Abstract
This study presents a new approach to the optimization of Natural Language Processing (NLP) techniques for medical entity recognition and disease classification. By leveraging patient queries and PubMed article abstracts, the research uses advanced extraction methods to identify biomedical entities and diseases from medical texts. Diseases are grouped using a combination of TF-IDF and K-means clustering, and classification models are then applied to predict disease clusters based on known entities. A key innovation of this work is the use of Stochastic Dynamic Optimization to fine-tune parameters, significantly enhancing clustering and classification performance. Experimental results demonstrate that the proposed method improves the accuracy of extraction and classification, outperforming traditional methods in terms of precision and scalability. This scalable and efficient approach to biomedical data analysis has the potential to support future clinical decision-making, enable personalized medicine, and provide valuable healthcare insights, ultimately contributing to improved patient outcomes and more effective research workflows.