Method and system for predicting new atrial fibrillation risk of ICD implantation patient
By constructing a multi-factor statistical analysis model, combining parameters such as gender, left atrial diameter and pulmonary artery systolic pressure, and establishing a mathematical correlation equation, the problem of insufficient prediction accuracy of new-onset atrial fibrillation after ICD implantation was solved, and a more accurate risk assessment was achieved.
Patent Information
- Application Number
- CN202510766654.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-06-10
- Publication Date
- 2025-09-26
AI Technical Summary
In the existing technology, the prediction factors for new-onset atrial fibrillation after ICD implantation in people at high risk of sudden cardiac death are single, the prediction accuracy is insufficient, and there is a lack of dedicated prediction tools for ICD implanted people.
A multivariate statistical analysis model was constructed, combining parameters such as gender, left atrial diameter, diabetes indicators, and pulmonary artery systolic pressure to establish a mathematical correlation equation, calculate the individual risk of new-onset atrial fibrillation, and provide a visual risk assessment.
The AUC value for predicting new-onset atrial fibrillation was increased to 0.793, with significantly improved sensitivity and specificity, providing a dedicated prediction tool for ICD implanted populations and improving prediction accuracy.
Smart Images
Figure CN120708864A_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of medical data analysis, and in particular relates to a method and a system for predicting the risk of new-onset atrial fibrillation in ICD implanted patients. Background Art
[0002] Atrial fibrillation (AF) is one of the most common arrhythmias in clinical practice, significantly increasing the risk of stroke, heart failure, and death. Early screening and intervention are crucial to improving prognosis. Traditional opportunistic screening may miss paroxysmal or asymptomatic AF. Clinical new-onset AF risk prediction models can help diagnose AF early, and early intervention can improve its clinical outcomes. For the general population, the main prediction models can be divided into two categories: traditional scoring systems and machine learning models. In traditional clinical scoring systems, such as the CHARGE-AF model, 12 parameters such as age, BMI, blood pressure, and diabetes are included. （1） There are also scoring systems such as CHA2DS2-VASc and HARMS2-AF models; machine learning models such as AI-ECG model, WARN deep learning model, multimodal fusion model, etc.
[0003] Previous research has shown that patients undergoing vascular implantable electronic device (CIED) surgery have numerous risk factors for new-onset atrial fibrillation, including baseline characteristics, cardiac structural indicators, implantation-related parameters, and laboratory markers. However, there are few studies on prediction models for new-onset atrial fibrillation after ICD implantation in patients at high risk of sudden cardiac death. Only a few risk predictors have been identified, such as left atrial volume index (LAVI). When LAVI ≥ 38.5 mL / m² is used as the cutoff value, the AUC is 0.71, with a sensitivity of 69.4% and a specificity of 56.7%. （2） .
[0004] The existing technical defects include the following two points: (1) The current predictive factors for new-onset atrial fibrillation after ICD implantation in high-risk populations of sudden cardiac death are single, with AUC values generally lower than 0.75, indicating insufficient prediction accuracy; (2) New-onset atrial fibrillation prediction models are mostly focused on cardiovascular implantable electronic devices (CIEDs), and there is a lack of dedicated prediction tools for ICD implanted populations.
[0005] Prior art literature: 1) Li Y , Li Q , Wang L ,et al.The mC2HEST Score for Incident AtrialFibrillation: MESA (Multi-Ethnic Study of Atherosclerosis)[J].Jacc: Advances,2025, 4(2).DOI:10.1016 / j.jacadv.2024.101521. 2) Kim BS , Chun KJ , Hwang JK ,et al.Predictors and long-term clinical outcomes of newly developed atrial fibrillation in patients with cardiac implantable electronic devices[J].Medicine, 2016, 95(28):e4181.DOI:10.1097 / MD.0000000000004181. Summary of the Invention
[0006] The present invention aims to solve the problem of constructing a multifactor model to predict the risk of new-onset atrial fibrillation after ICD implantation in people at high risk of sudden cardiac death. The AUC is 0.793 (95% CI: 0.705-0.855), which has higher sensitivity and specificity than traditional single-factor prediction. It is also a special prediction tool for ICD implanted people, providing a method and prediction system for predicting the risk of new-onset atrial fibrillation in ICD implanted patients.
[0007] The technical contents adopted are: In a first aspect, a method for predicting the risk of new-onset atrial fibrillation in an ICD implanted patient comprises the following steps: Data acquisition: clinical characteristic parameters of ICD implanted patients are collected, including at least gender, left atrial diameter, diabetes index, and pulmonary artery systolic pressure; Data preprocessing: Standardization of clinical characteristic parameters, including binarization, retention of continuous values, and coding of categorical variables; Model construction: Based on a multivariate statistical analysis model, a mathematical equation was established to associate the clinical characteristic parameters with the risk of new-onset atrial fibrillation. The equation included a weighted combination of gender, left atrial diameter, diabetes index, and pulmonary artery systolic pressure. Risk calculation: Calculate the risk probability of new-onset atrial fibrillation for an individual based on the mathematical correlation equation and output a risk value between 0 and 1; Result output: Generate visual risk assessment results.
[0008] Preferably, the left atrial structural parameter is the left atrial inner diameter, and the preset threshold is 34 mm; when the left atrial inner diameter is greater than the threshold, it is marked as 1, otherwise it is marked as 0.
[0009] Preferably, the presence of diabetes diagnosis is marked as 1, and otherwise marked as 0.
[0010] Preferably, the mathematical correlation equation is: X=a+b×sex+c×LAD+d×diabetes+e×PASA; Among them, the gender parameter is 1 for females and 0 for males; the coefficient e of PASP in its prediction equation is 0.040, the coefficient c of LAD ≥ 34 mm is 1.740, the coefficient b of females is 1.000, and the coefficient d of diabetes is 1.370.
[0011] Preferably, the visual risk assessment result is a nomogram, which converts each clinical characteristic parameter into a corresponding score, and associates the risk probability of new-onset atrial fibrillation through the total score.
[0012] Preferably, a model validation step is also included: the predictive performance of the multivariate statistical analysis model is verified by ROC curve, calibration curve and decision curve. When predicting the risk of new-onset atrial fibrillation in a population at high risk of sudden cardiac death after ICD implantation, when the Youden index on the ROC curve is the largest, the prediction probability cut-off value is 0.23, the accuracy is 76.1%, the sensitivity is 72.7%, and the specificity is 78.5%; AUC=0.793 (95%CI: 0.705-0.855).
[0013] In the second aspect, the present invention also provides a system for predicting the risk of new-onset atrial fibrillation in patients with ICD implants, comprising: a data acquisition module for obtaining clinical characteristic parameters of patients with ICD implants, the parameters including at least gender, left atrial structure parameters, basic metabolic disease indicators, and pulmonary artery pressure parameters; a data processing module for performing standardized preprocessing on the clinical characteristic parameters; a model calculation module for storing a multi-factor statistical analysis model and calculating the risk probability of new-onset atrial fibrillation based on the preprocessed parameters; and a result display module for outputting visual risk assessment results.
[0014] Preferably, the data acquisition module is connected to the hospital HIS system to automatically capture gender, left atrial diameter, diabetes diagnosis code and pulmonary artery systolic pressure data in the electronic medical record.
[0015] Preferably, the model calculation module integrates a risk equation calculator, supports real-time input of parameters and outputs a risk probability value in the range of 0-1.
[0016] In a third aspect, the present invention further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method for predicting the risk of new-onset atrial fibrillation in ICD implanted patients as described in the first aspect.
[0017] Compared with the prior art, the present invention has the following advantages: A multivariate model was constructed to predict the risk of new-onset atrial fibrillation after ICD implantation in people at high risk of sudden cardiac death. The AUC was 0.793, and the 95% CI was 0.705-0.855. Compared with the traditional univariate prediction, the model had higher sensitivity and specificity, and was a special prediction tool for people with ICD implantation. BRIEF DESCRIPTION OF THE DRAWINGS
[0018] Figure 1 The prediction model for new-onset atrial fibrillation in high-risk SCD patients after ICD implantation provided in this application; Figure 2 The ROC curve for the risk assessment of new-onset AF after ICD implantation in patients at high risk of sudden cardiac death provided in this application; Figure 3 This is the calibration curve of the new AF prediction model provided in this application; Figure 4 This is the DCA decision curve for the new-onset AF prediction model provided in this application. DETAILED DESCRIPTION
[0019] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be understood as limiting the present invention.
[0020] The following describes an embodiment of the present invention based on its overall structure.
[0021] Data collection: By connecting to the hospital's HIS system, the LAD (left atrial diameter, unit: mm), PASA (pulmonary artery systolic pressure, unit: mmHg), diabetes diagnosis code (ICD-10 E11), and gender fields are captured; parameter preprocessing: LAD is binarized (>34mm is marked as 1), and pulmonary artery systolic pressure retains the original continuous value; risk calculation: The risk equation calculator is called to output a 0-1 risk value; result display: The risk value is output on the doctor's workstation / pacemaker programmer.
[0022] Model construction principle 1. Variable screening and data preprocessing Data source and feature extraction: Based on the electronic medical records of ICD implanted patients, parameters including demographic characteristics (gender, age), cardiovascular indicators (left atrial diameter (LAD), pulmonary artery systolic pressure (PASP), underlying diseases (history of diabetes), and dynamic follow-up data (atrial fibrillation status) were extracted.
[0023] Data standardization: Continuous variable processing: LAD was binarized (>34 mm was marked as 1), and pulmonary artery systolic pressure retained its original continuous value.
[0024] Categorical variables were coded as follows: sex (female = 1, male = 0), diabetes status (present = 1, absent = 0), and LAD ≥ 34 mm (yes = 1, no = 0).
[0025] Missing value processing: Multiple imputation method was used to fill missing data to ensure data integrity.
[0026] 2. Univariate and multivariate analysis Univariate Logistic Regression Candidate variables: Based on clinical guidelines and literature review, 12 parameters were initially selected, including demographic characteristics (gender), cardiac structural indicators (LAD, LVEDD, LVEF), functional indicators (NYHA classification, PASP), biomarkers (NT-proBNP, UA), medication history (loop diuretics, digoxin, amiodarone) and underlying disease status (diabetes). Variables significantly associated with atrial fibrillation (gender, LAD ≥ 34 mm, diabetes, PASP) were screened out.
[0027] Multivariate Logistic Regression and Model Optimization Factors with P values < 0.05 in the multivariate logistic regression (see Table 1 ) were included in the final model.
[0028] 3. Model construction: A prediction equation was constructed by combining gender, left atrial diameter (LAD ≥ 34 mm), diabetes status, and pulmonary artery systolic pressure (PASA): X = -4.886 + (1.000 × gender) + (1.740 × LAD) + (1.370 × diabetes) + (0.040 × PASA) Risk probability P = e X / (1+e X ) Where: ①P=e x / (1+e x ), where e is a natural number. ②X = -4.886 + (1.000 × gender) + (1.740 × LAD) + (1.370 × diabetes) + (0.040 × pulmonary artery systolic pressure).
[0029] In the above equation, LAD is in mm, pulmonary artery systolic pressure is in mmHg, female is 1, male is 0; LAD>34 mm is 1, LAD≤34 mm is 0, diabetes is 1, otherwise it is 0.
[0030] 4. Model Validation and Performance Evaluation Nomogram construction: Based on the multivariate logistic regression coefficient, the four variables of gender, LAD ≥ 34 mm, diabetes, and pulmonary artery systolic pressure (PASP) were converted into a visual scoring system, such as Figure 1 The total score corresponds to the individualized risk probability of new-onset atrial fibrillation (AF).
[0031] Discriminability verification: ROC curve: The AUC of the model combining the four indicators is 0.793, 95%CI: 0.705-0.855, and the meaning of 95%CI is the 95% confidence interval. Figure 2 As shown, it is significantly higher than the prediction of a single indicator.
[0032] Optimal cutoff value: When the Youden index is maximum, the risk probability threshold (Cut-off=0.23) corresponds to a sensitivity of 72.7%, a specificity of 78.5%, and an accuracy of 76.1%, which is suitable for clinical risk stratification.
[0033] Calibration Verification: Bootstrap internal validation: After 1000 resamplings, the adjusted AUC was 0.79 (95% CI: 0.72-0.86), and the calibration slope = 0.89 (the ideal value is 1), indicating that the deviation between the model prediction value and the actual risk is controllable. Figure 3 shown.
[0034] Nagelkerke R²: 0.35, suggesting that the model can explain approximately 35% of the AF risk variation.
[0035] Hosmer-Lemeshow test: P=0.62, confirming that the predicted probability did not deviate significantly from the actual observed risk.
[0036] Clinical practicality verification: Decision curve analysis (DCA): When the threshold probability is 5%-85%, the net benefit of using this model to guide clinical intervention (such as enhanced monitoring or preventive treatment) is significantly higher than that of the "full intervention" or "no intervention" strategy, such as Figure 4 shown.
[0037] Implementation 2 Case 1: High-risk patient Patient A: female (1 point), LAD=38mm (1 point), diabetes (1 point), PASP=52mmHg (52 points).
[0038]
[0039] = 67.6% (high risk, requiring initiation of anticoagulation and intensive follow-up) Case 2: Medium-risk patients Patient B: male (0 points), LAD=36mm (1 point), no diabetes (0 points), PASP=42mmHg (42 points).
[0040]
[0041] =44.3% (medium risk, Holter monitoring recommended) Case 3: Low-risk patient Patient C: male (0 points), LAD=30mm (0 points), no diabetes (0 points), PASP=32mmHg (32 points).
[0042] (Low risk, routine follow-up is sufficient).
[0043] Table 1 Multivariate logistic regression analysis of influencing factors of new-onset AF in high-risk SCD population after ICD implantation
[0044] Note: LAD: left atrial diameter, LVEDD: left ventricular end-diastolic diameter, LVEF: left ventricular ejection fraction, PASA: pulmonary artery systolic pressure, NT-proBNP: N-terminal pro-B-type natriuretic peptide, UA: uric acid. * indicates that clinical parameters were independently associated with new-onset atrial fibrillation in univariate logistic regression, with statistically significant differences (P < 0.05).
[0045] The above-described embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them. Although the present application has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. These modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present application, and should all be included in the scope of protection of the present application.
Claims
1. A method for predicting the risk of new-onset atrial fibrillation in patients with ICD implantation, characterized in that: The following steps are involved: Data acquisition: clinical characteristic parameters of ICD implanted patients are collected, including at least gender, left atrial diameter, diabetes index, and pulmonary artery systolic pressure; Data preprocessing: Binarization of clinical characteristic parameters, standardization of continuous values, and coding of categorical variables; Model construction: Based on a multivariate statistical analysis model, a mathematical equation was established to associate the clinical characteristic parameters with the risk of new-onset atrial fibrillation. The equation included a weighted combination of gender, left atrial diameter, diabetes index, and pulmonary artery systolic pressure. Risk calculation: Calculate the risk probability of new-onset atrial fibrillation for an individual based on the mathematical correlation equation and output a risk value between 0 and 1; Result output: Generate visual risk assessment results.
2. The prediction method according to claim 1, characterized in that The left atrial structural parameter is the left atrial inner diameter, and the preset threshold is 34 mm; when the left atrial inner diameter is greater than the threshold, it is marked as 1, otherwise it is marked as 0.
3. The prediction method according to claim 1, wherein: The presence of diabetes diagnosis is marked as 1, otherwise it is marked as 0.
4. The prediction method according to claim 1, wherein: The equation for calculating the risk value is as follows: X = a + b × sex + c × LAD + d × diabetes + e × PASA; In the prediction equation, the coefficient e of PASP was 0.040, the coefficient c of LAD ≥ 34 mm was 1.740, the coefficient b of gender was 1.000, and the coefficient d of diabetes was 1.
370.
5. The prediction method according to claim 1, wherein: The visualized risk assessment result is a nomogram, which converts each clinical characteristic parameter into a corresponding score and associates the risk probability of new-onset atrial fibrillation with the total score.
6. The prediction method according to claim 1, characterized in that It also includes a model validation step: the predictive performance of the multivariate statistical analysis model is verified by ROC curves, calibration curves, and decision curves. When predicting the risk of new-onset atrial fibrillation in people at high risk of sudden cardiac death after ICD implantation, when the Youden index on the ROC curve is the largest, the prediction probability cut-off value is 0.23, the accuracy is 76.1%, the sensitivity is 72.7%, and the specificity is 78.5%; the AUC is 0.793, and the 95% CI is 0.705-0.
855.
7. A system for predicting the risk of new-onset atrial fibrillation in patients with ICD implants, characterized in that: include: A data acquisition module is used to obtain clinical characteristic parameters of ICD implanted patients, wherein the parameters include at least gender, left atrial structural parameters, basic metabolic disease indicators, and pulmonary artery pressure parameters; A data processing module, configured to perform standardized preprocessing on the clinical characteristic parameters; Model calculation module, which stores a multi-factor statistical analysis model and calculates the risk probability of new-onset atrial fibrillation based on preprocessed parameters; The result display module is used to output visual risk assessment results.
8. The prediction system according to claim 7, characterized in that The data acquisition module is connected to the hospital HIS system to automatically capture gender, left atrial diameter, diabetes diagnosis code and pulmonary artery systolic pressure data in the electronic medical record.
9. The prediction system according to claim 7, characterized in that The model calculation module integrates a risk equation calculator, supports real-time parameter input and outputs a risk probability value in the range of 0-1.
10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method for predicting the risk of new-onset atrial fibrillation in an ICD implanted patient as described in any one of claims 1 to 6 is implemented.