Epilepsia

Improved wearable devices for detecting epileptic seizures using artificial intelligence

Updated

Abstract

Detection of 900 seizures from 166 patients achieved a maximum sensitivity of 83.9% using a CNN-LSTM-based model.

  • Wearable sensors recorded electrodermal activity, accelerometry, and blood volume pulse signals.
  • Nineteen out of 28 seizure types could be detected by at least one data modality.
  • The best performance was achieved with a fusion of accelerometry and blood volume pulse data, which resulted in a false positive rate of 35.3%.
  • The study analyzed 13,254 hours of sensor data to evaluate seizure detection performance.

Simplified

Full Text

Full text is available at the source.

Funding

Competing interests

S.H., S.Y., and J.T. are part of patent applications describing technology to detect and classify epileptic seizures using EEG data and video data. S.V. is part of a patent application covering technology for seizure forecasting. T.L. is part of patents and patent applications to detect and predict clinical outcomes, and to detect, manage, diagnose, and treat neurological conditions, epilepsy, and seizures, but none of these has been licensed. He has received device donations to Boston Children's Hospital for research purposes from various companies, including Empatica. In the past, he has received research support paid to Boston Children's Hospital from Empatica for research unrelated to this study. The remaining authors have no conflicts of interest.
PubMed

What Lands in Your Inbox Each Week:

  • 📚7 fresh studies
  • 📝plain-language summaries
  • ✅direct links to original studies
  • 🏅top journal indicators
  • 📅weekly delivery
  • 🧘‍♂️always free