Patient Satisfaction Feedback Analysis: Facial Emotion Recognition Using Convolutional Neural Network

Abstract

Telehealth has become very crucial for remote healthcare services, yet measuring patient satisfaction poses a snag. This study introduces a feedback-enabled telehealth system that adopts Convolutional Neural Networks (CNNs) for real-time facial emotion recognition. Using the CK+ dataset, a MobileNet-based CNN was trained to categorized patient emotions during remote consultations. The model achieved 65.78% accuracy, with performance affected by imbalanced class distributions across facial expressions. These results demonstrate the feasibility of CNN-based emotion recognition as a tool for having real-time access to patient satisfaction in telehealth. Future work will be in the direction of improving the accuracy of the result through balanced datasets and applying privacy-preserving techniques such as federated learning. This research contributes to advancing patient-centric telehealth systems by integrating AI-driven feedback mechanisms into remote healthcare services.