What it does
Epilepsy affects over 70M people, causing physical injuries and psychosocial isolation. We propose a seizure detection and treatment system with an AI-powered wearable tiny EEG patch for real-time detection and a vagus nerve stimulator for seizure inhibition.
Your inspiration
Yazhou was a research intern working on data analysis for epilepsy patients at Shenzhen University General Hospital when she was a freshman. She felt a personal connection to the patients and their families as she reviewed data that included patients' names and detailed information. This experience makes her desire to help those affected by epilepsy. Last year, she proposed a rapid seizure detection algorithm at EMBC2023. To ensure clinical practicality, she began developing a real-time seizure detection and treatment system. That is the beginning of the idea.
How it works
Our project is organized into three parts: 1. Algorithm and Software: Software for seizure detection. Currently, we are using a CNN-BiLSTM deep learning structure. It is trained on open datasets, including the BONN dataset, the CHB-MIT dataset, and data collected from our partner hospital, HKU-Shenzhen Hospital. 2. Wearable EEG Sensors: Wearable clinical EEG sensors embedding our algorithm for real-time sensing, computation, and alerting. These sensors contain 4 electrodes and can be hidden in hats and wigs. 3. Stimuli Integration [Ongoing]: Embedding a stimuli component into the device and algorithm to enable real-time intervention and management of epileptic seizures.
Design process
At first, it was just an algorithm for seizure detection, trained on the open BONN dataset with data from only two patients. Afterward, the algorithm was improved using a larger open dataset. We also reached out to neurosurgeons for additional EEG data. Currently, we have non-public data from 20 epilepsy patients from our partner hospital. In the meantime, we developed the hardware. The first version had four electrodes that could be freely attached to patients' heads using hydrogel. Later, we combined the four electrodes into a small patch that could be attached to the head.
How it is different
Unlike current clinical EEG caps, it offers superior wearability, making it comfortable and easy to use. Our design is very small, enabling it to be hidden in hats and wigs. It provides clinical-level accuracy, allowing patients to use our design during their normal daily activities, including at home and in public areas. Furthermore, it is more affordable compared to high-cost systems like the EEG cap offered by g.tec. Additionally, it supports therapeutic applications, which most competitors lack. This combination of clinical utility, affordability, comfort, and versatility makes our product a standout choice in the EEG sensor market.
Future plans
We founded a startup and would like to keep working on it. Here is the future plan: 2024: Clinical tests for AI algorithms. Hardware development. The first generation prototype. 2025: The first generation product. FDA approval for the AI algorithm. 2027: Clinical tests for the first generation product. The second generation product. FDA approval for the first generation product. 2028: Clinical tests for the second generation product. NMPA approval for the first generation product. Enter the Chinese market. 2030: FDA approval for the second generation product. Enter the US market. Achieve payback.
Awards
Tam Wing Fan Innovation Fund, Philomathia Foundation Innovation Fund, University of Hong Kong, April. 2024 Winner, Pitch New Ideas 2024, University of Hong Kong, Mar. 2024 Ideation Seed Fund, Hong Kong Science and Technology Parks Corporation, Nov 2023
Share this page on