Sentiment Analysis of Doctor’s Responses to Patient Inquiries in a Medical Chatbot: A Logistic Regression Approach

Abstrak

This study addresses the challenge of improving doctor-patient communication in medical chatbot systems by integrating sentiment analysis to classify doctor responses as positive or negative. The primary objective was to develop a model that enhances the emotional intelligence and appropriateness of chatbot interactions using Logistic Regression. The model achieved 98.63% accuracy, 99.68% precision, 95.90% recall, and 97.75% F1-score, demonstrating its high reliability in classifying sentiments with minimal misclassifications. While the model performs well, further improvements could focus on reducing false negatives to increase recall. The implications of this research are significant for digital healthcare, as the model enables chatbots to provide more empathetic, context-aware responses, improving patient engagement and overall communication. The novelty of this study lies in applying sentiment analysis within medical chatbot systems, contributing to the growing field of emotional intelligence in digital healthcare. The findings highlight the potential of sentiment analysis to enhance patient interactions, making medical chatbots more effective and human-like. This study provides a solid foundation for further advancements in healthcare chatbots, demonstrating the potential of machine learning to improve the quality of doctor-patient communication in a digital context.