provides a method and system for assisting in screening electrocardiogram abnormalities based on patient information, which is used to assist in screening electrocardiogram abnormalities based on patient information to improve the accuracy and efficiency of early diagnosis of heart health problems. The system combines multiple health-related parameters, including but not limited to the age, gender, vision status, cardiovascular health indicators, liver function, thyroid health, gynecological health and kidney function of the subject. These parameters are comprehensively analyzed to train a symbolic regression model to construct a highly interpretable prediction formula, and the formula constructed by the model is used to predict the possibility of electrocardiogram abnormalities.

[0004] In one aspect, the present invention provides a method for assisting in screening electrocardiogram abnormalities based on patient information, comprising the following steps: S1. Collect patient information, which refers to health-related parameters, including the subject's age, gender, vision, cardiovascular health indicators, liver function, thyroid health, gynecological health, and renal function information; S2. By comprehensively analyzing the above-mentioned various health-related parameters, a highly interpretable symbolic regression model is constructed; S3 evaluates the patient using the symbolic regression model to predict the likelihood of an electrocardiogram abnormality.

[0005] Another aspect of the present invention provides a system for assisting in screening electrocardiogram abnormalities based on patient information, comprising: A receiving module for receiving patient information, used for receiving the subject's age, gender, vision status, cardiovascular health index, liver function, thyroid health, gynecological health and renal function parameters; The training module is used to train the regression model through various parameters to build a symbolic regression model with strong interpretability; An evaluation module, used to evaluate the patient using the symbolic regression model, input relevant pathological parameters, and predict the possibility of electrocardiogram abnormality; The notification module is used to determine the probability value of electrocardiogram abnormality from the prediction results obtained by the symbolic regression model, and send a reminder for heart examination to the patient and the doctor if the probability value is higher than a predetermined threshold.

[0006] Specifically, the system first receives patient information, which can be obtained in a variety of ways, including self-entry by the patient or extracted from medical records. Then, a symbolic regression model is constructed by comprehensively analyzing this information. This model is trained using previous health data, including electrocardiogram data and health-related parameter data from multiple patients.

[0007] The symbolic regression model is used to evaluate the patient to predict the likelihood of an ECG abnormality. The model generates a prediction result that reflects the likelihood of the patient having an ECG abnormality. If the prediction result shows a high likelihood of abnormality, the system will send a reminder for a heart check to the patient and the doctor to further confirm or exclude the ECG abnormality. In addition, the system may also include a user interface that allows the patient to view their health information and the likelihood of an ECG abnormality, thereby promoting patient participation in health management. DETAILED DESCRIPTION

[0008] Example 1: Patient information collection and model construction In the first specific embodiment of the present invention, we will discuss in detail the process of collecting patient information and the construction of the symbolic regression model.

[0009] 1. Patient information collection In order to start the screening of ECG abnormalities, the first task is to collect various parameters related to the patient's health. These parameters will be used to build a symbolic regression model for the prediction of ECG abnormalities. Patient information collection is the first step of the system, and its key steps are as follows: 1.1 Data Source Patient information can be obtained from multiple sources, including but not limited to the following: Patient self-entry: Patients can use the application or website provided by the system to self-enter health-related information such as age, gender, vision status, etc. This information can be easily entered through a user-friendly interface to establish a patient's personal health profile; Medical record extraction: For patients who already have medical records, the system can extract the patient's health data through the electronic health record (EHR) system or hospital information system. This method helps to obtain more comprehensive and accurate patient information.

[0010] 1.2 Parameter Types The patient information collected includes not only basic information but also a variety of health-related parameters. Here are some examples of the types of parameters that may be collected: Age: This is an important parameter as heart health problems are often associated with age.

[0011] Gender: Gender has an important influence on the risk of cardiovascular disease and is therefore essential information.

[0012] Vision Conditions: Vision problems may affect a patient's interpretation of an EKG.

[0013] Cardiovascular health indicators: including parameters such as blood pressure, cholesterol levels, body mass index (BMI), etc., which are used to assess cardiovascular health status.

[0014] Liver function: Abnormal liver function may affect drug metabolism and heart health.

[0015] Thyroid health: Abnormal thyroid function may lead to problems such as irregular heartbeat.

[0016] Gynecological health (for women): Gynecological issues may affect heart health, especially during pregnancy.

[0017] Kidney function: Kidney health is related to fluid balance and electrolyte homeostasis in the body, which is critical to heart function.

[0018] 1.3 Data Privacy and Security The privacy and security of patient information are important aspects of system design. Appropriate encryption, access control, and data protection measures should be fully considered to ensure that patient information is not accessed or disclosed without authorization.

[0019] 2. Symbolic regression model construction After collecting patient information, the next step is to build a highly interpretable symbolic regression model. This model will be trained using previous health data, including electrocardiogram data and health-related parameter data from multiple patients.

[0020] 2.1 Data Preprocessing Before building a model, the data needs to be preprocessed, which includes steps such as data cleaning, missing value processing, and standardization to ensure the quality and consistency of the data.

[0021] 2.2 Initializing the population Define a set of functions: select possible functions such as addition, subtraction, multiplication, division, etc.

[0022] Define the set of terminal symbols: select possible input variables and constants.

[0023] Randomly generated trees: Use a set of functions and a set of terminal symbols to randomly generate individuals in a tree structure, which will become members of the initial population.

[0024] 2.3 Model Training and Validation The model training process involves dividing the existing health data into a training set and a validation set. The training set is used to learn the model, while the validation set is used to evaluate the performance of the model. The performance indicators of the model may include accuracy, sensitivity, specificity, etc.

[0025] 2.4 Evaluating Fitness For each individual, calculate its fitness, that is, its performance in solving the regression problem. The fitness function can be the mean square error, correlation coefficient, or other regression performance indicators. The higher the fitness, the closer the individual is to the goal of solving the problem.

[0026] 2.5 Crossover Operation The selected parent individuals will create offspring through a crossover operation. The crossover operation swaps part of the tree structure of two individuals to produce new individuals. This simulates the crossover of genes.

[0027] 2.6 Mutation Operation For some individuals, a mutation operation is performed, which involves changing part of the individual's tree structure to introduce new diversity. This simulates the mutation of genes.

[0028] 2.7 Evaluation of New Individuals For the generated offspring individuals and the unselected parent individuals, their fitness is calculated.

[0029] 2.8 Environment Selection Select a portion of individuals from the parent and offspring as the next generation population. Usually, the selection method gives priority to individuals with high fitness to ensure that the solution to the problem is gradually optimized.

[0030] 2.9 Termination Conditions Check if the termination condition is met. This could be reaching a certain fitness threshold, reaching a maximum number of iterations, or other conditions. If the condition is met, the algorithm terminates.

[0031] 2.10 Output the best individual When the algorithm terminates, the individual with the highest fitness is output, which represents the best solution to the symbolic regression model.

[0032] 2.11 Interpreting the results Interpret the best individuals to obtain the final symbolic regression model. This involves translating the tree structure into mathematical expressions or program code.

[0033] These steps constitute the basic process of genetic programming. To actually implement GP to build a symbolic regression model, you need to write the corresponding code to perform the above steps and choose the appropriate set of functions, terminal symbols, and parameter configuration to solve the specific regression problem. In addition, you need to carefully adjust the parameters of the algorithm to obtain the best results.

[0034] 2.4 Model Interpretability The interpretability of the model is very important for doctors and patients. The system should provide tools or interfaces so that users can understand the prediction results of the model. The model has strong visibility and can present the internal mechanism of the judgment, so that users can better understand the possibility of ECG abnormality.

[0035] The above is a detailed description of the first specific embodiment of the present invention, covering the process of collecting patient information and constructing a symbolic regression model.

[0036] Example 2: Electrocardiogram abnormality prediction and reminder system In the second specific embodiment of the present invention, we will discuss in detail the use of symbolic regression model and the prediction and reminder system of electrocardiogram abnormalities.

[0037] 1. Use of symbolic regression model The constructed symbolic regression model will be the core of this system. The model should be able to accept patient information as input and generate predictions about the likelihood of an ECG abnormality.

[0038] 1.1 Patient information input First, the user needs to enter the patient's information into the system. This can be achieved through a user interface where patients can easily enter or update their health-related data.

[0039] 1.2 Model Evaluation Next, the system evaluates the patient using a symbolic regression model that takes into account a variety of parameters, including age, gender, vision, cardiovascular health, liver function, thyroid health, gynecological health, and kidney function, to predict the likelihood of an abnormal ECG.

[0040] 2. ECG abnormality prediction and reminder system Once the model has completed its evaluation, it generates a prediction reflecting the likelihood of the patient having an abnormal ECG. The system then takes action based on this prediction.

[0041] 2.1 Interpretation of prediction results The system should provide a user-friendly interface to explain the model’s predictions. This should include explaining the likelihood of the predictions and the factors associated with various health parameters.

[0042] 2.2 Reminder System If the model's predictions indicate a high likelihood of abnormality, the system will send a reminder to the patient and doctor for a heart checkup. This can be done in a variety of ways, including mobile app notifications, emails, text messages, or phone calls.

[0043] 2.3 User Interface The user interface should allow patients to view their health information and the likelihood of ECG abnormalities. This interface may provide a graphical display to help users better understand their health status.

[0044] The above is a detailed description of the second specific embodiment, covering the use of the symbolic regression model and the operation process of the prediction and reminder system for abnormal electrocardiograms. Next, we will explore the third specific embodiment.

[0045] Example 3: Automatic model update and optimization In the third specific embodiment of the present invention, we will discuss in detail the automatic updating and optimization process of the symbolic regression model.

[0046] 1. Automatic model update To keep the model accurate and useful, the system should have the ability to automatically update the model, which means the model can reflect the patient's latest health data.

[0047] 1.1 Continuous Data Collection The system needs to collect new health data from patients on a regular basis. This can be achieved through integration with medical record systems or through user self-updates.

[0048] 1.2 Model Retraining Once enough new data is available, the system should trigger a retraining of the symbolic regression model. This process will use the new data to update the model to ensure it reflects the latest health trends for the entire population of patients.

[0049] 2. Model Optimization The performance of symbolic regression models can be further improved through continuous optimization. This includes increasing the diversity and number of data sources to improve the predictive performance of the model.

[0050] 2.1 Data Diversity The system should strive to collect patient data from different regions, age groups, and health conditions to increase the data diversity of the model.

[0051] 2.2 Data quantity The performance of a model is usually correlated to the amount of training data. Therefore, the system should constantly strive to collect more data to improve the accuracy of the model.

[0052] The above is a detailed description of the third specific implementation method, covering the process of automatic model update and optimization. This process ensures the continuous performance improvement and adaptability of the system.

[0053] The detailed description of these three specific embodiments provides a deep understanding of how to collect patient information, build models, predict ECG abnormalities and optimize the performance of the system. These methods together constitute the core of the present invention and provide a solid foundation for providing a more comprehensive, accurate and interpretable cardiac health assessment.

Claims

1. A method for assisting in screening electrocardiogram abnormalities based on patient information, characterized in that: The method comprises the following steps: S1. Collect patient information, which refers to health-related parameters, including the subject's age, gender, vision, cardiovascular health indicators, liver function, thyroid health, gynecological health, and renal function information; S2. By comprehensively analyzing the above-mentioned various health-related parameters, a highly interpretable symbolic regression model is constructed; S3 evaluates the patient using the symbolic regression model to predict the likelihood of an electrocardiogram abnormality.

2. A method for assisting in screening electrocardiogram abnormalities based on patient information according to claim 1, characterized in that: In step S1, the patient information is obtained by inputting it into the online system and extracting medical records online.

3. The method for assisting in screening electrocardiogram abnormalities based on patient information according to claim 1, characterized in that: The model building in step S2 specifically includes: 2.1 Data Preprocessing Preprocess the collected patient data, including data cleaning, missing value processing, and standardization; 2.2 Initializing the population Define a set of functions: select possible functions such as addition, subtraction, multiplication, and division; Define the set of terminal symbols: select possible input variables and constants; Random tree generation: Use a set of functions and a set of terminal symbols to randomly generate individuals in a tree structure, which will become members of the initial population; 2.3 Model Training and Validation The existing health data is divided into a training set and a validation set. The training set is used for model learning, and the validation set is used to evaluate the performance of the model. 2.4 Evaluating Fitness For each individual, calculate its fitness; 2.5 Crossover Operation The selected parent individuals will create offspring through a crossover operation; the crossover operation swaps parts of the tree structure of two individuals to produce new individuals; 2.6 Mutation Operation For some individuals, mutation operations are performed; 2.7 Evaluation of New Individuals For the generated offspring individuals and the unselected parent individuals, calculate their fitness; 2.8 Environment Selection Select a portion of individuals from the parent and offspring generations as the next generation population; 2.9 Termination Conditions Check whether the termination conditions are met, such as reaching a certain fitness threshold, reaching the maximum number of iterations or other conditions. If the conditions are met, the algorithm terminates; 2.10 Output the best individual When the algorithm terminates, the individual with the highest fitness is output.

4. The method for assisting in screening electrocardiogram abnormalities based on patient information according to claim 1, characterized in that: The S3 includes: 3.1 Determine the probability value of electrocardiogram abnormality from the prediction results obtained by the symbolic regression model; 3.2 If the probability value is higher than a predetermined threshold, a reminder for a heart checkup is sent to the patient and the doctor.

5. A system for assisting in screening electrocardiogram abnormalities based on patient information, characterized in that: include: A receiving module for receiving patient information, used for receiving the subject's age, gender, vision status, cardiovascular health index, liver function, thyroid health, gynecological health and renal function parameters; The training module is used to train the regression model through various parameters to build a symbolic regression model with strong interpretability; An evaluation module, used to evaluate the patient using the symbolic regression model, input relevant pathological parameters, and predict the possibility of electrocardiogram abnormality; The notification module is used to determine the probability value of electrocardiogram abnormality from the prediction results obtained by the symbolic regression model, and send a reminder for heart examination to the patient and the doctor if the probability value is higher than a predetermined threshold.

6. A system for assisting in screening electrocardiogram abnormalities based on patient information according to claim 5, characterized in that: The system also includes a user interface for the patient to view his or her health information and the possibility of abnormal ECG.

7. The system for assisting in screening electrocardiogram abnormalities based on patient information according to claim 5, characterized in that: The notification module can send reminders to patients and doctors via mobile applications, emails, text messages or phone calls.