Abstract—As patient-generated content becomes more widespread on platforms such as WebMD, sentiment analysis has proven to be an effective approach for capturing user experiences with medications. This research conducts a comparative assessment of classical Machine Learning (ML) techniques versus Deep Learning (DL) approaches for categorizing drug reviews, utilizing a newly proposed composite metric—the Drug Effectiveness-Satisfaction Score (DESS)—calculated as the average of user-rated effectiveness and satisfaction. A diverse set of ML algorithms (Random Forest, Support Vector Machines (SVM), XGBoost) and DL architectures (Long Short-Term Memory (LSTM), Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM), Bidirectional Encoder Representations from Transformers (BERT), and DistilBERT) were tested using multiple feature extraction and embedding methods, including CountVectorizer, Term Frequency-Inverse Document Frequency (TF-IDF), Global Vectors for Word Representation (GloVe), and transformer-based embeddings. The results indicate that BERT attained the highest F1-Score at 87.22%, whereas eXtreme Gradient Boosting (XGBoost) demonstrated a good trade-off between accuracy and computational cost. Despite robust results, performance was slightly below benchmarks reported in other domains, likely due to the complexity and ambiguity introduced by the DESS metric and the unstructured nature of real-world medical reviews. This work underscores the potential of combining ML and DL for healthcare sentiment analysis and highlights future opportunities in domain-specific fine-tuning, ensemble modeling, and explainable AI.
 
Keywords—Natural Language Processing (NLP), user feedback, sentiment analysis, healthcare data, word embeddings, pharmacovigilance, text classification, Bidirectional Encoder Representations from Transformers (BERT)


Copyright © 2026 by the authors. This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited (CC BY 4.0).