ScholarLens Back to search
2023•Research SquareOpen access

Evaluating the Ability of ChatGPT in Generating Differential Diagnosis in Clinicopathological Conferences of Neurodegenerative Disorders

Shunsuke Koga, Nicholas B. Martin, Dennis W. Dickson

Open full text 2 citations

Abstract

Abstract This study explores the utility of the artificial intelligence model ChatGPT in predicting neuropathologic diagnoses from clinical summaries presented at clinicopathological conferences. Twenty-five cases of various neurodegenerative disorders were analyzed using ChatGPT versions 3.5 and 4. The primary neuropathological diagnoses predicted by the models aligned with the official neuropathologic diagnoses in 32% (ChatGPT-3.5) and 52% (ChatGPT-4) of cases. When considering all proposed diagnoses, accuracy increased to 76% and 84% for ChatGPT-3.5 and ChatGPT-4, respectively. These findings highlight the potential of artificial intelligence tools like ChatGPT in neuropathology, suggesting they may facilitate more comprehensive discussions in clinicopathological conferences.

Open-access reader

About this research paper

What this paper is about

Abstract This study explores the utility of the artificial intelligence model ChatGPT in predicting neuropathologic diagnoses from clinical summaries presented at clinicopathological conferences. Twenty-five cases of various neurodegenerative disorders were analyzed using ChatGPT versions 3.5 and 4. The primary neuropathological diagnoses predicted by the models aligned with the official neuropathologic diagnoses in 32% (ChatGPT-3.5) and 52% (ChatGPT-4) of cases. When considering all proposed diagnoses, accuracy increased to 76% and 84% for ChatGPT-3.5 and ChatGPT-4, respectively. These findings highlight the potential of artificial intelligence tools like ChatGPT in neuropathology, suggesting they may facilitate more comprehensive discussions in clinicopathological conferences.

Why it matters

OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

Abstract This study explores the utility of the artificial intelligence model ChatGPT in predicting neuropathologic diagnoses from clinical summaries presented at clinicopathological conferences. Twenty-five cases of various neurodegenerative disorders were analyzed using ChatGPT versions 3.5 and 4. The primary neuropathological diagnoses predicted by the models aligned with the official neuropathologic diagnoses in 32% (ChatGPT-3.5) and 52% (ChatGPT-4) of cases. When considering all proposed diagnoses, accuracy increased to 76% and 84% for ChatGPT-3.5 and ChatGPT-4, respectively. These findings highlight the potential of artificial intelligence tools like ChatGPT in neuropathology, suggesting they may facilitate more comprehensive discussions in clinicopathological conferences.

Key concepts: Differential (mechanical device), Differential diagnosis, Medicine, Pathology, Neuroscience, Psychology, Physics, Thermodynamics

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
Evaluating the Ability of ChatGPT in Generating Differential Diagnosis in Clinicopathological Conferences of Neurodegenerative Disorders — Research Paper | ScholarLens