30 days mortality prediction and risk factor analysis of Asian patients with ACS using interpretable machine learning algorithm

Abstract Background Thrombolysis in Myocardial Infarction (TIMI) is used to predict the mortality rate patients with acute coronary syndrome (ACS). TIMI was developed limited data on Asian cohort and based Western cohort. STEMI NSTEMI have separate scores. There has been research ACS using interpretable machine learning (ML) algorithms. Purpose To construct a single 30-day risk scoring system, as well identify analyse factors ASIAN ACS, that applicable both patients, an ML algorithm. Methods The National Cardiovascular Disease Database registry of 9054 used. 70% for algorithm development, remaining 30% validation Fifty-four parameters were considered, demographics, cardiovascular risk, medications, clinical variables. provide better guidance advice judgement, gradient boosting (XGBoost) classification analysis SHapley Additive exPlanation (SHAP) value graphs Each indicator's SHAP indicates impact model output (mortality) calculated XGBoost model. performance evaluation metric area under curve (AUC). validated dataset compared conventional score NSTEMI. Results top ten predictors from for; (AUC = 0.8534, 95% CI: 0.8226–0.8842, Accuracy: 0.8053, Sensitivity: 0.73125, Specificity: 0.81355) 0.8145, 0.77–0.8589, 0.7972, 0.64356, 0.81232) outperformed (STEMI AUC 0.785, 0.543). Killip class, age, heart rate, fasting blood glucose, ACEI, creatine kinase, systolic pressure, HDLC, cardiac catheterization, oralhypogly are chosen by feature selection ascending order. Cardiac catheterization pharmacotherapy drugs selected improve prediction TIMI. variable names displayed y-axis order importance. average shown next them. x-axis. colour represents feature, ranging small large, allowing comprehension distribution values each (Figure 1). We can see having high killip class being older linked lower survival patients. procedures, use ACEI OHA, patient 2). Conclusions A would classify than TIMI, which requires two distinct In population, be Funding Acknowledgement Type funding sources: Public grant(s) – budget only. Main source(s): Technology Development Fund 1