Machine Learning Methods for Managing Parkinson’s Disease

Abstract

A neurodegenerative disorder without permanent cure, Parkinson’s Disease (PD) increasingly hinders motor and cognitive abilities. Timely intervention of neuroprotective therapies can help minimize the early impairments in PD. Early diagnosis would play major role in facilitating such proactive treatment plan. However, the conventional methods of PD diagnosis suffer from less accessibility, high costs, human bias and patient inconvenience. Moreover, there is a dearth of high-frequency monitoring systems to track the progression. Deficient monitoring and management of the progression diminishes both quality of life and life expectancy of the patient. The challenges and concerns in conventional methods of diagnosis and treatment of PD call for use of advanced technology like Machine Learning (ML) and Internet of Things (IoT). The proposed chapter is aimed at giving insights into robust and effective practical implementation of ML with IoT in PD care. Non-invasive biomarkers data from human voice and keystrokes (tapping) are demonstrated as promising base for early diagnosis. These illustrations focus on ease of building cost efficient and scalable PD prediction systems. In addition, a multitude of contemporary developments and inspiring future opportunities for managing PD with ML are highlighted.