Extraction of Disease Symptoms from Free Text Using Natural Language Processing Techniques

Natural language processing (NLP) has emerged as a transformative technology in the field of medicine, offering innovative solutions to the challenges of efficiently extracting, analyzing and using information from the vast amounts of textual data generated in healthcare systems. NLP techniques enable the automated extraction of valuable information from clinical notes, medical literature and other textual sources, facilitating tasks such as clinical decision support, information retrieval and patient data management. With NLP, healthcare professionals can promote the integration of the patient as a key contributor to improving diagnosis accuracy through the expression of his state of health using free text. Such an approach could provide researchers and healthcare professionals with a wealth of actionable information, helping to reshape modern medicine. However, the adoption of NLP in medicine also presents challenges in terms of data quality and model interpretability. This paper discusses the application of NLP on medical prescriptions of all kinds, from simple notes to pure free text. The main objective is to be able to extract symptoms which will then be used to feed the classification model responsible for generating medical predictions. Experimental results show that the proposed method, combined with appropriate word processing, is highly effective in extracting medical vocabulary from even unstructured free text. Performance measurements have given very good values, demonstrating the usefulness of such an approach in modern medical diagnosis. This kind of diagnosis allows patients to freely express their feelings and daily experiences without any intervention, which could improve the accuracy of the diagnosis.