Serum and Cerebrospinal Fluid Cytokine Biomarkers for Diagnosis of Multiple Sclerosis

Figure 5

Schematic representation of the proposed model. Dataset 1: IL-8, IL-12p40, CCL2, CCL5, CXCL1, CXCL9, HGF, IFN-α2, b-NGF, and SCGF-b (cytokines uniquely affected in serum). Dataset 2: IL-1α, IL-4, IL-18, CCL7, CCL27, INFγ, LIF, M-CSF, TNF-α, and SCF (cytokines found commonly affected both in serum and CSF of MS). Dataset 3: IL-1β, IL-2, IL-7, IL-10, IL-9 IL-12(p70), IL-16, CCL3, CCL4, GM-CSF, PDGF-bb, and TRAIL (cytokines uniquely affected in CSF). Dataset 4 consists of dataset 1 and 2; Dataset 5 consists of dataset 2 and 3. Five cytokines from the dataset 1, 2, and 3 can be estimated in serum and CSF and then will be given as input to into the five (gNB, KNN, DT, XGB, and RF) machine learning models to predict the MS. Dataset 4 and dataset 5 in serum and CSF, respectively, can be estimated to discriminate between progressive (primary and secondary progressive) and relapse remitting forms of MS.