Using Machine Learning to Unveil Early Signs of Parkinson’s Disease: A Review

Parkinson’s is a brain degenerative disease striking mostly over 60 s, which gradually steals movement. Tremors, stiffness, slowness, and walking troubles mark its motor symptoms. Later, fatigue, sleep issues, and thinking changes creep in. Degeneration in a brain region called Substantia Nigra, a midbrain region responsible for dopamine production, is the culprit. The cell death is caused by abnormal accumulation of proteins into Lewy bodies. This lack of dopamine disrupts both movement and other brain functions. Unfortunately, no cure has been found for this ailment till date, but early detection of the disease can help doctors to act at the right time in stimulating the deeper parts of the brain with L-DOPA and MAO inhibitors and the patient can return to a near normal lifestyle. Due to the deficit of dopamine in the brain, the communication between synapses is hindered and functions like speech, tremor, bradykinesia, rigidity, and stooped gait are the common symptoms. This is where machine learning comes in. With the help of the machine learning models like Support Vector Machine (SVM) and Random Forest (RF) and voice (tremor) dataset of patients with Parkinson’s, high accuracy results in early detection can be reached. Precise application of steps like data preprocessing and feature extraction helped in improving the performance metrics.