System and method for evaluating potential complications of department of cardiology

Abstract

The invention relates to the technical field of medical evaluation, and provides an evaluation system and method for potential complications of the department of cardiology, and the method comprises the following steps: S1, collecting information data of a patient, and carrying out the data verification and data preprocessing; s2, based on the obtained information, classifying types of complications which may occur on the patient; and S3, based on a complication type classification result, risk factors are extracted, and a risk factor network is constructed. By classifying the complication types, constructing the risk factor network and comprehensively considering multiple factors to calculate the risk index, compared with a traditional assessment mode, the risk of potential complication of a patient can be assessed more comprehensively and accurately, misdiagnosis and missed diagnosis caused by incomplete assessment are reduced, and the risk assessment efficiency is improved based on the risk score, the risk index and the risk type. The drug dosage or the measure intensity is adjusted through the patient specificity index, the prevention measures are dynamically adjusted in combination with the treatment time, and the pertinence and effectiveness of the prevention scheme are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of medical evaluation, and particularly to an evaluation system and method for potential complications in cardiology. Background Art

[0002] Currently, the evaluation of potential complications in cardiology mainly relies on doctors' clinical experience, lacking a systematic, comprehensive and accurate evaluation method. Traditional evaluation methods often only consider a single factor or a few factors, making it difficult to accurately judge the overall risk of patients. At the same time, there are differences in the experience and judgment criteria of different doctors, resulting in the lack of consistency and reliability of evaluation results.

[0003] In terms of technical means, although technologies such as electrocardiogram, echocardiogram, and blood biochemical index detection have been widely used in clinical practice, these data are often used in isolation without fully integrating and exploring their potential value. With the development of medical informatization, a large amount of medical data has been generated and stored, but how to effectively use these data for complication risk assessment has become a difficult problem. Summary of the Invention

[0004] Aiming at the problems existing in the prior art, the purpose of the present invention is to provide an evaluation system and method for potential complications in cardiology to solve the problems raised in the above background art.

[0005] To achieve the above purpose, the present invention provides an evaluation method for potential complications in cardiology, including the following steps:

[0006] S1. Collect patient information data, and perform data verification and data preprocessing;

[0007] S2. Classify the types of complications that may occur to the patient based on the obtained information;

[0008] S3. Extract risk factors based on the classification result of complication types, and construct a risk factor network;

[0009] S4. Calculate the risk index of the patient's occurrence of a specific complication based on the constructed risk factor network;

[0010] S5. Generate a corresponding prevention plan based on the risk score, risk index, and risk type;

[0011] S6. Monitor the complications that may occur to the patient in real time, identify abnormal trends in the occurrence of complications, and set an early warning threshold to trigger an early warning when the threshold is exceeded;

[0012] S7. Track the implementation effect of the prevention plan and collect it into the database as reliable data for model optimization.

[0013] Preferably, in the step S1, collecting and organizing data includes the following steps:

[0014] S11. Obtain structured electronic medical record data, electrocardiograms, color ultrasounds, and real-time vital sign data of patients, and calculate the data integrity. The formula is:

[0015] ,

[0016] In the formula, is the data integrity rate, is the number of actually collected data items, is the number of data items that should be collected;

[0017] S12. After normalizing the collected data, use the multiple imputation method to supplement the missing values of the data. The formula is:

[0018] ,

[0019] In the formula, is the filled value of the data, is the collected data, is the observed data, is the random error term;

[0020] S13. Check whether the data is within the medically reasonable range, check the logical relationship between the data, and score the data consistency. The formula is:

[0021] ,

[0022] is the consistency score value, is the consistent data item, is the total number of comparison data items;

[0023] S14. Optimize the data features through feature engineering to improve the model performance;

[0024] S15. Perform entity matching based on the patient's unique identifier, and use weighted average to fuse multi-source data. The formula is:

[0025] ,

[0026] In the formula, is the fused data, is the data source 's weight, is the th value of the sample data point.

[0027] Preferably, in the step S14, optimizing the data features by feature engineering includes the following steps:

[0028] S141. Calculate the risk factors of the patient's cardiovascular complications and judge the risk probability of complications. The formula is:

[0029] ,

[0030] In the formula, is the risk factor of complications, is the total cholesterol, is the high-density lipoprotein cholesterol, is the triglyceride, is the age;

[0031] S142. Select features using the importance score of random forest. The formula is:

[0032] ,

[0033] is the importance score of feature selection, is the feature in the th tree, is the number of trees;

[0034] S143. Evaluate whether to retain features through feature variance and perform dimensionality reduction on features. The formula is:

[0035] ,

[0036] In the formula, is the value of the th sample data point, is the number of samples, is the sample mean, is the variance measuring the dispersion of sample data.

[0037] Preferably, in step S2, identifying the types of complications existing in the department of cardiology includes the following steps:

[0038] S21. Construct an association matrix of complications and risk features and calculate the association degree score of each complication of the patient. The formula is:

[0039] ,

[0040] In the formula, is the association degree score of the complication, is the quantified value of the th risk feature, is the important weight of the th risk feature, is the total number of risk features;

[0041] S22. Integrate the evidence of clinical features, laboratory indicators, and treatment history, calculate the confidence level, and perform multimodal evidence fusion. The formula is:

[0042] ,

[0043] In the formula, is the confidence level of the complication , is the basic probability assignment of the evidence to the proposition .

[0044] S23. Calculate the comprehensive priority based on the association score and confidence level of the complication, and rank the priority of complication types. The formula is:

[0045] ,

[0046] In the formula, is the comprehensive priority score, and are adjustable parameters;

[0047] S24. Set the risk threshold, and output a risk priority list according to multiple complications, sorted in descending order of the comprehensive priority score. The formula is:

[0048] ,

[0049] In the formula, is the set risk threshold;

[0050] S25. Update the risk feature weights according to the real-time monitoring data. The formula is:

[0051] ,

[0052] In the formula, is the weight of feature at time , is the change amount of feature , is the influence coefficient of feature on the complication, is the learning rate.

[0053] Preferably, in step S3, the risk factor network includes the following steps:

[0054] S31. Use an independent sample test to calculate the significance of the mean difference between data groups. At the same time, analyze the single-factor correlation through advantage comparison. The formula is:

[0055] ,

[0056] In the formula, is the test statistic for measuring the difference in the means of two sets of continuous variables, and are the means of the two sets of samples respectively, is the statistic obtained by synthesizing the dispersion degrees of the two sets of sample data, and are the numbers of the two sets of samples respectively, is the index reflecting the association strength between the exposure factor and the disease, and are the numbers of individuals exposed to and not exposed to a certain factor in the case group respectively, and are the numbers of individuals exposed to and not exposed to a certain factor in the control group respectively;

[0057] S32. Screen independent risk factors, calculate the regression coefficient and the risk ratio, and judge the significance of the interaction effect by introducing the interaction term. The formula is:

[0058] ,

[0059] In the formula, is the probability of complication occurrence, is the risk factor, is the regression coefficient, is to map the probability value to the real number domain after taking the logarithmic transformation of the complication occurrence probability , is the intercept term, is to measure the influence degree and direction of each risk factor on the log odds , is the chi-square statistic for testing whether the interaction term is significant, is the likelihood value reflecting the goodness of fit of the model to the data under the assumption of no interaction, is the likelihood value reflecting the goodness of fit of the model to the data when the interaction is included;

[0060] S33. Compress the coefficients of irrelevant variables to achieve feature sparsity, focus on the key risk factors of potential complications in cardiology, improve the model performance, and quantify the feature importance to screen key features to provide a basis for disease condition assessment and intervention. The formula is:

[0061] ,

[0062] In the formula, is to find the combination of regression coefficients that can minimize the objective function value to optimize the model, is the total number of patient samples participating in the evaluation and analysis, and are the probabilities of actual and predicted occurrence of potential complications for the th patient, is the regularization parameter, is the total number of risk factors included in the evaluation system, is the th regression coefficient corresponding to the th risk factor, is the importance score of the th risk factor for predicting potential complications in the cardiology department, is the total number of patient samples used by the th decision tree in the random forest model, and are the th risk factor before splitting and after splitting along the coefficients of the nodes;

[0063] S34. Calculate the conditional probability between the risk factor and the complication, and construct a directed acyclic graph. The formula is:

[0064] ,

[0065] In the formula, is the conditional probability of occurrence of cardiology complications under the condition that the risk factor occurs, is the joint probability of simultaneous occurrence of the risk factor and cardiology complications, is the conditional probability of occurrence of cardiology complications, is the joint probability of simultaneous occurrence of risk factors.

[0066] Preferably, in the step S4, calculating the risk index of the patient's specific complication includes the following steps:

[0067] S41. Draw a causal diagram of the risk factor and the complication, and identify the confounding variables. The formula is:

[0068] ,

[0069] In the formula, is the risk factor effect value, is the complication, is the target factor, is the set of confounding variables;

[0070] S42. For the time-varying indicators, extract the time series features, analyze the recent fluctuation coefficient and trend slope, and evaluate the time-dependent risk after the patient's surgery or medication. The formula is:

[0071] ,

[0072] In the formula, is the short-term fluctuation coefficient for measuring the short-term fluctuation degree of complications, and as well as are respectively the maximum value, the minimum value and the average value of the complication occurrence probability within one week, is the trend slope reflecting the trend of complications changing over time, and are respectively the th time point and the time average value, and are respectively the complication occurrence probability value corresponding to the th time point and the average probability of complications occurring in patients within one week, is the risk value of the patient after surgery or medication, is the risk coefficient, is the number of days after surgery or medication, is the weight coefficient, is the complication history index;

[0073] S43. Train the model by combining the data of multiple hospitals and establish a vertical federated learning framework. The formula is:

[0074] ,

[0075] In the formula, is the global model, and are respectively the model parameters of the th participating hospital at the and th round of training, is the learning rate, is the loss function with respect to the model parameter gradient;

[0076] S44. Adjust the patient-specific parameters, and then adjust the global model coefficients to improve the prediction accuracy of the model. The formula is:

[0077] ,

[0078] is the th model coefficient after personalized adjustment of the patient, is the th coefficient in the global model, is the th personalized offset, is the biomarker ratio;

[0079] S45. Integrate the machine learning score and the clinical guideline score, and obtain the comprehensive risk score through dynamic parameter adjustment. The formula is:

[0080] ,

[0081] is the comprehensive risk score, is the fusion weight parameter, is the machine learning score, is the clinical guideline score.

[0082] Preferably, in the step S5, generating the prevention plan includes the following steps:

[0083] S51. Divide the risk levels according to the comprehensive risk score, and clarify the very low, low, medium, high, and extremely high risk intervals and corresponding thresholds;

[0084] S52. Establish a prevention measure library including drug treatment, lifestyle intervention, and surgical treatment categories, and code the applicable scenarios and contraindications according to the guideline standards;

[0085] S53. Exclude the contraindicated plans and generate a quantified recommended plan;

[0086] S54. According to the patient-specific indicators, adjust the drug dosage or measure intensity to ensure individualized and safe drug use, and dynamically attenuate or enhance the prevention measures in combination with the treatment time to adapt to the changes in the condition;

[0087] S55. Check the compliance of the plan against the latest clinical guidelines to ensure that there are no omissions or conflicts in the key measures, and generate a visual explanation report to present the recommendation basis in structured language.

[0088] An assessment system for potential complications in cardiology, applied to the assessment method for potential complications in cardiology described in any one of the above, includes:

[0089] A data integration module, used for the system data entry, integrating the functions of patient information collection, data verification, and data preprocessing;

[0090] A complication type identification module, based on the patient's underlying diseases, treatment methods, and the structured data output by the data integration module, classifies the possible complication types;

[0091] A risk factor analysis module, based on the target complications output by the complication type identification module and the relevant risk factors extracted from the structured data of the data integration module, constructs a risk factor network;

[0092] A multi-dimensional risk assessment module, based on the results of the risk factor analysis module, uses a multi-dimensional assessment model to calculate the risk index of the patient having a specific complication;

[0093] A preventive measure generation module, based on the risk scores and complication types output by the multi-dimensional risk assessment module, matches corresponding general preventive measures from a preset preventive measure knowledge base to generate a personalized prevention plan.

[0094] A dynamic monitoring module monitors the patient in real time, compares the monitoring data with the baseline data, identifies abnormal trends, and triggers an early warning mechanism.

[0095] An effect feedback module is used to track the implementation effect of preventive measures and optimize the risk assessment model and preventive measure library.

[0096] The beneficial effects of an evaluation system and method for potential complications in cardiology provided by the present invention are as follows:

[0097] 1. By collecting multi-dimensional information of the patient, including structured electronic medical record data, electrocardiogram, color Doppler ultrasound, real-time vital sign data, etc., and performing data verification and preprocessing to ensure the integrity and accuracy of the data. On this basis, classify the complication types, construct a risk factor network, and comprehensively consider various factors to calculate the risk index. Compared with traditional evaluation methods, it can more comprehensively and accurately evaluate the risk of potential complications of the patient, reducing misdiagnosis and missed diagnosis caused by incomplete evaluation.

[0098] 2. Based on the risk scores, risk indices, and risk types, the system can generate a personalized prevention plan for the patient from a preventive measure library including categories such as drug treatment, lifestyle intervention, and surgical treatment. Adjust the drug dosage or measure intensity through patient-specific indicators, and dynamically adjust preventive measures in combination with the treatment time, improving the pertinence and effectiveness of the prevention plan, ensuring individualized safe medication, and better meeting the treatment needs of different patients. BRIEF DESCRIPTION OF THE DRAWINGS

[0099] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0100] Figure 1 It is a schematic flowchart of the method for an evaluation system and method for potential complications in cardiology provided by this application;

[0101] Figure 2 It is a schematic diagram of the system modules of an evaluation system and method for potential complications in cardiology provided by this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0102] The following further describes in detail the specific implementation manners of the present invention in conjunction with the accompanying drawings of the specification and embodiments. The following embodiments are only used to illustrate the present invention, but cannot be used to limit the scope of the present invention.

[0103] As Figure 1 - Figure 2 shown, this embodiment proposes an evaluation method for potential complications in cardiology, including the following steps:

[0104] S1. Collect patient information data, and perform data verification and data preprocessing;

[0105] S2. Classify the types of complications that may occur to the patient based on the obtained information;

[0106] S3. Extract risk factors based on the classification results of complication types, and construct a risk factor network;

[0107] S4. Calculate the risk index of the patient having a specific complication based on the constructed risk factor network;

[0108] S5. Generate corresponding prevention plans based on the risk score, risk index, and risk type;

[0109] S6. Monitor in real time the complications that may occur to the patient, identify abnormal trends in the occurrence of complications, and set an early warning threshold to trigger an early warning when the threshold is exceeded;

[0110] S7. Track the implementation effect of the prevention plan and collect it into the database as reliable data for model optimization.

[0111] Specifically, by collecting multi-dimensional information of the patient, including structured electronic medical record data, electrocardiogram, color Doppler ultrasound, real-time vital sign data, etc., and performing data verification and preprocessing to ensure the integrity and accuracy of the data. On this basis, classify the types of complications, construct a risk factor network, and comprehensively consider various factors to calculate the risk index. Compared with the traditional evaluation method, it can more comprehensively and accurately evaluate the risk of the patient having potential complications, and reduce misdiagnosis and missed diagnosis caused by incomplete evaluation.

[0112] In this embodiment, in step S1, the collection and collation of data include the following steps:

[0113] S11. Obtain structured electronic medical record data, electrocardiogram, color Doppler ultrasound, and the patient's real-time vital sign data, and calculate the data integrity, and the formula is:

[0114] ,

[0115] In the formula, is the data integrity rate, is the number of actually collected data items, is the number of data items to be collected;

[0116] S12. After normalizing the collected data, use the multiple imputation method to supplement the missing values of the data. The formula is:

[0117] ,

[0118] In the formula, is the imputed value of the data, is the collected data, is the observed data, is the random error term;

[0119] S13. Check whether the data is within the medically reasonable range, check the logical relationship between the data, and score the data consistency. The formula is:

[0120]

[0121] is the consistency score value, is the consistent data item, is the total number of comparison data items;

[0122] S14. Optimize the data features through feature engineering to improve the model performance;

[0123] S15. Perform entity matching based on the patient's unique identifier, and adopt weighted average to fuse multi-source data. The formula is:

[0124] ,

[0125] In the formula, is the fused data, is the data source 's weight, is the th value of the sample data point.

[0126] Specifically, collect the patient's basic information through multiple channels to ensure the integrity of key basic data and the unity of the quantification standard, covering multi-dimensional information of the patient's individual characteristics, providing raw data support for subsequent data processing and risk assessment.

[0127] In this embodiment, in step S14, the feature engineering to optimize the data features includes the following steps:

[0128] S141. Calculate the risk factors of the patient's cardiovascular complications and judge the complication risk probability. The formula is:

[0129] ,

[0130] In the formula, is a complication risk factor, is total cholesterol, is high-density lipoprotein cholesterol, is triglyceride, is age;

[0131] S142. Select features using the random forest importance score. The formula is:

[0132] ,

[0133] is the importance score of feature selection, is the feature in the importance in the tree;

[0134] S143. Evaluate whether to retain features through feature variance and perform dimensionality reduction on features. The formula is:

[0135] ,

[0136] In the formula, is the value of the th sample data point, is the number of samples, is the sample mean, is the variance measuring the dispersion of sample data.

[0137] Specifically, through clinical means such as electrocardiogram, echocardiogram, blood biochemistry, and coagulation function tests, core indicators are extracted. Their accurate collection and quantification directly affect the accuracy of subsequent risk assessment, and particularly provide objective medical indicators for predicting complications such as heart failure and thrombosis. Specifically, through standardization processing, the original data is converted into high-quality structured data that can be used for modeling analysis, solving the problem of clinical data heterogeneity, ensuring that data from different sources and of different types can be uniformly processed, and at the same time improving data reliability through a data verification mechanism, providing guarantees for the stability and accuracy of the risk assessment model.

[0138] In this embodiment, in step S2, identifying the types of complications existing in the department of cardiology includes the following steps:

[0139] S21. Construct an association matrix of complications and risk features and calculate the association degree score of each complication of the patient. The formula is:

[0140] ,

[0141] In the formula, is the association degree score of the complication, is the The quantified value of a risk feature is the importance weight of the th risk feature;

[0142] S22. Integrate the evidence of clinical features, laboratory indicators, and treatment history, calculate the confidence level, and perform multimodal evidence fusion. The formula is:

[0143] ,

[0144] In the formula, is the confidence level of the complication , is the basic probability assignment of the evidence to the proposition ;

[0145] S23. Calculate the comprehensive priority based on the association degree score and confidence level of the complication, and rank the priority of complication types. The formula is:

[0146] ,

[0147] In the formula, is the comprehensive priority score, and are adjustable parameters;

[0148] S24. Set the risk threshold, and output a risk priority list according to multiple complications, sorted in descending order of the comprehensive priority score. The formula is:

[0149] ,

[0150] In the formula, is the set risk threshold;

[0151] S25. Update the risk feature weight according to the real-time monitoring data. The formula is:

[0152] ,

[0153] In the formula, is the weight of feature at time , is the change amount of feature , is the influence coefficient of feature on the complication, is the learning rate.

[0154] In this embodiment, in step S3, the risk factor network includes the following steps:

[0155] S31. Use independent sample tests to calculate the significance of the mean difference between data groups. At the same time, analyze the single-factor correlation through advantage comparison. The formula is:

[0156] ,

[0157] In the formula, is the test statistic for measuring the degree of mean difference between two groups of continuous variables. and are the means of the two groups of samples respectively. is the statistic obtained by synthesizing the dispersion degrees of the two groups of sample data. and are the numbers of the two groups of samples respectively. is the index reflecting the association strength between the exposure factor and the disease. and are the numbers of individuals exposed to and not exposed to a certain factor in the case group respectively. and are the numbers of individuals exposed to and not exposed to a certain factor in the control group respectively.

[0158] S32. Screen independent risk factors, calculate the regression coefficient and risk ratio, and judge the significance of the interaction effect by introducing interaction terms and using likelihood ratio tests. The formula is:

[0159] ,

[0160] In the formula, is the probability of complication occurrence. is the risk factor. is the regression coefficient. is to map the probability value to the real number domain after logarithmically transforming the probability of complication occurrence . is the intercept term. is to measure the influence degree and direction of each risk factor on the log odds . is the chi-square statistic for testing whether the interaction term is significant. is the likelihood value reflecting the goodness of fit of the model to the data under the assumption of no interaction. is the likelihood value reflecting the goodness of fit of the model to the data when including the interaction.

[0161] S33. Compress the coefficients of irrelevant variables to achieve feature sparsity, focus on the key risk factors of potential complications in cardiology, improve the model performance, and quantify the feature importance to screen key features to provide a basis for disease assessment and intervention. The formula is:

[0162] ,

[0163] In the formula, is to find the regression coefficients that can minimize the objective function value combination to optimize the model. is the total number of patient samples participating in the evaluation and analysis. and are respectively the probabilities of the actual and predicted occurrence of potential complications for the th patient. is the regularization parameter. is the total number of risk factors included in the evaluation system. is the regression coefficient corresponding to the th risk factor. is the importance score of the th risk factor for predicting potential complications in the cardiology department. is the total number of patient samples used by the th decision tree in the random forest model. and are respectively the coefficients of the node before splitting and the node after splitting along the th risk factor;

[0164] S34. Calculate the conditional probability between the risk factor and the complication, and construct a directed acyclic graph. The formula is:

[0165] ,

[0166] In the formula, is the conditional probability of the occurrence of a cardiology complication under the condition that the risk factor has occurred. is the joint probability of the simultaneous occurrence of the risk factor and the cardiology complication. is the conditional probability of the occurrence of a cardiology complication. is the joint probability of the simultaneous occurrence of the risk factors.

[0167] In this embodiment, in step S4, calculating the risk index of a patient's specific complication includes the following steps:

[0168] S41. Draw a causal diagram of the risk factor and the complication, and identify the confounding variables. The formula is:

[0169] ,

[0170] In the formula, is the risk factor effect value. is the complication. is the target factor. is the set of confounding variables;

[0171] S42. For metrics that change over time, extract time-series features, analyze the recent fluctuation coefficient and trend slope, and evaluate the time-dependent risk after the patient's surgery or medication. The formula is:

[0172] ,

[0173] In the formula, is the recent fluctuation coefficient that measures the recent fluctuation degree of complications, and as well as are respectively the maximum, minimum, and mean values of the complication occurrence probability within one week, is the trend slope that reflects the trend of complications changing over time, and are respectively the th time point and the time mean, and are respectively the complication occurrence probability value corresponding to the th time point and the mean complication occurrence probability of the patient within one week, is the risk value after the patient's surgery or medication, is the risk coefficient, is the number of days after surgery or medication, is the weight coefficient, is the complication history index;

[0174] S43. Combine the data of multiple hospitals to train the model and establish a vertical federated learning framework. The formula is:

[0175] ,

[0176] In the formula, is the global model, and are respectively the model parameters of the th participating hospital at the and rounds of training, is the learning rate, is the loss function with respect to the model parameter gradient;

[0177] S44. Adjust the patient-specific parameters, and then adjust the global model coefficients to improve the model prediction accuracy. The formula is:

[0178] ,

[0179] is the th model coefficient after the patient's personalized adjustment, is the th coefficient in the global model, is the th personalized offset, is the biomarker ratio;

[0180] S45, fusing the machine learning score and the clinical guideline score, and obtaining the comprehensive risk score through dynamic parameter adjustment. The formula is:

[0181] ,

[0182] is the comprehensive risk score, is the fusion weight parameter, is the machine learning score, is the clinical guideline score.

[0183] Specifically, the specific implementation logic of constructing the risk assessment module is designed, including the risk factor assignment rules and the comprehensive risk score. By combining clinical experience with quantitative calculation, a risk assessment model that can be standardized and applied is formed, which can quickly output the risk level based on the individual characteristics of patients, providing an intuitive and quantifiable reference basis for clinical decision-making, and is particularly applicable to the risk stratification of common cardiovascular complications such as heart failure and thrombosis.

[0184] In this embodiment, in step S5, generating the prevention plan includes the following steps:

[0185] S51. Divide the risk levels according to the comprehensive risk score, and clarify the very low, low, medium, high, and extremely high risk intervals and the corresponding thresholds;

[0186] S52. Establish a prevention measure library including drug treatment, lifestyle intervention, and surgical treatment categories, and code the applicable scenarios and contraindications according to the guideline standards;

[0187] S53. Exclude the contraindicated plans and generate a quantitative recommended plan;

[0188] S54. According to the patient-specific indicators, adjust the drug dosage or the intensity of the measures to ensure individualized and safe drug use, and dynamically attenuate or enhance the prevention measures in combination with the treatment time to adapt to the changes in the condition;

[0189] S55. Check the compliance of the plan against the latest clinical guidelines to ensure that there are no omissions or conflicts in the key measures, and generate a visual explanation report to present the recommended basis in structured language.

[0190] An assessment system for potential cardiovascular complications, which is applied to an assessment method for potential cardiovascular complications in any one of the above, includes:

[0191] A data integration module, used for the system data entry, integrating the functions of patient information collection, data verification, and data preprocessing;

[0192] A complication type identification module, based on the patient's underlying diseases, treatment methods, and the structured data output by the data integration module, classifies the possible complication types;

[0193] A risk factor analysis module, based on the target complications output by the complication type identification module and extracting relevant risk factors from the structured data of the data integration module, constructs a risk factor network;

[0194] A multi-dimensional risk assessment module, based on the results of the risk factor analysis module, adopts a multi-dimensional assessment model to calculate the risk index of the patient having a specific complication;

[0195] A preventive measure generation module, based on the risk score and complication type output by the multi-dimensional risk assessment module, matches the corresponding general preventive measures from the preset preventive measure knowledge base to generate a personalized prevention plan;

[0196] A dynamic monitoring module, monitors the patient in real time, compares the monitoring data with the baseline data, identifies abnormal trends, and triggers an early warning mechanism;

[0197] An effect feedback module, used to track the implementation effect of preventive measures and optimize the risk assessment model and preventive measure library.

[0198] Specifically, based on the risk score, risk index, and risk type, the system can generate a personalized prevention plan for the patient from a preventive measure library including categories such as drug treatment, lifestyle intervention, and surgical treatment, adjust the drug dosage or measure intensity through patient-specific indicators, dynamically adjust preventive measures in combination with the treatment time, improve the pertinence and effectiveness of the prevention plan, ensure individualized safe medication, and better meet the treatment needs of different patients.

[0199] The above embodiments are only used to illustrate the present invention, rather than limiting the present invention. Although the present invention has been described in detail with reference to the embodiments, those of ordinary skill in the art should understand that various combinations, modifications, or equivalent replacements of the technical solutions of the present invention do not depart from the spirit and scope of the technical solutions of the present invention, and should all be covered within the scope of the claims of the present invention.

Claims

1. A method for evaluating potential complications in cardiology department, characterized in that, It includes the following steps: S1. Collect patient information data, and perform data verification and data preprocessing; S2. Based on the obtained information, classify the types of complications that the patient may develop; S3. Based on the classification results of the complication types, extract risk factors and construct a risk factor network; S4. Based on the constructed risk factor network, calculate the risk index of the patient developing a specific complication; S5. Based on the risk score, risk index, and risk type, generate corresponding prevention plans; S6. Monitor in real time the complications that the patient may develop, identify abnormal trends in the occurrence of complications, and set warning thresholds to trigger warnings when the thresholds are exceeded; S7. Track the implementation effect of the prevention plan and collect it into the database as reliable data for model optimization.

2. The evaluation method of potential complications in cardiology department according to claim 1, characterized in that In step S1, the collection and collation of data include the following steps: S11. Obtain structured electronic medical record data, electrocardiograms, color Doppler ultrasounds, and real-time vital sign data of the patient, and calculate the data integrity. The formula is: ， Wherein, is the data integrity rate, is the actual number of collected data items, is the number of data items to be collected; S12. After normalizing the collected data, use the multiple imputation method to supplement the missing values of the data. The formula is: ， Wherein, is the filling value of the data, is the collected data, is the observed data, is the random error term; S13. Check whether the data is within the medically reasonable range, check the logical relationship between the data, and score the data consistency. The formula is: ， is the consistency evaluation score value, is the consistent data item, is the total number of comparison data items; S14. Optimize the data features through feature engineering to improve the model performance; S15. Perform entity matching based on the patient's unique identifier, and use weighted average to fuse multi-source data. The formula is: ， Wherein, is the fused data, is the data source weight, is the value of the sample data point 3. The assessment method for potential complications in cardiology department according to claim 2, characterized in that, In step S14, the optimization of data features by feature engineering includes the following steps: S141. Calculate the risk factors of the patient's cardiovascular complications and judge the risk probability of complications. The formula is: ， wherein, is the complication risk factor, is the total cholesterol, is the high-density lipoprotein cholesterol, is the triglyceride, is the age; S142. Select features using the importance score of random forest. The formula is: ， For feature selection importance scoring, For features In the importance in the th tree, where S143. Evaluate whether to retain features through feature variance, and perform dimensionality reduction processing on the features. The formula is: ， In the formula, is the value of the th sample data point, is the number of samples, is the sample mean, is the variance that measures the dispersion of the sample data.

4. The evaluation method for potential complications in cardiology department according to claim 1, characterized in that, In step S2, the identification of the types of complications existing in the department of cardiology includes the following steps: S21. Construct an association matrix of complications and risk characteristics, and calculate the association degree score of each complication of the patient. The formula is: ， Wherein, is the correlation score of the complication, is the quantification value of the th risk feature, is the importance weight of the th risk feature, is the total number of risk features; S22. Integrate the evidence of clinical characteristics, laboratory indicators, and treatment history, calculate the confidence level, and perform multi-modal evidence fusion. The formula is: ， In the formula, is the confidence level of the complication , is the basic probability assignment of the evidence to the proposition ; S23. Based on the association degree score and confidence level of the complications, calculate the comprehensive priority, and rank the priority of the complication types; ， wherein, is the comprehensive priority score, and are adjustable parameters; S24. Set the risk threshold, and according to multiple complications, output a list of risk priorities, sorted in descending order of the comprehensive priority score. The formula is: ， In the formula, is the set risk threshold; S25. Update the risk feature weights according to the real-time monitoring data. The formula is: ， wherein, is the feature weight at time ; is the change amount of the feature ; is the influence coefficient of the feature on the complication; is the learning rate.

5. The assessment method for potential complications in cardiology department according to claim 1, characterized in that, In step S3, the risk factor network includes the following steps: S31. Use an independent sample test to calculate the significance of the mean difference between data groups. At the same time, through advantage comparison, analyze the single-factor correlation. The formula is: ， In the formula, is the test statistic for measuring the difference in the means of two sets of continuous variables, and are the means of the two sets of samples respectively, is the statistic obtained by synthesizing the dispersion degrees of the two sets of sample data, and are the numbers of the two sets of samples respectively, is the index reflecting the association strength between the exposure factor and the disease, and are the numbers of individuals exposed to and not exposed to a certain factor in the case group respectively, and are the numbers of individuals exposed to and not exposed to a certain factor in the control group respectively; S32. Screen independent risk factors, calculate the regression coefficient and risk ratio, and judge the significance of the interaction effect by introducing interaction terms and using the likelihood ratio test. ， In the formula, is the probability of complication occurrence, is the risk factor, is the regression coefficient, After the probability of complication occurrence is logarithmically transformed, the probability value is mapped to the real number domain, is the intercept term, is to measure each risk factor on the log odds The degree and direction of the influence, is the chi-square statistic for testing whether the interaction term is significant, is the likelihood value reflecting the goodness of fit of the model to the data under the assumption of no interaction, is the likelihood value reflecting the goodness of fit of the model to the data when the interaction is included; S33. Compress the coefficients of irrelevant variables to achieve feature sparsity, focus on the key risk factors of potential complications in cardiology, improve the model performance, quantify the feature importance, screen the key features to provide a basis for disease assessment and intervention. The formula is as follows: ， In the formula, is to find the regression coefficients that can minimize the objective function value to optimize the model. is the total number of patient samples participating in the evaluation and analysis. and are respectively the probabilities of the th patient actually and predicted to have potential complications. is the regularization parameter. is the total number of risk factors included in the evaluation system. is the th regression coefficient corresponding to the th risk factor. is the importance score of the th risk factor for predicting potential complications in cardiology. is the th total number of patient samples used by the and are respectively the coefficients of the node before splitting and the node after splitting along the th risk factor. Coefficient; S34. Calculate the conditional probability between risk factors and complications, and construct a directed acyclic graph. The formula is as follows: ， Wherein, is the conditional probability of a cardiovascular complication occurring under the condition that a known risk factor occurs, is the joint probability of the risk factor and the cardiovascular complication occurring simultaneously, is the conditional probability of a cardiovascular complication occurring, is the joint probability of the risk factors occurring simultaneously.

6. The evaluation method for potential complications in cardiology department according to claim 1, characterized in that, In the step S4, calculating the risk index of a patient's specific complication includes the following steps: S41. Draw a causal graph of risk factors and complications, and identify confounding variables. The formula is as follows: ， In the formula, is the risk factor effect value, is the complication, is the target factor, is the set of confounding variables; S42. For indicators that change over time, extract time-series features, analyze the recent fluctuation coefficient and trend slope, and evaluate the time-dependent risk after the patient's surgery or medication. The formula is as follows: ， In the formula, is the short-term fluctuation coefficient for measuring the short-term fluctuation degree of complications, and and are respectively the maximum value, minimum value and mean value of the complication occurrence probability within one week, is the trend slope reflecting the trend of complication change over time, and are respectively the th time point and the time mean value, and are respectively the complication occurrence probability value corresponding to the th time point and the mean value of the complication occurrence probability of the patient within one week, is the risk value of the patient after surgery or medication, is the risk coefficient, is the number of days after surgery or medication, is the weight coefficient, is the complication history index; S43. Train the model by integrating data from multiple hospitals, and establish a longitudinal federated learning framework. The formula is as follows: ， Wherein, is the global model, and are respectively the model parameters of the th participating hospital in the and round of trainers, is the learning rate, is the loss function with respect to the model parameter gradient; S44. Adjust the patient-specific parameters, and then adjust the global model coefficients to improve the model prediction accuracy. The formula is as follows: ， The nth model coefficient adjusted for the patient's personalization, is the nth coefficient in the global model, is the nth personalized offset, and is the biomarker ratio; S45. Integrate the machine learning score and the clinical guideline score, and obtain the comprehensive risk score through dynamic parameter adjustment. The formula is as follows: ， is the comprehensive risk score, is the fusion weight parameter, is the machine learning score, is the clinical guideline score.

7. The assessment method of potential complications in cardiology department according to claim 6, characterized in that, In the step S5, generating a prevention plan includes the following steps: S51. Divide the risk levels according to the comprehensive risk score, and clarify the very low, low, medium, high, and extremely high risk intervals and corresponding thresholds; S52. Establish a prevention measure library including drug treatment, lifestyle intervention, and surgical treatment categories, and code the applicable scenarios and contraindications according to the guideline standards; S53. Exclude the contraindicated plans and generate a quantitative recommendation plan; S54. According to the patient-specific indicators, adjust the drug dosage or measure intensity to ensure individualized and safe medication, and dynamically attenuate or enhance the prevention measures in combination with the treatment time to adapt to the changes in the condition; S55. Check the compliance of the plan against the latest clinical guidelines to ensure that key measures are not omitted or in conflict, and generate a visual explanation report to present the recommendation basis in structured language.

8. An evaluation system for potential complications in cardiology, applied to an evaluation method for potential complications in cardiology according to any one of claims 1-7, characterized in that, Including: A data integration module, which is used for the system data entry, integrating the functions of patient information collection, data verification, and data preprocessing; A complication type identification module, which classifies the possible complication types based on the patient's underlying diseases, treatment methods, and the structured data output by the data integration module; A risk factor analysis module, which constructs a risk factor network based on the target complications output by the complication type identification module and the relevant risk factors extracted from the structured data of the data integration module; A multi-dimensional risk assessment module, which calculates the risk index of a patient's occurrence of a specific complication by using a multi-dimensional assessment model based on the results of the risk factor analysis module; A prevention measure generation module, which matches the corresponding general prevention measures from the preset prevention measure knowledge base based on the risk score and complication type output by the multi-dimensional risk assessment module, and generates a personalized prevention plan; A dynamic monitoring module, which monitors the patient in real time, compares the monitoring data with the baseline data, identifies abnormal trends, and triggers an early warning mechanism; An effect feedback module, which is used to track the implementation effect of prevention measures and optimize the risk assessment model and prevention measure library.

Images