Background information

As part of my second year (2025-2026) Design and Professional Practice module, we were tasked with creating a device that tracks Parkinsonian symptoms and make it usable for regular patients.

Design Choices

After consultations with doctors and parkinson patients, we decided that predicting the onset of Parkinson's proved too difficult with this requiring an algorithm that detects Parkinson symptoms out of nowhere. We therefore decided to track the progression of the disease with a watch based wearable to focus on tremor. Many other wearable alternatives were discussed such as a simple pocket watch or thigh attachment but upon consultation it was discovered that Parkinsonian tremor is the biggest tell tale factor that can be used to detect Parkinsons. A watch wearable was designed.

What I did

I was responsible for PCB design + the tremor detection pipeline. Whilst the PCB design was simple - the tremor pipeline was where the real fun was. We wanted an algorithm to differentite regular tremors from Parkinsonian tremors but lacked the data to map our IMU measurements to Parkinson tremors. As a result we created our synthetic dataset by using physics equations to convert the clinical Parkison measurement standard (MDRS-UPDRS) into actual IMU data for training for our MLP model.

If at this point you are still reading this blog, I applaud you so without furter ado see below for our report on our methodologies: