ObjectiveTo systematically evaluate risk prediction models for ventricular arrhythmia (VA) following percutaneous coronary intervention (PCI) in patients with acute myocardial infarction (AMI), aiming to provide references for the development, optimization, and application of the models. MethodsDatabases including CNKI, Wanfang, VIP, Chinese Biomedical Literature Database, PubMed, Embase, and Cochrane Library were searched for studies on VA prediction models after PCI in AMI patients from inception to September 2025. Two researchers independently screened the literature, extracted data, and assessed the quality of included studies using the prediction model risk of bias assessment tool. Meta-analysis of common predictors was performed using Stata 18.0 software, and the area under the curve (AUC) of the models was statistically analyzed using MedCalc software. ResultsA total of 12 studies were included, establishing 12 models involving 3411 patients. The incidence of VA ranged from 11.0% to 50.8%, with an overall incidence of approximately 24.5%. The AUC values of the 12 models ranged from 0.717 to 0.983, indicating good predictive performance. However, the overall risk of bias in the included studies was high. Statistical analysis yielded a pooled AUC of 0.853 [95%CI (0.807, 0.899)]. Meta-analysis results showed that Killip class, left ventricular ejection fraction, thrombolysis in myocardial infarction flow grade, number of diseased coronary vessels, troponin levels, diabetes mellitus, J-wave on electrocardiogram, and serum potassium level were independent predictive factors for VA after PCI in AMI patients (P<0.05). ConclusionThe risk prediction models for VA after PCI in AMI patients demonstrate good overall discrimination. However, existing studies generally suffer from a high risk of bias, and the calibration and external validation of the models are severely insufficient, limiting their direct clinical applicability. Future multicenter, large-sample, prospective studies are needed to optimize study design and reporting processes, aiming to develop and validate more robust prediction models suitable for clinical practice, facilitating early identification and prevention of VA after PCI in AMI patients.