A wearable sensor based multi-criteria-decision-system for real-time seizure detection

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Epileptic seizures require timely detection for effective intervention.
• Existing seizure detection methods often suffer from false alarms or missed detections.
• Wearable technology offers a promising avenue for continuous, unobtrusive physiological monitoring.

Purpose of the Study:

• To develop and evaluate a novel, low-power, low-cost wearable system for real-time epileptic seizure detection.
• To improve the accuracy of seizure detection by integrating multiple physiological parameters.
• To reduce false alarms and true negatives in seizure detection.

Main Methods:

• A two-part wearable system (chest and hand-worn) was developed to collect Electro-cardiograph (ECG), Electro-dermal Activity (EDA), body motion, and breathing rate (BR) data.
• A Multi-Criteria-Decision-System (MCDS) was designed to process these parameters.
• Long-Short-Term-Memory (LSTM) based anomaly detection and logistic classifiers were employed on a smartphone application for real-time analysis via Bluetooth Low Energy (BLE 4.0).

Main Results:

• The system demonstrated high performance on synthetic data.
• Achieved 96% precision, indicating a low rate of false positives.
• Achieved 90% recall, indicating a high rate of true positives detected.

Conclusions:

• The proposed wireless wearable system effectively detects epileptic seizures in real-time.
• The integration of multiple physiological parameters and advanced algorithms enhances detection accuracy and reduces false alarms.
• This technology holds potential for improving patient care and safety for individuals with epilepsy.