Tremor

Inspiration

The Unified Parkinson’s Disease Rating Scale (UPDRS) is a critical tool for characterizing the progression of Parkinson’s disease. One of the key symptoms of Parkinson’s is hand tremors, and the UPDRS motor test aims to visually inspect and assign a severity rating to these tremors. This rating determines the appropriate treatment for the patient. However, the process can be highly subjective, relying heavily on human judgment. Inspired by the need for a more objective and consistent assessment, we set out to develop a tool that uses machine learning to quantify tremor severity accurately and reliably.

What it does

Our tool is powered by a machine learning model that predicts the severity of hand tremors on a scale of 0 to 4 (as defined by the UPDRS) based on live sensor readings from a microcontroller. The microcontroller is attached to the hand of the person being tested, providing real-time data. This eliminates the subjectivity of visual assessments, enabling more accurate diagnoses and treatment planning.

How we built it

Using a microcontroller, we captured live acceleration readings from hand movements. These readings were transmitted to a serial port in real-time. On the software side, we retrieved and processed this data concurrently using threading. The data was filtered and analyzed to produce key features such as mean, max, min, standard deviation, and other metrics, which served as input for our machine-learning model. The model was trained to classify tremor severity based on this data. Finally, we designed a simple yet effective GUI to display the predictions along with live analytics of the tremor data.

Challenges we ran into

Integrating the frontend and backend while ensuring smooth real-time data transmission was a significant challenge. It required us to delve into multithreading and synchronization techniques to handle concurrent data retrieval and processing efficiently.

Accomplishments that we're proud of

We are incredibly proud of creating a working product that combines hardware and software knowledge. Our tool has the potential to make a meaningful impact by providing an objective and consistent method to assess hand tremors, benefiting patients and healthcare professionals alike. Successfully integrating live sensor data with machine learning and a real-time GUI is a major accomplishment for our team.

What we learned

We deepened our understanding of threading to manage live data processing efficiently.

What's next for Tremor

We aim to enhance the user interface to make it even more intuitive and efficient for healthcare professionals. We also want to explore incorporating more advanced analytics, such as frequency domain analysis, to provide deeper insights into tremor characteristics. Expanding the tool to support other types of tremors or movement disorders is also a future goal, potentially making it a versatile tool for neurology.