Epilepsy is a neurological disorder that affects 600,000 people in France. Almost a third of these patients are known as “drug-resistant”, which means that they continue to have seizures despite drug treatments. Consequently, they have very few alternatives. These unpredictable seizures are a source of stress and a daily accident risk for patients. Besides, this is a complex disorder and there are more than 80 distinct epilepsies.
Aura is a non-profit association, the aim of which is to improve the quality of life for people with epilepsy. By means of an open science approach, the project brings together a multi-disciplinary community for the purpose of developing a connected thoracic patch enabling epilepsy seizures to be detected and the patient to be warned. For a person with epilepsy, a better understanding of their disorder represents a first step in regaining their independence and everyday peace of mind.
A key component of the device is the design of a customised algorithm for detecting epilepsy seizures. This algorithm is based on statistical methods and makes use of machine learning. In order to overcome this technical challenge, we have put in place an automated processing chain based entirely on an open source stack (Airflow, MLFlow, InfluxDB, Grafana) and a set of dedicated open source libraries for processing the cardiac signal. We will explain how this approach, which is still too marginal in the health sector, guarantees the transparency, reproducibility and robustness of our results, as well as the dissemination of our work among the scientific community.
Open science in the service of people with epilepsy