Many parents who have children have epilepsy are worried about their children at school, and many teachers might lack the knowledge to recognize a seizure when it happens. We can to give the parents a peace of mind, give the children a better childhood, and give doctors more data to help cure epilepsy.
What it does
The device measures the acceleration of children's arms and uses the data to detect whether a child is having a seizure. If a seizure is detected, a red led alarm lights up on the device to signal for help. If the alarm is not disabled by pressing a button on the device within 5 seconds, a text message is sent to the parent. The data is also recorded in a database for future data analytics and machine learning.
How I built it
We used the Mbed connect cloud to connect to the internet and measure acceleration. It pipes the raw data to the python server. The server then performs FFT on the data and looks at the high-frequency components to determine whether the child is having a seizure. It then uses some APIs to send a text message.
Challenges I ran into
There were many problems when programming the Mbed, as c++ has many potential pitfalls. For example, we initially did the HTTP requests in the main loop, but that didn't work due to multithreading reasons. Therefore we had to use an RTOS timer to do the requests.
Accomplishments that I'm proud of
We coded a complete solution for the problem in 24 hours and developed an algorithm that is reliable enough to exclude most of the false positives.
What I learned
I learned a lot about Mbed programming and how to debug things in c++.
What's next for Epilepsy Detector
We want to make an app that lets parents and doctors view the seizure histories of their children.