Emotion- and Sentiment-Aware Healthcare Text Analytics for Sustainable and Inclusive Health Systems

Electronic health records, patient feedback systems, social media, telemedicine, and online healthcare forums are all sources of large quantities of unstructured textual data that are created by healthcare organizations. Currently, traditional healthcare analytics approaches mainly focus on clinical indicators and ignore emotion/sentiment information contained within textual communication. In this paper, a new Emotion- and Sentiment-Aware Healthcare Text Analytics (ESHTA) framework is proposed to provide sustainable and inclusive healthcare systems through integrating natural language processing, machine learning, and emotion recognition technologies. The proposed framework preprocesses healthcare related text, classifies the sentiment, detects the emotion, extracts the healthcare topic, and performs decision-support analytics. A series of benchmark healthcare review datasets and patient feedback records were used for experimental evaluation. The proposed framework was able to perform sentiment classification and emotion recognition with an accuracy of 92.4% and 89.1% respectively, surpassing conventional machine learning approaches.