Method for researching similarities and differences of left heart failure and right heart failure in cardiopulmonary exercise test

Abstract

The invention discloses a method for researching similarities and differences of left heart failure and right heart failure in a cardiopulmonary exercise test, and relates to the technical field of data analysis. Comprising the following steps: S1, selecting and grouping research objects: selecting heart failure patients with stable clinical symptoms, and grouping according to the level II-IV of Newlink heart functions; dividing the patients into a left heart failure group and a right heart failure group according to clinical diagnosis results of the patients; s2, performing CPET operation standardization: performing quality control on the equipment, including environment control and equipment regular calibration; a subject is firstly subjected to static lung function examination, and then is tested through the testing device. According to the method, a multivariate statistical method and a machine learning algorithm are introduced and improved, so that the precision, reliability and clinical applicability of CPET data analysis are remarkably improved; compared with a traditional method, the method not only can more deeply excavate and understand the complex physiological characteristics of the heart failure patient shown in the CPET, but also can provide a more accurate prediction model.

Description

Technical Field

[0001] The invention relates to the technical field of data analysis, and in particular to a method for studying the similarities and differences between left heart failure and right heart failure in cardiopulmonary exercise testing. Background Art

[0002] In studies exploring the use of cardiopulmonary exercise testing (CPET) in patients with heart failure, traditional approaches often focus on a single physiological parameter or simple statistical analysis, which limits a comprehensive understanding of the complexity of the disease.

[0003] After searching, the application scheme of Chinese patent application number CN202410268819.X discloses a multi-parameter cardiopulmonary function test equipment and its data processing method, which belongs to the field of electrical digital data processing technology, wherein the method comprises: step 1: obtaining the test mode of the cardiopulmonary function test equipment; step 2: obtaining the first acquisition parameter according to the test mode based on the cardiopulmonary exercise test scheme; step 3: obtaining the regression data model; step 4: obtaining the second acquisition parameter according to the analysis requirements based on the regression data model; step 5: performing data processing based on the parameter type of the second acquisition parameter. The data processing method in the above patent has the following deficiencies: the accuracy of data analysis is poor, and the universality of the data processing strategy is still insufficient. Summary of the invention

[0004] The purpose of the present invention is to solve the shortcomings of the prior art and to propose a method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A method to study the similarities and differences between left heart failure and right heart failure during cardiopulmonary exercise testing includes:

[0007] S1: Selection and grouping of research subjects: Patients with heart failure with stable clinical symptoms were selected and enrolled according to New York Heart Function Classification II to IV; according to the clinical diagnosis results of the patients, the patients were divided into left heart failure group and right heart failure group;

[0008] S2: Standardization of CPET operation: Quality control of the equipment, including environmental control and regular calibration of the equipment; the subjects first undergo a static lung function test, followed by testing with a test device, with respiratory, ECG, and blood pressure vital signs monitored throughout the process;

[0009] S3: Data collection and analysis: Based on the CPET indicators, continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes and detailed records of gas exchange rate were carried out; statistical software was used to process the data; partial correlation analysis was introduced to evaluate the correlation between different variables;

[0010] S4: Result analysis: Comparative analysis of the differences in CPET performance between patients with left heart failure and right heart failure.

[0011] Preferably: in S1, the selection and grouping of research subjects include:

[0012] S11: Screening criteria formulation: Inclusion and exclusion criteria were formulated based on medical history, clinical symptoms and preliminary examination results;

[0013] S12: Clinical evaluation: Clinical evaluation of patients who initially meet the screening criteria;

[0014] S13: Grouping: Patients were divided into corresponding study groups according to the clinical evaluation results;

[0015] S14: Baseline characteristics record: Record the patient's baseline characteristics in detail;

[0016] In S2, CPET operation standardization includes:

[0017] S21: Static pulmonary function test: The subjects were tested for vital capacity, slow vital capacity, and maximum minute ventilation in a sitting position;

[0018] S22: Dynamic load test: set the parameters of the test device according to the actual situation of the patient;

[0019] S23: Full-time monitoring and recording: Starting from the resting state, continuously monitor and record the patient's heart rate, blood pressure, respiratory rate, and blood oxygen saturation indicators until the end of the recovery period.

[0020] Preferably: in S3, data collection and analysis includes:

[0021] S31: Data collection: CPET index collection, and continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes and gas exchange rate indicators;

[0022] S32: Statistical analysis: Statistical software was used to process the data, and independent sample t-test, variance analysis, and partial correlation analysis were used to analyze the relationship between the data;

[0023] S33: Subgroup analysis: Subgroup analysis was performed based on the baseline characteristics of the patients to explore the differences in CPET indicators among different subgroups and their potential influencing factors;

[0024] S34: Sensitivity analysis: Monte Carlo simulations were used to assess the confidence intervals and sensitivity of statistics.

[0025] Preferably: in said S32, the statistical analysis comprises:

[0026] Use multivariate analysis of variance to explore the relationship between multiple response variables and grouping variables;

[0027] Based on the random forest algorithm, feature selection and importance scoring were performed to identify the most important parameters for classification of left and right heart failure in CPET;

[0028] Regression analysis was used to evaluate the predictive ability of CPET parameters for HF type.

[0029] Preferably: in said S33, subgroup analysis comprises:

[0030] According to the patient's age, gender, and disease severity, a clustering algorithm was used to divide the patients into subgroups based on CPET parameters;

[0031] Comparative analyses of CPET parameters were performed within each subgroup.

[0032] Preferably, in S32, multivariate variance analysis is used to explore the relationship between multiple response variables and grouping variables, as follows:

[0033] Preparation phase: Collect CPET data, including continuous variables such as heart rate and blood pressure, and record the grouping information of each patient;

[0034] Data preprocessing: handling missing values ​​and performing variable transformation;

[0035] Perform MANOVA: Set heart rate and blood pressure as dependent variables and grouping variables as independent variables in the statistical software for multivariate analysis of variance;

[0036] Result evaluation: Wilks' lambda value, F value and significance level were analyzed to evaluate the inter-group differences and the correlation between response variables.

[0037] Preferably, in S32, regression analysis is used to evaluate the predictive ability of CPET parameters for the type of heart failure, as follows:

[0038] Select variables: Select CPET parameters that affect the type of heart failure;

[0039] Model establishment: A logistic regression model was constructed with the type of heart failure as the dependent variable and CPET parameters as the independent variables;

[0040] Model evaluation: The explanatory power and predictive ability of the model were evaluated through goodness-of-fit indices, prediction accuracy, and likelihood ratio tests;

[0041] Model optimization: Select the optimal feature subset through stepwise regression method or based on information criteria.

[0042] Preferably: In S32, feature selection and importance scoring are performed based on a random forest algorithm, as follows:

[0043] Data preparation: All parameters of CPET are used as features, and the classification label of heart failure is used as the target variable;

[0044] Divide the data set: divide the data into training set and test set;

[0045] Model training: Use the random forest algorithm to train the model;

[0046] Feature importance evaluation: rank the importance of each feature by the feature importance score provided by the algorithm;

[0047] Feature selection: Remove features with low importance scores and select features that contribute more to model prediction.

[0048] Preferably, in S33, a decision tree is applied to discover key distinguishing features within the subgroup, as follows:

[0049] Determine subgroups: Divide patients into subgroups based on clinical characteristics or CPET results;

[0050] Data organization: For each subgroup, organize CPET parameters as input for decision tree or GBM models;

[0051] Model construction: Build a decision tree model or GBM model, and adjust the tree depth and learning rate to optimize performance;

[0052] Identification of key features: Analyze model results to identify features with key discriminatory power for HF classification within subgroups.

[0053] Preferably, in S34, the confidence interval and sensitivity of the statistical data are evaluated using Monte Carlo simulation, as follows:

[0054] Determine the analysis objectives: identify the statistics needed to assess uncertainty;

[0055] Monte Carlo simulation: set the distribution under preset conditions, perform random sampling, and calculate the statistics for each simulation;

[0056] Bootstrap resampling: repeatedly sampling the original data set and calculating the statistics obtained from each sampling;

[0057] Construct confidence intervals: Calculate the confidence interval of the statistic based on simulation or resampling results, and take the 2.5% and 97.5% quantiles as the 95% confidence interval;

[0058] Sensitivity analysis: Assessing the sensitivity of a statistic to changes in assumptions by changing the assumptions of the simulation or resampling.

[0059] The beneficial effects of the present invention are:

[0060] 1. This invention significantly improves the accuracy, reliability and clinical applicability of CPET data analysis by introducing and improving multivariate statistical methods and machine learning algorithms.

[0061] 2. Compared with traditional methods, the present invention can not only more deeply explore and understand the complex physiological characteristics of heart failure patients in CPET, but also provide a more accurate prediction model.

[0062] 3. This invention further enhances the stability and credibility of the research results through sensitivity analysis and uncertainty assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 The present invention provides a flow chart of a method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing. DETAILED DESCRIPTION

[0064] The technical solution of the present invention is further described in detail below in conjunction with specific implementation methods.

[0065] Embodiment 1:

[0066] A method to study the similarities and differences between left heart failure and right heart failure during cardiopulmonary exercise testing includes:

[0067] S1: Selection and grouping of research subjects: Patients with heart failure with stable clinical symptoms were selected and enrolled according to New York Heart Function Classification II to IV; according to the clinical diagnosis results of the patients, the patients were divided into left heart failure group and right heart failure group;

[0068] S2: Standardization of CPET operation: quality control of equipment, including environmental control, regular calibration of equipment, etc.; subjects first undergo static lung function test, and then test with test equipment, such as incremental load test on power bicycle, and monitor vital signs such as breathing, electrocardiogram, blood pressure, etc. throughout the process;

[0069] S3: Data collection and analysis: Based on the CPET indicators, continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes and detailed records of gas exchange rate were carried out; statistical software, such as SPSS26.0, was used to process the data; partial correlation analysis was introduced to evaluate the correlation between different variables;

[0070] S4: Result analysis: Comparative analysis of the differences in CPET performance between patients with left heart failure and right heart failure.

[0071] Among them, in S1, the selection and grouping of research subjects include:

[0072] S11: Screening criteria formulation: Inclusion and exclusion criteria were formulated based on medical history, clinical symptoms and preliminary examination results;

[0073] S12: Clinical evaluation: Patients who initially meet the screening criteria will undergo clinical evaluation, including but not limited to electrocardiogram, echocardiogram and other examinations;

[0074] S13: Grouping: Patients were divided into corresponding study groups according to the clinical evaluation results;

[0075] S14: Baseline characteristics record: Record the patient's baseline characteristics in detail, including age, gender, weight, height, disease course, etc., to provide basic data for subsequent analysis.

[0076] Among them, in S2, CPET operation standardization includes:

[0077] S21: Static pulmonary function test: The subjects were required to undergo static pulmonary function tests such as vital capacity, slow vital capacity, and maximum minute ventilation in a sitting position;

[0078] S22: Dynamic load test: set the parameters of the test device according to the actual situation of the patient, such as the starting load and increasing rate of the power bicycle;

[0079] S23: Full-time monitoring and recording: Starting from the resting state, continuously monitor and record the patient's heart rate, blood pressure, respiratory rate, blood oxygen saturation and other indicators until the end of the recovery period.

[0080] Among them, in S3, data collection and analysis include:

[0081] S31: Data collection: CPET index collection, and continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes, gas exchange rate and other indicators;

[0082] S32: Statistical analysis: SPSS26.0 and other statistical software were used for data processing, and independent sample t-test, variance analysis, partial correlation analysis and other methods were used to analyze the relationship between data;

[0083] S33: Subgroup analysis: Subgroup analysis was performed based on the baseline characteristics of the patients to explore the differences in CPET indicators among different subgroups and their potential influencing factors;

[0084] S34: Sensitivity analysis: Monte Carlo simulation or bootstrap resampling is used to assess the confidence interval and sensitivity of statistics.

[0085] Wherein, in said S32, the statistical analysis includes:

[0086] Multivariate analysis of variance (MANOVA) was used to explore the relationship between multiple response variables (such as heart rate, blood pressure, etc.) and grouping variables (left heart failure and right heart failure);

[0087] Based on the random forest algorithm, feature selection and importance scoring were performed to identify the most important parameters for classification of left and right heart failure in CPET;

[0088] Regression analysis was used to evaluate the predictive ability of CPET parameters for heart failure type (left heart failure and right heart failure).

[0089] Wherein, in said S33, subgroup analysis includes:

[0090] According to factors such as age, gender, and severity of the disease, clustering algorithms such as K-means or hierarchical clustering are used to divide patients into subgroups based on CPET parameters;

[0091] Comparative analyses of CPET parameters were performed within each subgroup.

[0092] Among them, in S32, multivariate variance analysis is used to explore the relationship between multiple response variables and grouping variables, as follows:

[0093] Preparation phase: Collect CPET data, including continuous variables such as heart rate and blood pressure, and record the group information of each patient (left heart failure and right heart failure);

[0094] Data preprocessing: handling missing values ​​and performing variable transformation (such as logarithmic transformation of heart rate to meet analysis assumptions);

[0095] Perform MANOVA: Set heart rate, blood pressure, etc. as dependent variables and grouping variables (left heart failure and right heart failure) as independent variables in the statistical software for multivariate analysis of variance;

[0096] Result evaluation: Wilks' lambda value, F value and significance level were analyzed to evaluate the inter-group differences and the correlation between response variables.

[0097] Wherein, in S32, regression analysis is used to evaluate the predictive ability of CPET parameters for heart failure types (left heart failure and right heart failure), as follows:

[0098] Select variables: Select CPET parameters that affect the type of heart failure (such as peak oxygen uptake, heart rate recovery time, etc.);

[0099] Model establishment: A logistic regression model was constructed with the type of heart failure as the dependent variable and CPET parameters as the independent variables;

[0100] Model evaluation: Evaluate the explanatory power and predictive ability of the model through goodness-of-fit indicators (such as R-squared), prediction accuracy, likelihood ratio test, etc.

[0101] Model optimization: Select the optimal feature subset through stepwise regression method or based on information criteria (such as AIC, BIC).

[0102] In S32, feature selection and importance scoring are performed based on the random forest algorithm, as follows:

[0103] Data preparation: All parameters of CPET are used as features, and the classification labels of heart failure (left heart failure and right heart failure) are used as target variables;

[0104] Divide the data set: Divide the data into training set and test set, with a ratio of 70% training set and 30% test set;

[0105] Model training: Use the random forest algorithm to train the model;

[0106] Feature importance evaluation: rank the importance of each feature by the feature importance score provided by the algorithm;

[0107] Feature selection: Remove features with low importance scores and select features that contribute more to model prediction.

[0108] Among them, in S33, a decision tree is applied to discover key distinguishing features within the subgroup, as follows:

[0109] Determine subgroups: Divide patients into subgroups based on clinical characteristics or CPET results;

[0110] Data organization: For each subgroup, organize CPET parameters as input for decision tree or GBM models;

[0111] Model construction: Build a decision tree model or GBM model, and adjust hyperparameters such as tree depth and learning rate to optimize performance;

[0112] Identification of key features: Analyze model results to identify features with key discriminatory power for HF classification within subgroups.

[0113] Wherein, in said S34, the confidence interval and sensitivity of the statistical data are evaluated by Monte Carlo simulation, as follows:

[0114] Determine the analysis objective: Identify the statistic for which uncertainty needs to be assessed, such as the mean difference of a CPET parameter;

[0115] Monte Carlo simulation: set the distribution under preset conditions (such as normal distribution), perform random sampling, and calculate the statistics of each simulation;

[0116] Bootstrap resampling: repeatedly sampling the original data set (with replacement) and calculating the statistics obtained from each sampling;

[0117] Construct confidence intervals: Calculate the confidence interval of the statistic based on simulation or resampling results, usually taking the 2.5% and 97.5% quantiles as the 95% confidence interval;

[0118] Sensitivity analysis: By changing the assumptions of the simulation or resampling (such as changing the distribution assumptions), the sensitivity of the statistic to the changes in the assumptions is evaluated.

[0119] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing, characterized in that: include: S1: Selection and grouping of research subjects: Patients with heart failure and stable clinical symptoms were selected according to New York Heart Classification II to IV. According to the clinical diagnosis of patients, the patients were divided into left heart failure group and right heart failure group; S2: Standardization of CPET operation: Quality control of the equipment, including environmental control and regular calibration of the equipment; the subjects first undergo a static lung function test, followed by testing with a test device, with respiratory, ECG, and blood pressure vital signs monitored throughout the process; S3: Data collection and analysis: Based on the CPET indicators, continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes and detailed records of gas exchange rate were carried out; statistical software was used to process the data; partial correlation analysis was introduced to evaluate the correlation between different variables; S4: Result analysis: Comparative analysis of the differences in CPET performance between patients with left heart failure and right heart failure.

2. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing according to claim 1, characterized in that: In S1, the selection and grouping of research subjects include: S11: Screening criteria formulation: Inclusion and exclusion criteria were formulated based on medical history, clinical symptoms and preliminary examination results; S12: Clinical evaluation: Clinical evaluation of patients who initially meet the screening criteria; S13: Grouping: Patients were divided into corresponding study groups according to the clinical evaluation results; S14: Baseline characteristics record: Record the patient's baseline characteristics in detail; In S2, CPET operation standardization includes: S21: Static pulmonary function test: The subjects were tested for vital capacity, slow vital capacity, and maximum minute ventilation in a sitting position; S22: Dynamic load test: set the parameters of the test device according to the actual situation of the patient; S23: Full-time monitoring and recording: Starting from the resting state, continuously monitor and record the patient's heart rate, blood pressure, respiratory rate, and blood oxygen saturation indicators until the end of the recovery period.

3. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing according to claim 2, characterized in that: In S3, data collection and analysis include: S31: Data collection: CPET index collection, and continuous dynamic monitoring of blood oxygen saturation, electrocardiogram changes and gas exchange rate indicators; S32: Statistical analysis: Statistical software was used to process the data, and independent sample t-test, variance analysis, and partial correlation analysis were used to analyze the relationship between the data; S33: Subgroup analysis: Subgroup analysis was performed based on the baseline characteristics of the patients to explore the differences in CPET indicators among different subgroups and their potential influencing factors; S34: Sensitivity analysis: Monte Carlo simulations were used to assess the confidence intervals and sensitivity of statistics.

4. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing according to claim 3, characterized in that: In said S32, the statistical analysis includes: Use multivariate analysis of variance to explore the relationship between multiple response variables and grouping variables; Based on the random forest algorithm, feature selection and importance scoring were performed to identify the most important parameters for classification of left and right heart failure in CPET; Regression analysis was used to evaluate the predictive ability of CPET parameters for HF type.

5. The method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise test according to claim 3, characterized in that: In S33, subgroup analysis includes: According to the patient's age, gender, and disease severity, a clustering algorithm was used to divide the patients into subgroups based on CPET parameters; Comparative analyses of CPET parameters were performed within each subgroup.

6. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing according to claim 3, characterized in that: In S32, multivariate analysis of variance is used to explore the relationship between multiple response variables and grouping variables, as follows: Preparation phase: Collect CPET data, including continuous variables such as heart rate and blood pressure, and record the grouping information of each patient; Data preprocessing: handling missing values ​​and performing variable transformation; Perform MANOVA: Set heart rate and blood pressure as dependent variables and grouping variables as independent variables in the statistical software for multivariate analysis of variance; Result evaluation: Wilks' lambda value, F value and significance level were analyzed to evaluate the inter-group differences and the correlation between response variables.

7. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise testing according to claim 3, characterized in that: In S32, regression analysis is used to evaluate the predictive ability of CPET parameters for the type of heart failure, as follows: Select variables: Select CPET parameters that affect the type of heart failure; Model establishment: A logistic regression model was constructed with the type of heart failure as the dependent variable and CPET parameters as the independent variables; Model evaluation: The explanatory power and predictive ability of the model were evaluated through goodness-of-fit indices, prediction accuracy, and likelihood ratio tests; Model optimization: Select the optimal feature subset through stepwise regression method or based on information criteria.

8. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise test according to claim 3, characterized in that: In S32, feature selection and importance scoring are performed based on the random forest algorithm, as follows: Data preparation: All parameters of CPET are used as features, and the classification label of heart failure is used as the target variable; Divide the data set: divide the data into training set and test set; Model training: Use the random forest algorithm to train the model; Feature importance evaluation: rank the importance of each feature by the feature importance score provided by the algorithm; Feature selection: Remove features with low importance scores and select features that contribute more to model prediction.

9. The method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise test according to claim 3, characterized in that: In S33, a decision tree is applied to discover key distinguishing features within the subgroups, as follows: Determine subgroups: Divide patients into subgroups based on clinical characteristics or CPET results; Data organization: For each subgroup, organize CPET parameters as input for decision tree or GBM models; Model construction: Build a decision tree model or GBM model, and adjust the tree depth and learning rate to optimize performance; Identification of key features: Analyze model results to identify features with key discriminatory power for HF classification within subgroups.

10. A method for studying the differences and similarities between left heart failure and right heart failure in cardiopulmonary exercise test according to claim 3, characterized in that: In S34, Monte Carlo simulation is used to evaluate the confidence interval and sensitivity of the statistical data, as follows: Determine the analysis objectives: identify the statistics needed to assess uncertainty; Monte Carlo simulation: set the distribution under preset conditions, perform random sampling, and calculate the statistics for each simulation; Bootstrap resampling: repeatedly sampling the original data set and calculating the statistics obtained from each sampling; Construct confidence intervals: Calculate the confidence interval of the statistic based on simulation or resampling results, and take the 2.5% and 97.5% quantiles as the 95% confidence interval; Sensitivity analysis: Assessing the sensitivity of a statistic to changes in assumptions by changing the assumptions of the simulation or resampling.

Images