A study on the Simplification of Methods for Diagnosing Parkinson’s disease

Abstract

In this paper, a study was conducted to simplify the method for diagnosing Parkinson’s disease. In 2015, Parkinson’s disease affected 6.2 million people and resulted in about 117,400 deaths globally. Parkinson’s disease typically occurs in people over the age of 60, of whom about one percent are affected. Because of the mild and unnoticed symptoms, early diagnosis and medication of Parkinson’s disease is essential. As there is no complete cure, if early diagnosis and treatment are performed, it might help Parkinson’s disease patient to relieve symptoms. UPDRS (Unified Parkinson’s Disease Rating Scale) and Speech Examination, these are the two major features when diagnosing Parkinson’s disease. Using Microsoft Azure ML, applying two-class support vector machine and two-class boosted decision tree, 77.4% accuracy was shown with speech examination feature and with UPDRS feature 97.4% accuracy appeared finally, using both features 97.4% accuracy was given in diagnosing Parkinson’s disease. Which means that speech examination feature is less effective when diagnosing Parkinson’s disease patient.Applying the result proposed in this paper, doctors can shorten the time when diagnosing Parkinson’s patients and further diagnose other potential Parkinson patients.