AI-Powered Neurogenetics: Supporting Patient’s Evaluation with Chatbot

Abstract

BACKGROUND/OBJECTIVES: Artificial intelligence and large language models like ChatGPT and Google's Gemini are promising tools with remarkable potential to assist healthcare professionals. This study explores ChatGPT and Gemini's potential utility in assisting clinicians during the first evaluation of patients with suspected neurogenetic disorders. METHODS: By analyzing the model's performance in identifying relevant clinical features, suggesting differential diagnoses, and providing insights into possible genetic testing, this research seeks to determine whether these AI tools could serve as a valuable adjunct in neurogenetic assessments. Ninety questions were posed to ChatGPT (Versions 4o, 4, and 3.5) and Gemini: four questions about clinical diagnosis, seven about genetic inheritance, estimable recurrence risks, and available tests, and four questions about patient management, each for six different neurogenetic rare disorders (Hereditary Spastic Paraplegia type 4 and type 7, Huntington Disease, Fragile X-associated Tremor/Ataxia Syndrome, Becker Muscular Dystrophy, and FacioScapuloHumeral Muscular Dystrophy). RESULTS: According to the results of this study, GPT chatbots demonstrated significantly better performance than Gemini. Nonetheless, all AI chatbots showed notable gaps in diagnostic accuracy and a concerning level of hallucinations. CONCLUSIONS: As expected, these tools can empower clinicians in assessing neurogenetic disorders, yet their effective use demands meticulous collaboration and oversight from both neurologists and geneticists.

Ähnliche Arbeiten