# Towards Long-term Non-invasive Monitoring for Epilepsy via Wearable EEG Devices

arXiv: https://arxiv.org/abs/2106.08008 · HTML: https://mlsanity.com/paper/2106.08008/

## Key facts

- Cited by 6 papers in the map (data through 14 Oct 2026).
- Builds on 2 of its 25 references that are in the map.
- First posted on arXiv 15 Jun 2021.

## Builds on (most cited first, 2 of 2)

- [ECG-TCN: Wearable Cardiac Arrhythmia Detection with a Temporal Convolutional Network](https://mlsanity.com/paper/2103.13740/) (2021)
- [Systematic Assessment of Hyperdimensional Computing for Epileptic Seizure Detection](https://mlsanity.com/paper/2105.00934/) (2021)

## Built on by (most cited first, 6 of 6)

- [SzCORE: A Seizure Community Open-source Research Evaluation framework for the validation of EEG-based automated seizure detection algorithms](https://mlsanity.com/paper/2402.13005/) (2024)
- [Energy-Efficient Tree-Based EEG Artifact Detection](https://mlsanity.com/paper/2204.09577/) (2022)
- [Ensemble learning using individual neonatal data for seizure detection](https://mlsanity.com/paper/2204.07043/) (2022)
- [BISeizuRe: BERT-Inspired Seizure Data Representation to Improve Epilepsy Monitoring](https://mlsanity.com/paper/2406.19189/) (2024)
- [EpiDeNet: An Energy-Efficient Approach to Seizure Detection for Embedded Systems](https://mlsanity.com/paper/2309.07135/) (2023)
- [Skilog: A Smart Sensor System for Performance Analysis and Biofeedback in Ski Jumping](https://mlsanity.com/paper/2309.14455/) (2023)

## Abstract (arXiv)

We present the implementation of seizure detection algorithms based on a minimal number of EEG channels on a parallel ultra-low-power embedded platform. The analyses are based on the CHB-MIT dataset, and include explorations of different classification approaches (Support Vector Machines, Random Forest, Extra Trees, AdaBoost) and different pre/post-processing techniques to maximize sensitivity while guaranteeing no false alarms. We analyze global and subject-specific approaches, considering all 23-electrodes or only 4 temporal channels. For 8s window size and subject-specific approach, we report zero false positives and 100% sensitivity. These algorithms are parallelized and optimized for a parallel ultra-low power (PULP) platform, enabling 300h of continuous monitoring on a 300 mAh battery, in a wearable form factor and power budget. These results pave the way for the implementation of affordable, wearable, long-term epilepsy monitoring solutions with low false-positive rates and high sensitivity, meeting both patient and caregiver requirements.