Classificação fuzzy de padrões não-motores e indicação da severidade da Doença de Parkinson

Abstract

Parkinson's disease is an age-related neurodegenerative disease. About 1% of individuals over the age of 65 develop the disease. Recent research on incipient Parkinson's disease detection has indicated subtle changes in voice, hyposmia and sleep disorders as the first indicators of the disease. This work considers analyses of speech amplitudes in certain frequencies and computational intelligence algorithms for incipient detection of non-motor patterns of the Parkinson's disease. The large number of data and variables involved and the uncertainty about exact values make expert analyses difficult and imprecise. Clustering algorithms, viz. Fuzzy C-Means and Gustafson-Kessel, were implemented to analyse attributes extracted from a database provided by the University of Oxford. The algorithms have presented results regarding the inference of the severity of the Parkinson's disease for each individual considering the UPDRS (Unified Parkinson's Disease Rating Scale). Particularly, the Gustafson-Kessel algorithm has provided the best results in terms of correct classifications according to severity levels.