Enhancing Medical Text Classification and Disease Diagnosis with BERT Advances in Deep Learning for Healthcare

Abstract

This study explores the use of deep learning models, particularly BERT, for medical text classification and disease diagnosis. The research aims to evaluate the performance of different models, including CNN, Transformer, and BERT, in terms of their ability to accurately classify medical texts. Experimental results show that BERT outperforms other models across all key metrics, including accuracy, recall, precision, and F1-score. The model's ability to effectively capture long-range dependencies and complex semantic structures in medical data plays a crucial role in achieving superior classification performance. The findings suggest that BERT, despite its high computational cost, is a promising tool for improving medical text classification, with potential applications in clinical decision support and disease prediction. Further research on optimizing BERT for real-time applications will be necessary to enhance its practical applicability.