DeepSeek: Another step forward in the diagnosis of oral lesions
Márcio Diniz Freitas, Pedro Diz Dios
Abstract
Artificial intelligence (AI) is increasingly being explored as a tool for medical diagnosis, particularly in fields with limited specialized training, such as oral medicine. This study evaluates the performance of DeepSeek-R1, an open-source large language model (LLM), in diagnosing oral diseases and conditions using text-based case descriptions from the New England Journal of Medicine's "Image Challenge." Results indicate that DeepSeek-R1 achieved a diagnostic accuracy of 91.6 %, slightly outperforming ChatGPT-4o (88.9 %) and significantly exceeding the 47.8 % accuracy of the journal' readers. While DeepSeek lacks direct image interpretation capabilities, it demonstrates high proficiency in textual diagnostic tasks. These findings suggest that DeepSeek-R1 could be a valuable aid for medical professionals in diagnosing oral diseases.