Parkinson’s disease diagnosis using neural networks: Survey and comprehensive evaluation

Parkinson’s disease (PD) is a chronic neurodegenerative disease of that predominantly affects the elderly in today’s world. For the diagnosis of the early stages of PD, effective and powerful automated techniques are needed by recent enabling technologies as a tool. In this study, we present a comprehensive review of papers from 2013 to 2021 on the diagnosis of PD and its subtypes using artificial neural networks (ANNs) and deep neural networks (DNNs). We present detailed information and analysis regarding the usage of various modalities, datasets, architectures and experimental configurations in a succinct manner. We also present an in-depth comparative analysis of various proposed architectures. Finally, we present a number of relevant future directions for researchers in this area. •We provide the most relevant information collected from 143 papers published from 2013–2021 on diagnosis and classification of Parkinson’s disease.•We used artificial and deep neural networks in a highly compact manner within this paper.•We design this paper in a manner that enables a reader to objectively compare the network architectures used by the researchers.•We provide insights on various aspects of deep networks and their training configurations used by researchers and discuss their efficacy.•We provide numerous future directions by the help of our discussions and supporting materials.