Establishment and validation of a risk prediction model for malignant ventricular arrhythmia in acute myocardial...

To develop and validate a nomogram for predicting malignant ventricular arrhythmias (MVA) risk in acute myocardial infarction (AMI), utilizing parameters derived from 24-hour dynamic electrocardiogram (Holter) monitoring.

We enrolled 279 AMI patients with MVA diagnosed via Holter monitoring. Key parameters from 24-hour Holter monitoring, such as heart rate variability (HRV) and QT interval variability (QTV), were extracted for subsequent analysis. These parameters were subsequently incorporated with clinical indicators to develop a risk prediction model. Electrocardiographic parameters and clinical indicators associated with MVA in the multivariable logistic regression analysis were used to construct a predictive nomogram for risk visualization. The nomogram model underwent internal and external validation for discrimination, calibration, and clinical utility. SHAP was used for model interpretation.

Of the 279 AMI patients, 37 cases (13.3%) developed MVA. Multivariate analysis showed that SDNN, 24h-QTV, night-QTV, day-QTV, TnI and Killip classification were independent predictors. The AUC of the model was 0.95 (0.92-0.98). The nomogram demonstrated good calibration, with predicted probabilities aligning well with actual outcomes (Hosmer-Lemeshow test, P=0.781). Clinical net benefit of the model was observed over a wide threshold probability range of 0.10 to 0.80 in the decision curve analysis.

The nomogram based on 24-hour Holter parameters can effectively identify a risk of MVA in AMI patients during hospitalization and provide an objective basis for clinical decision-making.