Computer-aided diagnosis of multiple sclerosis

The study aims to develop a computer-assisted decision support based on cerebrospinal fluid (CSF) and blood findings to improve their value and ease the diagnostic procedure of chronic inflammatory diseases (CIDs) of central nervous system (CNS). Data were collected from patients suffering from multiple sclerosis (MS, n=73), from another CID of the CNS (n=22), or a psychiatric disease (control group, CTRL, n=12). Univariate and multivariate analyses were performed using multiple logistic regression and artificial neural networks. Differentiating between MS and CID, no parameter could be disclosed that could provide a meaningful decision support. However, multivariate analysis obtained a statistically significant classification (sensitivity=84.9%, specificity=54.5%, p0.001). On the contrary, multivariate analysis based on the differentiation between MS vs. CTRL, gave good results (sensitivity=95.9%, specificity=83.3%, p0.001). It became evident from standard laboratory findings that there is a significant potential for computer-aided decision support.