Abstract
Novel coronavirus pneumonia (COVID-19) poses a major threat to human health as a global public health problem. Currently, the morbidity and mortality rate of myocardial injury in COVID-19 patients is as high as 59.6%; however, clinical prediction models for myocardial injury in COVID-19 patients are not well developed. This study used a retrospective analysis to include 1737 COVID-19 patients who attended Thousand Buddha Mountain Hospital in Shandong Province from December 2022 to December 2023. Data collection was performed through a medical big data system, and the patients were randomly divided into a training group (1216 cases) and a validation group (521 cases). In this study, 1-factor logistic regression, optimal subset regression, and least absolute shrinkage and selection operator regression were used to screen risk factors for myocardial infarction, and a prediction model was constructed based on the results of multifactor logistic regression. The predictive efficacy and clinical utility of the model were further evaluated using area under the receiver operating characteristic curve, calibration curve, and decision curve analysis. (1) Predictor variables screened by one-way logistic regression, optimal subset regression, and least absolute shrinkage and selection operator regression were included in multifactorial logistic regression, respectively, and the results all showed that age, history of alcohol consumption, diastolic blood pressure, heart rate, body mass index, and cystatin C were important risk factors affecting the occurrence of myocardial injury in patients with new crowns. (2) receiver operating characteristic curves were drawn based on the risk factors screened and the results showed that the area under the curve for the prediction set was 0.78 (0.75-0.81). (3) The calibration curves show that the model has good accuracy, with a mean error of 0.02 for both the training set as well as the validation set models. In this study, a myocardial injury prediction model for COVID-19 patients based on clinical parameters was successfully constructed used age, history of alcohol consumption, diastolic blood pressure, heart rate, body mass index, and cystatin C.