Sentiment Analysis and Topic Modeling of Patient Reviews on Psychiatric Medications Using Cloud-Based Architecture

This study conducts an in-depth investigation of patient feedback regarding psychiatric medications sourced from WebMD, with the objective of exploring the relationships between customer sentiment and drug ratings, as well as identifying recurring themes within the reviews. Employing sentiment analysis models, specifically Valence Aware Dictionary for sEntiment Reasoning (VADER) and RoBERTa, the research evaluates patient sentiments, while the Bidirectional Encoder Representations from Transformers (BERT) model is utilized to uncover prevalent topics within the feedback. The analysis reveals key themes, including drug effectiveness, side effects, and ease of use. The cloud-based infrastructure of the study, implemented via Docker and deployed on Amazon Web Services (AWS), facilitates scalable and efficient data processing, rendering it suitable for the analysis of extensive datasets. The results demonstrate significant correlations between patient sentiment and overall drug ratings, highlighting the importance of patient reviews as a valuable source of real-world insights into medication efficacy and user experience. This approach establishes a robust framework for future research in patient sentiment analysis and drug efficacy, thereby contributing to a more profound understanding of patient experiences with psychiatric medications.