AI aces diagnosis of chronic ataxias

A bespoke virtual assistant based on artificial intelligence (AI) has outperformed neurologists in the diagnosis of chronic ataxias in a recent study published in Movement Disorders. Lucas Alessandro and colleagues developed a chatbot that could respond to queries by retrieving information from a database and using generative AI to maximize the relevance and clarity of responses. The diagnostic accuracy of the virtual assistant was 91%, whereas 21 neurologists achieved an accuracy of just 18% and the GPT-4 large language model achieved 19%. The tool could, therefore, be a valuable tool for supporting diagnosis, and similar virtual assistants could be developed for other neurological disorders.

Comments (0)

No login
gif