Many people with a mild to extreme case of tourette's syndrome are likely to be in unbearable pain. With the cause unknown, there are uncertainties when "treating" tourette's. Our motivation is to provide a platform that will both monitor and provide musical therapy to patients.
T-Solvies comes in two components: desktop, arm band (outside use). It is scientifically proven that the playing of music helps ease a turette's syndrome attack. All three components detect the occurrence of an attack, and respond with the playing of relaxing music.
The desktop application is built to run in the background every time the patient is at his or her table. Using the webcam, erratically fast motion--the type which occur during physical tics of a turette's attack--can be detected with blur recognition.
The arm band, on the other hand, detects sharp movements with its accelerometer and gyroscope. In the event where the patient is outside, or away from a computer, he or she may opt for the arm band.
Both components store relevant data regarding the attacks, potentially making it easier for medical professionals to gain insight on the lives of the patients.
Challenges we ran into
The once planned third component of T-Solvies, an audio recognition AI, was ultimately scrapped because of the limited audio data on people with turette's syndrome to do a binary classification.
We also had trouble tuning the hardware. Originally, we had planned for a binary searching mechanism on the total magnitude of acceleration. The binary search tuning option did function properly, but the margin of error in our hardware made this redundant, and so we ultimately scrapped it.
Accomplishments that we're proud of
First and foremost, we are proud to have made a functioning product which can have the potential to help people with turette's syndrome with further development.
We are also proud of our willingness to scrap ideas, despite having poured hours in them.
What we learned
We put a lot of effort in removing acceleration due to gravity using linear algebra (with the Gyro), but the hardware's margin of error was too big for this to make a significant impact. We learned to do error analysis before considering optimization.
With this effort, we learned a ton of math!
What's next for T-Solvies
We would like to obtain better hardware to truly have a adjustable sensitivity based on user feedback.
Additionally, we would like to obtain more data for our for now scrapped audio recognition AI.