An ensemble-based sentiment analysis approach for precision medicine recommendation.

The rapid growth of online healthcare platforms has resulted in an unprecedented volume of patient-generated medical reviews, creating new opportunities for the development of intelligent and personalized medicine recommendation systems. However, extracting clinically meaningful insights from such heterogeneous and unstructured textual data remains a significant challenge. In this study, we propose a context-aware medicine recommendation framework that integrates patient reviews with demographic and clinical attributes, including age, gender, and medical condition, to enhance recommendation precision. The proposed framework combines conventional machine learning models, including Logistic Regression, Naïve Bayes, K-Nearest Neighbors, and Support Vector Machine, with deep learning (DL) architectures such as Simple Neural Network (SNN), Convolutional Neural Network (CNN), and Recurrent Neural Network with Long Short-Term Memory (RNN_LSTM). To further improve predictive performance, a novel stacked ensemble framework, termed SCL-MedStacker, is introduced, in which the DL models serve as base learners and a Random Forest classifier is employed as the meta-learner. To address class imbalance in the medical review dataset, random oversampling is applied prior to model training. Experimental results demonstrate that the proposed framework achieves an accuracy of 92%, outperforming existing state-of-the-art methods. Furthermore, the personalized recommendation mechanism exhibits improved contextual relevance and predictive reliability, highlighting its potential to support clinical decision-making, reduce prescribing errors, and enhance treatment effectiveness in real-world healthcare environments.