Clinical Risk Factor Prediction for Second Primary Skin Cancer: A Hospital-Based Cancer Registry Study

This study aimed to develop a risk-prediction model for second primary skin cancer (SPSC) survivors. We identified the clinical characteristics of SPSC and created awareness physicians screening high-risk patients among Using data from 1248 survivors extracted five registries, we benchmarked random forest algorithm against MLP, C4.5, AdaBoost, bagging algorithms several metrics. Additionally, in this study, leveraged synthetic minority over-sampling technique (SMOTE) issue imbalanced dataset, cost-sensitive learning risk assessment, SHAP analysis feature importance. The proposed outperformed other models, with an accuracy 90.2%, recall rate 95.2%, precision 86.6%, F1 value 90.7% category based on 10-fold cross-validation balanced dataset. Our results suggest that four features, i.e., age, stage, gender, involvement regional lymph nodes, which significantly affect output prediction model, need be considered next causal effect. In addition specific sites, these features allow further investigation secondary cancers