Probabilistic emotion and sentiment modelling of patient-reported experiences

Abstract: This study introduces a novel methodology for modelling patient emotions from online patient experience narratives. We employed metadata network topic modelling to analyse patient-reported experiences from Care Opinion, revealing key emotional themes linked to patient-caregiver interactions and clinical outcomes. We develop a probabilistic, context-specific emotion recommender system capable of predicting both multilabel emotions and binary sentiments using a naive Bayes classifier using contextually meaningful topics as predictors. The superior performance of our predicted emotions under this model compared to baseline models was assessed using the information retrieval metrics nDCG and Q-measure, and our predicted sentiments achieved an F1 score of 0.921, significantly outperforming standard sentiment lexicons. This method offers a transparent, cost-effective way to understand patient feedback, enhancing traditional collection methods and informing individualised patient care. Our findings are accessible via an R package and interactive dashboard, providing valuable tools for healthcare researchers and practitioners.

• The paper presents a probabilistic emotion recommender system using a naive Bayes classifier that achieved an F1 score of 0.921.
• The methodology employs metadata network topic modeling to decode key themes from patient narratives on a healthcare review website.
• The findings reveal that positive sentiments are linked to quality interactions and educational programs, while negative sentiments relate to administrative issues.

Overview of the Study

The integrity of healthcare systems relies heavily on patient feedback to ensure that medical services meet public expectations. Traditionally, this feedback has been collected through structured surveys and focus groups, which have their own set of limitations. A revolutionary approach has been introduced to analyze patient emotions and sentiments through patient-reported feedback using probabilistic modeling, offering an alternative to conventional methods.

Analyzing Patient Feedback

Researchers have utilized metadata network topic modeling to decode key themes in patient narratives collected from Care Opinion, a healthcare review website. This analysis uncovered significant relationships between patients' emotions and their experiences, particularly highlighting the influence of patient-caregiver interactions on emotional responses. The research team has observed that positivity in patient feedback is often tied to educational programs and quality interactions with healthcare personnel, while negative sentiments are associated with feelings of neglect and administrative concerns.

The Emotion Recommender System

To efficiently capture and analyze patient sentiment, the study presents a probabilistic emotion recommender system. This system, capable of predicting multi-label emotions and binary sentiments, was developed using a naive Bayes classifier. With a remarkable F1 score of 0.921, the system outperforms standard sentiment lexicons in interpreting patient feedback. It is now accessible via an R package and an interactive dashboard for use in further research and clinical practice.

Enhancing Patient-Centered Care

This novel methodology transforms unstructured patient-reported feedback into valuable insights. By integrating patient emotions with existing experience measures and incorporating them into surveys, healthcare providers can deliver more personalized and effective care. The transparent and interpretable approach can reshape how healthcare experiences are understood and pave the way for a profound improvement in the quality of healthcare delivery.