华中师范大学学报（自然科学版）2026，Vol.60Issue(2)：296-307,12.DOI:10.19603/j.cnki.1000-1190.2026.02.012
基于机器学习算法的黑色素瘤免疫治疗预后预测模型构建
Construction of a machine learning-based prognostic model for melanoma immunotherapy
摘要
Abstract
Immunotherapy resistance in melanoma correlates with dynamic heterogeneity in the tumor microenvironment(TME).This study aims to integrate gene expression profiles with clinical parameters through bioinformatics techniques,propose a TME classification strategy,and develop a melanoma prognosis prediction model(MMPS)to achieve dynamic prognostic stratification,thereby providing a theoretical basis for biomarker combination therapy.Leveraging the TCGA-SKCM cohort,this study quantifies TME features using algorithms such as ESTIMATE and CIBERSORT.Core prognostic genes were identified through unsupervised clustering and LASSO-Cox regression,leading to the construction of the melanoma microenvironment prognostic score(MMPS)model.Evaluation demonstrated the model's significant advantage in predicting 5-year overall survival,with a statistically significant 28％improvement in predictive performance compared to traditional TNM staging.This model provides a scientifically sound and reliable quantitative basis for assessing prognosis in melanoma patients.关键词
黑色素瘤/机器学习/预后预测模型/肿瘤微环境/免疫治疗Key words
melanoma/machine learning/prognostic prediction model/tumor microenvironment/immunotherapy分类
生物科学引用本文复制引用
董静,汪译函,卜佩,黄晓丹,陈珊,何岚,刘喜武,张月明..基于机器学习算法的黑色素瘤免疫治疗预后预测模型构建[J].华中师范大学学报（自然科学版）,2026,60(2):296-307,12.基金项目
湖南省卫生健康委员会卫生科研课题(202204083643) （202204083643）
湖南省自然科学基金项目(2023JJ60063) （2023JJ60063）
湖南中医药大学校级科研项目(2022XYLH151). （2022XYLH151）
