A nomogram based on the SVM machine learning model was constructed to predict the risk of COVID-19 severity. The model showed high accuracy and was validated using an external dataset. The study also explored the relationship between immunogenic cell death-related genes and COVID-19 severity.