Compression-ratio-based seizure detection.

Area of Science:

• Biomedical Engineering
• Signal Processing
• Wearable Technology

Background:

• Wireless seizure monitoring devices require efficient seizure detection and data compression under strict power constraints.
• Current methods often treat these tasks separately, increasing system complexity and power consumption.
• Maximizing battery life is crucial for continuous patient monitoring.

Purpose of the Study:

• To develop a single, low-power algorithm for simultaneous seizure detection and data compression in wireless monitoring systems.
• To investigate the feasibility of using compression ratio (CR) as a marker for seizure detection.
• To reduce overall system power consumption by integrating functionalities.

Main Methods:

• Developed a unified algorithm combining seizure detection and data compression functionalities.
• Utilized compression ratio (CR) as an indicator for seizure events.
• Evaluated the algorithm's performance without introducing additional computational overhead.

Main Results:

• Achieved 88% accuracy in seizure detection.
• Recorded a low rate of 5.5 false positives per hour (FPh).
• Demonstrated that seizure detection can be a byproduct of data compression, incurring no extra computational cost.

Conclusions:

• The proposed CR-based seizure detection method offers a power-efficient solution for wireless seizure monitoring.
• Integrating seizure detection with data compression significantly reduces system power requirements.
• This approach enhances the practicality of long-term, continuous seizure monitoring through improved battery life.