Using Genetic Algorithms for the Prediction of Cognitive Impairments

Cognitive impairments affect millions of persons worldwide, and especially elderly ones. These impairments may be one of the first signs of the arising of neurodegenerative diseases, such as Alzheimer’s and Parkinson’s, and it is expected that the incidence of this kind of diseases will dramatically increase worldwide in the near future. For this reason, the improvement of the tools currently used to diagnose these diseases is becoming crucial. Handwriting is one of the human skills affected by this kind of impairments, and anomalies such as micrographia have been adopted as diagnosis sign for the Parkinson’s disease. In a previous paper, we presented a study in which the handwriting of the subjects involved was recorded while they were performing some elementary tasks, such as the copy of simple words or the drawing of elementary forms. Then we extracted the features characterizing the dynamics of the handwriting and used them to train a classifier to predict whether the subject analyzed was affected by a cognitive impairment or not. In this paper, we present a system that uses a genetic algorithm to improve of the performance of the system previously presented. The genetic algorithm has been used to select the subset of tasks that allow improving the prediction ability of the previous system. The experimental results confirmed the effectiveness of the proposed approach.