Development and Validation of a Predictive Model for Radiation Dermatitis Risk in Cancer Patients During Radiotherapy

AbstractObjective To develop a risk prediction model for radiation dermatitis in cancer patients during radiotherapy, validate its predictive performance, and provide a theoretical basis for reducing the incidence of radiation dermatitis in clinical practice. Methods A prospective study design was adopted. Using convenience sampling, 296 cancer patients receiving radiotherapy at Fudan Tumor Minhang Branch from January 2023 to January 2025 were selected as the modeling cohort. Based on the occurrence of radiation dermatitis, they were divided into a dermatitis group (n = 150) and a non-dermatitis group (n = 146). Data from the two groups were compared, and a risk prediction model was constructed based on logistic regression analysis. The predictive performance of the model was evaluated using the area under the receiver operating characteristic (ROC) curve. From February to June 2025, 126 cancer patients undergoing radiotherapy were selected as a validation cohort for external validation of the model. Results In the modeling cohort of 296 patients, the mean age of the dermatitis group was 61.45 ± 9.67 years, and that of the non-dermatitis group was 57.06 ± 10.78 years. The incidence of radiation dermatitis was 50.67% (150/296) in the modeling cohort and 50% (63/126) in the validation cohort. Six predictors were finally included in the model: age (OR = 1.034, 95% CI: 1.008–1.061), total radiation dose (OR = 1.037, 95% CI: 1.004–1.071), body mass index (BMI) (OR = 7.123, 95% CI: 2.498–20.310), serum retinol (OR = 1.035, 95% CI: 1.016–1.055), diabetes comorbidity (OR = 0.357, 95% CI: 0.144–0.884), and serum albumin (OR = 0.248, 95% CI: 0.073–0.846) (all P < 0.05). The area under the ROC curve of the prediction model was 0.781, with a sensitivity of 0.79, specificity of 0.76, and Youden’s index of 0.55. External validation showed an area under the ROC curve of 0.768, sensitivity of 0.82, specificity of 0.68, and an accuracy of 87.3%, indicating good predictive performance. Conclusion The risk prediction model developed in this study demonstrates good predictive ability for the incidence of radiation dermatitis during radiotherapy in cancer patients. It can serve as a valuable tool for healthcare providers in designing and implementing targeted interventions.