Artificial intelligence-based decision-making: can ChatGPT replace a multidisciplinary tumour board?

Abstract

Artificial intelligence (AI) has been around for a while.Recent reports 1,2 have evaluated its role in assisting clinical decision-making, with promising results.After its launch in 2022 by OpenAI (San Francisco, California, USA), ChatGPT, the AI based on a large language model, has also been investigated as a tool, even in the clinical field 3 .Multidisciplinary tumour boards (MDTs) make decisions on oncological cases based on the knowledge of different specialties.These meetings can last many hours, consuming valuable specialists' time, and possibly contributing to their burnout 4 .The aim of this study was to assess the concordance between ChatGPT and the authors' MDT with regard to the therapeutic recommendation for patients with colorectal cancer.A cross-sectional study was undertaken that included 157 consecutive patients, aged between 18 and 75 years, with colorectal adenocarcinoma at any AJCC TNM stage, who were presented to the authors' MDT.Local ethics committee approval was obtained.Because ChatGPT was trained using information updated until September 2021, only patients diagnosed from 2020 were included.The clinical case summary submitted to the MDT was uploaded to ChatGPT 4.0 using a prompt designed to suggest therapy for each patient individually according to clinical guidelines available up to 2020.Analysis was performed separately according to preoperative and postoperative scenarios.Decisions made by ChatGPT and the MDT were categorized as follows: 'adjuvant' or 'follow-up' for postoperative scenarios, and 'surgery', 'neoadjuvant' or 'palliative' for preoperative ones.Concordance analysis was done using the κ coefficient for the

Ähnliche Arbeiten