Fig. 5

Results and relevance of the different features (tPRM metrics as inputs) used in the dictionary learning method. (A) Confusion Matrix showing the sensitivity and specificity of the ML model classifications for both the fast progressor (n = 985) and the slow progressor (n = 1929) classes in the test set. Green colored and red colored fields in the matrix represent agreement and disagreement, respectively, of the ML model with the actual decision. (B) Receiver Operating Characteristic (ROC) curve for our ML model and the logistic regression classifier with the corresponding Area Under the Curve (AUC) statistics. (C) Bar plot showing the feature importance score and feature ranking using the minimum redundancy maximum relevance method. (D) Plot showing the distribution of the features and their prediction accuracy over ten different training runs