Abstract Sentiment classification of patient feedback is a critical tool for healthcare providers seeking to understand and improve patient experiences. With the increasing volume of unstructured data generated on social media platforms, effective sentiment analysis has become more challenging yet more essential. This paper presents a competitive ensemble learning approach to sentiment classification, specifically tailored to analyze unstructured patient feedback from sources like Twitter. The proposed model integrates multiple machine learning classifiers, each contributing unique strengths to the ensemble, thereby enhancing classification accuracy and robustness. By focusing on both feature extraction techniques, such as TF-IDF and Word2Vec, and model integration strategies, the study achieves superior performance over traditional sentiment classification methods. The effectiveness of the model is demonstrated through extensive experiments on datasets related to healthcare topics, including diabetes and COVID-19, as well as benchmark datasets such as IMDB and Yelp reviews. Results indicate that the competitive ensemble approach not only improves accuracy but also offers better generalization across diverse datasets, making it a powerful tool for sentiment analysis in healthcare. This research highlights the potential of advanced machine learning techniques in transforming patient feedback into actionable insights for healthcare improvement.