Parkinson’s disease is a neurodegenerative disorder that affects approximately 0.2% of the population having motor disabilities as its most prominent feature. A symptom of the disease is lowered dopamine levels which often is countered by oral intake of a medication called Levodopa. However, for the dopamine levels to be steady, a patient would need to regularly take the medication throughout the day. As the disease and the treatment progresses, the correct medicine prescription becomes more difficult. This project is the continuation of a previous project done by students at Uppsala University, in which a Machine Learning model with the help of Support Vector Machine could classify data collected from a handheld accelerometer as the user being either under or overdosed for Parkinson’s Disease. The goal of this project was to achieve a similar result by developing a mobile app. The mobile app was supposed to allow the user to follow a path displayed on the screen with their finger, meanwhile the app would collect touch data in the form of coordinates and timestamp these. The app development proved to be successful, and the collected data was sent to a database hosted on the Google cloud service Firebase for storage. From there, the data could be downloaded and imported to MATLAB where an SVM model was set up and trained. Once trained using data collected from healthy individuals as well as patients suffering from Parkinson’s disease, the SVM could accurately differentiate between Parkinson’s disease data and healthy data with a success rate of 91.7%.
| Identifer | oai:union.ndltd.org:UPSALLA1/oai:DiVA.org:uu-483768 |
| Date | January 2022 |
| Creators | Olsson, Daniel, Eriksson, Jonathan, Soltani, Sedigheh |
| Publisher | Uppsala universitet, Institutionen för elektroteknik |
| Source Sets | DiVA Archive at Upsalla University |
| Language | English |
| Detected Language | English |
| Type | Student thesis, info:eu-repo/semantics/bachelorThesis, text |
| Format | application/pdf |
| Rights | info:eu-repo/semantics/openAccess |
| Relation | ELEKTRO-E ; 22005 |
Page generated in 0.0017 seconds