Homogenous Decision Tree Regressor Ensemble Model for Voice Features Modality based Early Diagnosis of Parkinson Disorder

Parkinson Disorder (PD) is a neurological disorder which is by nature progressive and degenerative. Dysphonia, a voice-based disorder is the most vital symptom exhibited by the 90% PD patients. PD has no cure and has no unique test. The delay in progression of PD can be made by the early diagnosis of the disease. The early diagnosis system can be made more accurate and effective by the incorporation of Artificial Intelligence (AI) technique. AI has a widespread application ranging from enterprise systems to small scale system. The proposed system aims to develop an AI based early diagnosis system based on voice features modality. The proposed system presents a Homogenous Decision Tree Regressor Ensemble model which predicts the Unified Parkinson Disorder Rating Score based on voice features. The proposed model is compared with the existing Decision Tree Regressor model. The suggested model is developed and tested with 42 PD patients voice features dataset. The evaluation metrics used are Mean Absolute Error, Mean Squared Error, and Co-efficient of Determination (R-Squared). It is evident from the results that the proposed model produces less error compared to the existing model.