Research Paper Volume 15, Issue 7 pp 2667—2688
Machine learning-based construction of immunogenic cell death-related score for improving prognosis and response to immunotherapy in melanoma
Figure 1. The work flow for the construction of ICDscore. Two clustering methods (consensus clustering and NMF clustering) were used for the molecular subtyping of melanoma patients, based on the gene expression of 34 ICD related genes. Two clusters (named as Cluster A and B) were identified and DEGs between these two clusters were analyzed. The LASSO regression model, multivariate Cox analyses were then used for the construction of ICDscore. The association between the ICDscore and prognosis, tumor immune microenvironment, or immunotherapy response was comprehensively investigated. The performance between ICDscore and other signatures was compared.