Inspiration
around 50 million people suffer from seizures globaly and 80% of them live in low-income contries and Invasive EEG-based predictive implant systems (devices that can detect a seizure before hapening)cost from 30000$ - 200000$ making them unaccessible to most people and only a few can afford them what more data focusing on seizures is very limited my site aims to fix that by providing a free and accessible way to track seizures and try to predict the risk of them happening in hopes that this can help with the issues mentioned before
What it does
The app uses a sklearn LogisticRegression that takes the user logs and adapts to their data (medication take, sleep, stress) based on these it tries to make better predictions (return Low/Elevated/High instead of an exact time) at first the model starts with standard weight based on real reaserch data then after the user fills enough daily and seizure logs that data will be fed to the model and it will start to adapt to the user so after a few days each user will have model adapted to their condition that gets better each day as long as they stay consistant with entering the their logs also the user can get a CSV doctor report based on their logs this can help when seeking medical aid (For more info about how the model works read model_description.txt)
How we built it
I build it using django for the back were all the data lives the front with react it is only used for the ux ui and user interaction
Challenges we ran into
making the Ml model was the hardest part i worked with similiar models previous and it was really exiting to get back to make one, i had to find a to make the stats reflect how they effect the user and the model
Accomplishments that we're proud of
As always finishing the project to the end is always something to be proud of, this is probably my most original and unique idea one it comes to a hackathon project
What we learned
learned more about how to handle security of data and users, also got to revise how logistic regression models work
What's next for SeizureLog
probably more user data like menstrual cycle that can be used to inhance the model
Log in or sign up for Devpost to join the conversation.