A Real-Time Seizure Detection Framework Using PSD and FFT for IoT-Based Healthcare Monitoring

Sri Harsha Grandhi; Veerandra Kumar R

doi:10.62647/

Authors

Sri Harsha Grandhi Intel, Folsom, California, USA Author
Veerandra Kumar R SNS College of Technology, Coimbatore, Tamil Nadu, India. Author

DOI:

https://doi.org/10.62647/

Keywords:

EEG Seizure Detection, CNN-LSTM Model, Edge Computing, Real-Time Monitoring, Signal Processing, Feature Extraction, Power Spectral Density (PSD), Fast Fourier Transform (FFT), IoT (Internet of Things), Low Latency, Healthcare Systems

Abstract

An advanced real-time EEG seizure detection technique, in this context, is the hybrid CNN-LSTM model at the edge devices, which improves speed and accuracy in seizure detection. The incorporation of signal processing techniques, namely FFT and Power Spectral Density (PSD), into our methodology significantly improves seizure frequency identification. The system performs low-latency inference such that caregivers receive timely notifications while generally having higher performance metrics such as precision, recall, and F1. Current approaches still experience problems such as latency over the network and slower response times. This research addresses the synergistic enhancement benefits of IoT, edge AI, and advanced signal processing for effective health monitoring systems and timely treatment of patients during seizure episode