Machine Learning Approaches for Early Parkinson’s Disease Detection: An Analytical Review

Parkinson’s disease is a neurological condition which requires early diagnosis and interventions for treatment because it worsens with time. New advances in machine learning have opened new opportunities for increasing the accuracy in diagnoses. For example, research for SPECT images of putamen and caudate regions of the brain using Artificial Neural Networks (ANN) has revealed enormous promise for the diagnosis of Parkinson’s disease. Similarly, voice data analysis which are recorded by smartphones and other appliances is showing great promise in the diagnosis process, with certain classifiers reaching nearly 86%.Another method uses Extreme Learning Machines (ELM) to predict speech samples with an accuracy of 90.76% on training data and 81.55% on independent datasets. Such advancements demonstrate how machine learning is helping doctors in making early and accurate diagnoses of the Parkinson’s disease.