Machine Learning Analysis of Immune Cells for Diagnosis and Prognosis of Cutaneous Melanoma

Figure 2

The immune cell-related diagnostic model of melanoma patients. (a–d) Volcano plots of TCGA dataset (a), GSE3189 (b), GSE15605 (c), and GSE46517 (d), which illustrated the differently-infiltrated immune cells between the melanoma and healthy controls. Red and blue plots mean statistical significance (). (e) The Upset plot of immune cells in different datasets. The amount of each dataset is shown by the dark bar on the left of the drawing. The black dots in the matrix on the right of the drawing indicate immune cell intersections. (f–i) Receiver operating characteristic (ROC) curves on multiple machine learning algorithms for the diagnostic model in TCGA dataset (f), GSE3189 (g), GSE15605 (h), and GSE46517 (i). AUC, area under ROC curve; LR, logistic regression; RF, random forests; SVM, support vector machines; LASSO, least absolute shrinkage and selection operator; NNET, neural network. (j) Principal component analysis of the expression of immune cells in the TCGA dataset. (k) Different distributions of diagnostic scores in multiple datasets. The box-violin plots indicate diagnostic scores at the median and interquartile range of value. means ; means ; means ; means .

(a)

(b)

(c)

(d)

(e)

(f)

(g)

(h)

(i)

(j)

(k)