2.4. Development of a Risk Prediction Model Based on DNA Repair Genes

80 UM patients from TCGA were randomly divided into a training group (n = 56) and an internal validation group (n = 24). In the training group, we selected 6 genes out of 52 DNA repair genes to construct a risk model through lasso regression. Patients were divided into the high risk group and low risk group by the median risk score. Then, we evaluated the risk model in the training group by survival analysis and 1-, 2-, and 3-year ROC curve. At the same time, we conducted validation in the internal validation group and the entire data set to validate the predictive effect of the model.