IoT-Enabled WBAN and Machine Learning for Speech Emotion Recognition in Patients

Area of Science:

• Healthcare technology
• Machine learning
• Signal processing

Background:

• Internet of Things (IoT)-enabled Wireless Body Area Networks (WBANs) integrate medical and non-medical devices for healthcare.
• Speech Emotion Recognition (SER) in healthcare faces challenges like low accuracy, high computational complexity, and real-time prediction delays.
• Identifying appropriate speech features for SER is crucial but difficult.

Purpose of the Study:

• To propose an emotion-aware IoT-enabled WBAN system using edge AI for real-time patient speech emotion prediction.
• To investigate the effectiveness of various machine learning and deep learning algorithms for SER.
• To develop and optimize hybrid deep learning models for improved accuracy and reduced computational complexity.

Main Methods:

• Developed a hybrid deep learning model combining Convolutional Neural Network (CNN) and Bidirectional Long Short-Term Memory (BiLSTM).
• Implemented a regularized CNN model and explored different optimization and regularization techniques.
• Evaluated models using standard performance metrics including accuracy, precision, recall, F1 score, and confusion matrix.

Main Results:

• One proposed hybrid deep learning model achieved approximately 98% prediction accuracy.
• The developed models demonstrated improved accuracy, reduced generalization error, and lower computational complexity compared to existing methods.
• Experimental results validated the efficiency and effectiveness of the proposed SER system.

Conclusions:

• The proposed emotion-aware IoT-enabled WBAN system effectively addresses challenges in healthcare SER.
• Hybrid deep learning models, particularly the CNN-BiLSTM combination, show significant promise for accurate and efficient emotion recognition.
• The system enables real-time prediction and captures emotional changes before and after treatment, enhancing patient monitoring.