Statistical approach to selecting the optimal parameters for diagnosis of some connective tissue diseases

In order to choose the optimal parameters for easier diagnosis of systemic autoimmune diseases, the authors focused on data dimensionality reduction, using both feature selection and feature extraction. The Multiple Correspondence Analysis was used as a feature extraction method, with the aim of exploring the underlying data structure and detecting the crucial latent variables. The obtained latent variables were used as an input for the Discriminant Analysis which correctly classified 86.5% of all analyzed cases. The high rate of correctly classified objects indicates that it would be possible to automate diagnostic processes, which would lead towards the development of decision support systems in this area of medicine. In addition to their knowledge and experience, clinical experts would have further help in decision support systems. That can allow easier learning, faster checking of diagnostic steps, lower rates of misdiagnosed cases and easier communication with experts from other medical centers.