Prediction of Parkinson Diseases at Early Stage Using Deep Learning Method

The second most prevailing neurodegenerative disorder is Parkinson’s disease (PD) that have been affected nearly 6 million of people in the world. The availability of emblematic treatment has the ability to increase the survivability of disease but there is no curing treatment available. The PD predominance and the disability life years are endured to increase steadily that may lead to a viable burden on patients and their families. Dopaminergic medications can typically slow down the PD progression whileused during the earlier stages. Theses treatment is frequently become less efficiency with the disease progress. Diagnosis of PD at early state is essential for immediate interventions which assist the patients to remain self-sufficient with long time period probably. Disturbances of balance, gait and posture are a hallmark of parkinsonian syndromes. Specifically, to uncover PD at an early stage, several indicators have been considered by testing the person by Neuro-radiological tools, cognitive tests, freezing of gait questionnaire and Movement Disorders Society- Unified Parkinson's Disease Rating Scale (MDSUPDRS). In this paper,the proposed fuzzy logic approach for prediction of PD and their severity classification withLong Short Term Memory (LSTM) is utilized for overcoming the curse dimensionality. The dataset involved consists of both numerical and categorical data which performs classification of PD patient at early stage by categorical data. Based on LSTM, the fuzzy rule get generated and the fuzzy logic computation technique is depend upon degree of truth. Using fuzzy logic the accurate prediction of PD at early stage and its level of severity is identified based on fuzzy rule with LSTM with accuracy of 96.5%.