Deep Learning-Based Stroke Disease Prediction System Using Real-Time Bio Signals

Area of Science:

• Biomedical Engineering
• Neuroscience
• Artificial Intelligence in Healthcare

Background:

• Aging populations and reduced birth rates necessitate advanced smart healthcare solutions.
• The COVID-19 pandemic accelerated the demand for contactless, non-face-to-face health services.
• Traditional stroke diagnosis using MRI/CT is costly, time-consuming, and requires specialized equipment.

Purpose of the Study:

• To develop a novel deep learning methodology for immediate application on raw electroencephalogram (EEG) data for stroke prediction.
• To overcome the limitations of traditional stroke diagnostic methods by utilizing non-invasive EEG measurements.
• To evaluate the performance of various deep learning models for stroke prediction using raw EEG data.

Main Methods:

• Proposed a deep learning-based stroke prediction model trained on real-time EEG sensor data.
• Implemented and compared LSTM, Bidirectional LSTM, CNN-LSTM, and CNN-Bidirectional LSTM models for time series data classification.
• Utilized raw EEG data directly, bypassing traditional frequency-based attribute extraction.

Main Results:

• The CNN-Bidirectional LSTM model achieved 94.0% accuracy in predicting stroke from raw EEG data.
• The model demonstrated a low False Positive Rate (FPR) of 6.0% and a low False Negative Rate (FNR) of 5.7%.
• Experimental results confirmed the feasibility of using raw EEG for real-time stroke prediction.

Conclusions:

• The developed deep learning model provides a highly accurate and efficient method for non-invasive stroke prediction.
• This approach significantly reduces cost and discomfort compared to conventional imaging techniques.
• The findings support the potential for real-time monitoring and early stroke detection in daily life settings.