Abstract
OBJECTIVE: The objective of this study was to create and validate a clinical prediction model for the incidence of major adverse cardiovascular events (MACE) within one year after percutaneous coronary intervention (PCI) in elderly patients diagnosed with acute coronary syndromes (ACS). METHODS: The study will use 70% of the 738 patients for model training and the remaining 30% for model validation. The feature recursive elimination algorithm (RFE) and the least absolute shrinkage selection operator (LASSO) regression technique will be used to identify the best combination of features. We compare the clinical prediction model we constructed with GRACE in terms of discrimination, calibration, recall, and clinical impact. RESULTS: We used the RFE and LASSO regression technique to select 8 key variables from 44 candidates for our predictive model. The predictive model was found to have a good fit based on the Hosmer-Lemeshow test results (χ (2) = 6.245). Additionally, the Brier score of the clinical prediction model was 0.1502, confirming its accuracy. When comparing our clinical prediction model to the widely used GRACE scoring system, the results showed that our model had slightly better predictive efficacy for the dataset involved in this study. The NRI was 0.6166, NRI + was 0.2262, NRI- was 0.3904, and IDI was 0.1272, with a P value of <0.001. The validation set's AUC was 0.787, indicating the prediction model has high differentiation and discriminative ability. CONCLUSION: This model assists in the early identification of the risk of MACE within one year after PCI for ACS in elderly patients.