Caso di studio sull'applicazione dell'Intelligenza Artificiale (AI) alla Medicina Narrativa per la scrittura di un breve racconto sulla Leptospirosi

This Narrative Medicine work describes the drafting of a short story on Leptospirosis by a ChatGPT Artificial Intelligence application. The bot was asked to invent a short story for non-expert readers to inform them about this zoonosis, at the same time giving them tips to avoid contracting the infection. The work started from a documentation collected in real infectious cases, by the Serology Laboratory of the Experimental Zooprophylactic Institute of Umbria and Marche (Italy). The initiative is part of the Ministry of Health collaborative writing activities Research Project: Micro Epidemic One Health. ChatGPT acquired the request to write a story of zoonosis, that face the proposed theme of Leptospirosis, reacting tof editors inputs and integrating medical information gradually over time. An important aspect is that interaction, after a simple subscription to the service, was possible through a Web interface without complex procedures and for free. The dialogue between requestor and bot was conducted in natural language, resembling what would have been established between a writer and an Editor. Illustrations, also generated by artificial intelligence (using DALL-E, OpenAI) have been included in this narrative on Leptospirosis. Generic images (boys bathing in the river, mice lapping at the water), have been processed satisfactorily. More specific ones, such as microphotographs of leptospira bacteria, were not produced correctly. The analysis of the text produced by Veterinarians has shown that the story is substantially correct, even though a little bit approximate. In it, for example, it seems not to take into account the difficulties of reclaiming the entire river for Leptospirosis On the linguistic level ChatGPI is able to operate really well. The linguist in charge to evaluate this work, points out that the english text generated by ChatGPT (using a bio-medical microlanguage) is rather appropriate and far better than the one we may have, for example, using Google Translator. Conclusion. The value of the software is relevant when we consider the simplicity of the human / machine interaction required. ChatGPT was able to generate a true narration regarding a case of Leptospirosis, furthermore providing in real time medical insights, on request. This bot can interactively translate , organize abstracts, explore a topic with the help of (non-scientific) literature. Many open questions remain in terms of "interaction of the project" with the general public. Today, with ChatGPT, new scenarios are emerging in which the interaction between man and machine seems to evolve to the point of simulating an effective editorial collaboration between autonomous subjects