To develop a novel predictive model based on commonly used emergency indicators for assessing the risk of in-hospital mortality among ST-segment elevation myocardial infarction (STEMI) patients who underwent percutaneous coronary intervention (PCI) at the emergency department.

A retrospective analysis was conducted based on the clinical data of 842 STEMI patients who received PCI treatment (from Dec. 26, 2020 to Dec. 25, 2021) at the emergency departments of Yonsei University College of Medicine and the Fourth People's Hospital of Shenzhen (From Jan. 1, 2021 to Dec. 31, 2021). 503 cases from Yonsei University were used as the training set, 251 cases as the internal validation set, and 88 cases from the Fourth People's Hospital of Shenzhen as the external validation set. LASSO regression and unconditional logistic stepwise regression were employed to screen predictive variables, to establish a prediction model, and to present it in the form of nomograms. The model's discrimination, calibration, and clinical applicability were evaluated using ROC curves, calibration curves, and clinical decision curves, respectively, followed by internal and external validation.

A total of 842 STEMI patients after PCI were included, with 40 cases of in-hospital deaths. Using LASSO regression and unconditional logistic stepwise regression methods, seven predictive factors were identified: left ventricular ejection fraction, systolic blood pressure, high-density lipoprotein cholesterol, creatine kinase isoenzyme, brain natriuretic peptide, age, and heart rate. A prediction model of in-hospital mortality after PCI in STEMI patients on basis of these factors were constructed as follows: logit (P)=-2.85810-0.10535×left ventricular ejection fraction+0.00551×creatine kinase isoenzyme-0.00002×brain natriuretic peptide+0.07765×age-0.01181×systolic blood pressure-0.03742×high-density lipoprotein cholesterol+ 0.01314×heart rate. The model demonstrated good predictive performance with ROC curve areas (AUC) of 0.88,0.83 for internal and external validation sets, respectively. Calibration curves indicated good agreement between predicted and observed in-hospital mortality values (HL test, P=0.648), and clinical decision curves showed that the model had good calibration and clinical applicability.

Based on easily accessible clinical indicators from the emergency department, the novel in-hospital mortality prediction model for STEMI patients after PCI demonstrates good discrimination, calibration, and clinical applicability.