EpiSense: The Epileptic Seizure Sensor

Inspiration

Our group was inspired by the rise of epilepsy cases in the 2010s and has identified the leading causes of seizure-related deaths that patients lacked access to health resources or communications to close people. We hoped to design a solution that closes this gap on a very present problem.

What it does

Epi-Sense is a wearable device that monitors biometric sensors (skin conductivity and motion) in real-time to detect and predict epileptic seizures. It integrates with a backend system that analyzes the data using neural networks and Fourier transforms to flag high-probability seizure events based on oscillations in the data. It also has an iPhone and Apple Watch app that provides instant notifications to healthcare providers/caregivers when a seizure is detected, along with a health chat-bot, appointment scheduling, and map integration.

How we built it

The device consists of wearable biometric sensors that measure skin conductivity and motion, connected to a custom microcontroller (ESP32) with WiFi connectivity. The microcontroller reads the sensor data and sends it to a Firebase Realtime Database in the cloud. The backend system analyzes the data using machine learning models (neural networks and Short-time Fourier Transform) to detect seizure patterns. The iPhone and Apple Watch app communicates with the backend to receive seizure alerts and provide additional features like a chat-bot and appointment scheduling.

Challenges we ran into

EEG machines are expensive and not easily interfaceable with microcontrollers. That’s why we decided to use cheaper GSR sensors for real time analysis to interface with the microcontroller and app. Due to the varying conductivity levels, we were not able to rely on if/else statements, and must build a Machine Learning Model to recognize patterns.

Accomplishments that we're proud of

Creating a fully-functional, multi-modal, full-stack product with the potential to impact the future of medical assistance. Being able to incorporate multiple technologies like SauceLab and Adobe.

What we learned

We learned how to integrate different technologies together and how to troubleshoot through all of them.

What's next for EpiSense: The Epileptic Seizure Sensor

We want to integrate deployable and wearable airbag garments for protection against falls. We also want to use more sophisticated datasets for our machine learning algorithms, focusing on more concrete EEG data.