Enhanced Drowsiness Prediction through a Fusion of ANFIS and BLSTM Models

This study proposes an innovative hybrid model to detect driver drowsiness by integrating an Adaptive Neuro-Fuzzy Inference System (ANFIS) with a Bidirectional Long Short-Term Memory (BLSTM) network. Given that drowsiness is a critical factor in road accidents, the development of accurate and efficient detection systems is essential for improving traffic safety. ANFIS demonstrates a rapid convergence within just 20 iterations, effectively addressing nonlinearities and uncertainties in behavioral data. BLSTM complements this by capturing complex temporal dependencies in both directions, achieving an accuracy of 98.48% and an F1 score of 0.985507. The fusion of these models further enhances performance, with the hybrid ANFIS-BLSTM model achieving a remarkable accuracy of 98.46% and reducing the mean squared error (MSE) to 0.019896. These results confirm the model’s ability to provide a robust and reliable solution for real-time drowsiness detection, ultimately contributing to safer driving conditions.