The established of a machine learning model for predicting the efficacy of adjuvant interferon alpha1b in patients with advanced melanoma.

Background: Interferon-alpha1b (IFN-α1b) has shown remarkable therapeutic potential as adjuvant therapy for melanoma. This study aimed to develop five machine learning models to evaluate the efficacy of postoperative IFN-α1b in patients with advanced melanoma. Methods: We retrospectively analyzed 113 patients with the American Joint Committee on Cancer (AJCC) stage III-IV melanoma who received postoperative IFN-α1b therapy between July 2009 and February 2024. Recurrence-free survival (RFS) and overall survival (OS) were assessed using Kaplan-Meier analysis. Five machine learning models (Decision Tree, Cox Proportional Hazards, Random Forest, Support Vector Machine, and LASSO regression) were developed and compared for their capacity to predict the outcomes of patients. Model performance was evaluated using concordance index (C-index), time-dependent receiver operating characteristic (ROC) curves, and decision curve analysis. Results: The 1-year, 2-year, and 3-year RFS rates were 71.10%, 43.10%, and 31.10%, respectively. For OS, the 1-year, 2-year, and 3-year OS rates were 99.10%, 82.30%, and 75.00%, respectively. The Decision Tree (DT) model demonstrated superior predictive performance with the highest C-index of 0.792. Time-dependent ROC analysis for predicting 1-, 2-, and 3-year RFS based on the DT model is 0.77, 0.79 and 0.76, respectively. Serum albumin emerged as the important predictor of RFS. Conclusions: Our study demonstrates the considerable efficacy DT model for predicting the efficacy of adjuvant IFN-α1b in patients with advanced melanoma. Serum albumin was identified as a key predictive factor of the treatment efficacy.