Parkinson’s disease detection using handwritten drawings and comparing it with voice dataset

• Main Authors: Sankar, Akshara R., Misra, Satya Kumar, Aswathy, S. U., Karthik, C., Chowdhury, Subrata
• Format: Conference Proceeding
• Language: English
• Subjects: Computer vision; Datasets; Diagnosis; Disease; Handwriting; Machine learning; Nervous system; Neurological diseases; Parkinson's disease; Photomacrographs; Signs and symptoms; Support vector machines; Voice
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0171239

Parkinson’s disease (PD) is one of the major neurological diseases affecting the nervous system of human body. Till now, there is no proper clinical examination that can diagnose a fully affected PD patient. But, the findings and reports states that the PD patients face disastrous changes in their handwriting. Hence, machine learning experts and research people have proposed different macrograph and computer vision (CV) based methods. Currently, it can take months to get an efficient and proper PD diagnosis, and symptoms that are to be noted and monitored effectively. Even on that note the probability of an improper diagnosis is approx. 20 Percent. I have used the Parkinson’s disease handwritten dataset and voice dataset. The results confirmed that handwriting is relevant in diagnosing and monitoring PD. Another set of voice dataset has been used to compare it with handwritten dataset. This is an attempt to find the disease as soon as possible by increasing the accuracy of previous results on the same by other researchers. Here I have used classifiers like LDA, KNN, SVM, RF and DT that can predict PD disease, from that SVM has shown greater result in both dataset but was giving greater accuracy in handwritten dataset.