Analysing the Patient Sentiments in Healthcare Domain Using Hybrid Deep Learning Model

Abstract

Patient sentiment analysis in healthcare plays a vital role in assessing service quality and improving
patient satisfaction. Traditional machine learning techniques often struggle to interpret domain-
specific medical language and subtle expressions of sentiment. Sentiment analysis in healthcare has
gained significant attention due to its potential to improve patient satisfaction, clinical outcomes, and
mental health monitoring. Existing methods primarily focus on rule-based and machine learning
approaches, which often fail to capture complex emotional nuances and contextual sentiments in
patient feedback. Additionally, traditional sentiment analysis models lack explainability, making it
difficult for healthcare providers to interpret the reasons behind sentiment classifications. This
research aims to develop an advanced sentiment analysis framework integrating deep learning models
with explainable AI (XAI) techniques to enhance the accuracy, interpretability, and reliability of
patient sentiment classification. This paper proposes a hybrid deep learning model that combines
Convolutional Neural Networks (CNN), Bidirectional Long Short-Term Memory (Bi-LSTM), and
attention mechanisms to classify patient sentiments from healthcare reviews and clinical notes.
Experiments on a dataset of 50,000 patient reviews demonstrate that the hybrid model outperforms
conventional machine learning and single deep learning models, achieving a high accuracy of 92.3%
and providing more nuanced insights into patient experiences.