Quality Prediction of Health Articles Based on the BERT-WWM Model

The quality of online health information is crucial for patients' decision-making and health outcomes. This paper primarily employs machine learning models to study and implement the automatic evaluation of the content quality of health promotion articles on online medical platforms. Utilizing a dataset of real health promotion articles and community user voting data, this paper introduces the prediction of health article quality based on the Chinese BERT-WWM model, along with a comparative analysis of various other machine learning algorithms. Significant differences in the predictive performance of different models are observed, with the fine-tuned BERT-WWM model achieving the best results. It not only reached the highest accuracy but also attained a recall rate of 62% for high-quality articles, while the worst-performing model achieved only 31%. This deep analysis of health promotion articles on online medical platforms using deep learning models contributes to the advancement of smart healthcare.