Construction and validation of an in-hospital cardiogenic shock prediction model for AMI patients based on machine learning

Objective To develop and validate a machine learning (ML) model for predicting in-hospital cardiogenic shock (CS) in acute myocardial infarction (AMI) patients undergoing percutaneous coronary intervention. Methods We included AMI patients who admitted to the First Hospital of Lanzhou University. Key predictive variables were screened using LASSO regression, and six ML predictive models were constructed. Model performance was evaluated using accuracy, sensitivity, precision, F1 score, AUROC, and AUPRC, and an online predictive application was developed. Results 851 patients were included, with 43 (5.05%) was CS. They were randomly divided into a training set (n = 595) and a test set (n = 256) at a ratio of 7:3. LASSO regression initially screened out 14 relevant variables. The LightGBM model performed optimally. After GSCV optimization, eight core modeling variables were finally determined: heart rate, D-dimer, Killip classification, creatinine, lipoprotein a, alanine aminotransferase, diastolic blood pressure, and apolipoprotein B/A. We established an ML prediction model based on LightGBM + LASSO + GSCV, Its AUROC = 0.87 (95% CI: 0.77–0.96) and AUPRC = 0.64 (95% CI: 0.36–0.86). We developed an online risk prediction application based on the Streamlit framework. Conclusion This study developed and validated a risk prediction model for in-hospital CS in AMI patients, providing a reliable risk assessment tool for clinical decision-making.