Fig. 6
From: Decoding distinctive features of plasma extracellular vesicles in amyotrophic lateral sclerosis
The size distribution and the amount of HSP90 in EVs help to correctly classify ALS patients. a-b machines learning frameworks used in the experiments. Scheme a is related to the basic framework, while scheme b is related to the advanced one. They both consist of three main steps: data handling, training of the models and testing of the models on an independent cohort. Those schemes differ in the data handling phase. The basic framework only preprocesses data, while the advanced one uses a tailored oversampling strategy to handle unbalanced data. TR is training set; TS is test set. c Precision-recall curves on size distribution for a binary comparison. d Precision-recall curves on oversampled data on size distribution for a binary comparison. e Precision-recall curves on size distribution and HSP90 for a binary comparison. f Precision-recall curves on oversampled data on size distribution and HSP90 for a binary comparison