Author response: Concordance of ChatGPT artificial intelligence decision-making in colorectal cancer multidisciplinary meetings: retrospective study

Correspondence to: Stephan Bischofberger (e-mail: stephan.bischofberger@kssg.ch)

Department of Surgery

Kantonsspital St. Gallen

St. Gallen

Switzerland

BJS Open, https://doi.org/10.1093/bjsopen/zraf040, published 07 May 2025

Thank you for the comments from Drs Daungsupawong and Wiwanitkit and the valuable input on our study1. Our study is one of the first studies with a larger number of cases that shows the possibilities of using a freely available AI tool like ChatGPT to support physicians in a multidisciplinary team meeting (MDT). The aim of our study was not to incorporate the recommendations of the used AI tool into treatment conception, but merely to correlate the recommendations of both the AI tool and the MDT. In our opinion, the retrospective design therefore has no disadvantages compared to prospective recruitment due to the consecutive patient selection in the selected study period.

As can be seen from the methodology, ChatGPT had exactly the same information available as the MDT, although ChatGPT only had the findings and not the images to assess (i.e. less information than the MDT). Figure 1 shows the prompt with the assessments of the original examinations. It will be exciting as soon as image data (e.g. MRI, CT) can also be interpreted by an AI tool.

You also raise very good points for discussion regarding the general use of AI systems (automation, resource limits, costs). In our opinion, we are in a development process that opens up many opportunities to further develop medical work. Geographical and socio-economic differences certainly exist in the various healthcare systems. Particularly when resources are scarce, such systems can possibly relieve the burden on doctors, which could save resources both in terms of personnel and financially. It should be noted that the ChatGPT version used is available free of charge.

As already mentioned, our study is an initial pilot study that does not aim to bring about a paradigm shift in the implementation of MDT, but rather to demonstrate the astonishing possibilities of a generally accessible AI tool such as ChatGPT. Of course, the next step would be a prospective intervention study, ideally in a multicentre setting. Such a study is already underway and we hope to be able to show this data in the course of time.