Abstract:
In the current busy lifestyle, people often neglect their health. As a result, they tend to suffer and spend unhealthy life due to many unknown health problems. One of those conditions is Parkinson’s disease, which is a neurological disease that mostly affects s the elderly. According to the UPDRS scale, Parkinson's disease has a variety of symptoms and can be detected using a variety of methods. Currently, there are a variety of methods to predict PD. Hindwing images, observing gait, MRI scanning, and speech recordings are used to predict Parkinson’s disease. The most effective way to diagnose PD is to have the patient draw something and identify the tremor signs they have. PDmate is a program designed to detect Parkinson’s disease at an early stage. It is designed using 3 publicly available datasets. This is a web-based project, which can use any age of healthy or Parkinson’s-affected patient. This thesis will propose a system that could detect Parkinson’s disease with the help of 3 different drawing patterns (Spiral, Meanders and Wave). Moreover, this research model is based on deep transfer learning, which can transfer knowledge from one model to another. So, this system may be able to help people affected with PD in an effective and accurate way.
