Deep learning-aided Parkinson's disease diagnosis from handwritten dynamics

Abstract

Parkinson's Disease (PD) automatic identification in early stages is one of the most challenging medicine-related tasks to date, since a patient may have a similar behaviour to that of a healthy individual at the very early stage of the disease. In this work, we cope with PD automatic identification by means of a Convolutional Neural Network (CNN), which aims at learning features from a signal extracted during the individual's exam by means of a smart pen composed of a series of sensors that can extract information from handwritten dynamics. We have shown CNNs are able to learn relevant information, thus outperforming results obtained from raw data. Also, this work aimed at building a public dataset to be used by researchers worldwide in order to foster PD-related research.