期刊名称:International Journal of Advanced Research In Computer Science and Software Engineering
印刷版ISSN:2277-6451
电子版ISSN:2277-128X
出版年度:2012
卷号:2
期号:3
出版社:S.S. Mishra
摘要:State-of-the-art research in the field of data min ing is having great impact in the area of medical diagnosis and disease prediction. This paper places emphasis on classifying the severity of Parkinson's disease (idiopathic Parkinsonism). This disease is a neurodegenerative disorder of the central nervous system. The brain cells (neurons) in the human brain produce dopamine in a particular area of the brain called the substantia nigra. Symptomatic identification includes loss of these specific brain cells and decline in dopamine concentration. Unified Parkinson Disease Rating Scale (UPDRS), captures multiple aspects of Parkinson Disease that include Mentation, Behaviour and mood, Activities of Daily Life (ADL), Motor Examination and complications of therapy. In the Parkinson Disease Tele-monitoring Dataset, the data consists of 16 biomedical voice measures with test subject information, motor UPDRS and total UPDRS scores. The main goal of this work is to predict the motor and total UPDRS scores from the voice measures. Predicting the scores can be done by extracting useful knowledge and thereby providing the scientific decision-making classification rules necessary for the diagnosis of disease severity. This is done by precisely classifying the given dataset and relegating them into people with high scores and low scores. Moreover this paper highlights the impact of six feature relevance algorithms and thirteen classification algorithms on the Parkinson Tele-monitoring dataset. We report 100 percent accurate classification by the Random Tree classification algorithm with the features filtered by the ReliefF algorithm.