Abstract

The integration of the Internet of Things (IoT) and artificial intelligence (AI) in healthcare is transforming patient care by enabling real-time monitoring and personalized treatment. Facial expression recognition (FER) plays a vital role in identifying emotional states, which can improve mental health diagnosis and patient engagement. However, existing FER systems face limitations in accuracy and responsiveness due to insufficient data processing capabilities and lack of adaptive learning models. This study proposes an IoT-driven FER system using a Convolutional Neural Network with Attention Mechanism (CNN-AM) to enhance emotion detection accuracy and system adaptability. IoT devices, including wearable sensors and smart cameras, capture real-time facial data, which is processed using the CNN-AM model to identify emotional states. The attention mechanism improves the model''''s ability to focus on critical facial features, reducing false detection rates. The system was tested on the FER-2013 dataset, achieving a recognition accuracy of 95.6%, outperforming existing methods such as Support Vector Machine (SVM) and ResNet by 3.2% and 2.1%, respectively. Results demonstrate that the proposed model enhances both detection speed and accuracy, offering a scalable and efficient solution for personalized healthcare in Industry 5.0.