Enhancing driver safety with ECG-based emotion recognition using BiLSTM networks

Emotions critically influence human decision making and behaviour, particularly in safety-sensitive contexts like driving. This study introduces an ECG-based emotion recognition framework suitable for online driver monitoring system that exclusively analyses electrocardiogram (ECG) signals through a Bidirectional Long Short-Term Memory (BiLSTM) network. The framework captures temporal dynamics in physiological features – including heart rate variability and signal entropy – to classify seven emotional states (neutral, happy, sad, angry, fear, surprise, disgust) with high accuracy. Beyond detection, the system incorporates an intelligent recommendation mechanism designed to mitigate emotional distractions demonstrating how emotion predictions could be translated into driver-support feedback. Experimental validation on synthetic ECG data demonstrates robust emotion classification performance in identifying complex emotional patterns from ECG data, outperforming conventional unimodal approaches. By bridging affective computing with intelligent transportation systems, this work advances the development of adaptive Driver Assistance Systems (DAS) that prioritize both road safety and user wellbeing. The proposed system’s real-time capability and nonintrusive design position it as a scalable solution for emotion aware environments, demonstrating the potential of ECG-based emotion recognition as a supporting component for future driver assistance systems. This research contributes to the growing field of affective human-machine interaction while demonstrating practical applications for intelligent transport systems.