Gait recognition in the classification of neurodegenerative diseases

Abstract

Incorrect disease diagnosis can lead to inappropriate treatment and serious impact on patient health. Neurodegenerative diseases diagnosis is currently based on neurologist observation, but, similarity in symptoms difficult early detection. This diagnosis can be supported by computational techniques such as classification by gait recognition. This has been well established in recent works for common disease like Parkinson, Alzheimer and Huntington, however, the efficiency of these techniques is unsatisfactory and only allow to classify one disease at a time. In this study we establish that meta-classifiers can be applied in diagnosis based on gait recognition for less commons diseases as Diabetic Neuropathy. We improve accuracy for ALS and we obtained the first results for Huntington with binary classification.