Automatic Diagnosis of Parkinson Disease through Handwriting Analysis: A Cartesian Genetic Programming Approach

Abstract

Early disease identification through non-invasive and automatic techniques has gathered increasing interest by the scientific community in the last decades. In this context, Parkinsons Disease (PD) has received particular attention in that it is a severe and progressive neurodegenerative disease and, therefore, early diagnosis would provide more prompt and effective intervention strategies. This, in turn, would successfully influence the life expectancy of the patients. However, the acceptance of computer-based diagnosis by doctors is hampered by the black-box approach implemented by the most performing systems, such as Artificial Neural Networks and Support Vector Machines, which do not explicit the rules adopted by the system. In this context, we propose a Cartesian Genetic Programming, aimed at automatically identify PD through the analysis of handwriting performed by PD patients and healthy controls. The use of such approach is particularly interesting in that it allows to infer explicit models of classification and, at same time, to automatically identify a suitable subset of features relevant for a correct diagnosis. The approach has been evaluated on the features extracted from the handwriting samples contained in the publicly available PaHaW dataset. Experimental results show that our approach compares favorably with state-of-the-art methods and, more importantly, provides an explicit model of the classification criteria.