Using AI to help with quality assurance in surgical studies

Surgery improves and saves millions of lives each year. Many operations are tested in randomised controlled trials (RCTs), which compare treatments to find out which works best. However, surgical RCTs can be difficult to design and run.

Surgeons often have different preferences about how to perform an operation, and they may have different levels of experience. This means the same procedure can be done in slightly different ways or to different standards. These differences can affect trial results and may introduce bias, which can reduce the reliability of the findings and impact patient care.

One way to reduce this problem is to include quality assurance (QA) in surgical studies. QA involves agreeing in advance what the key steps of an operation should be and checking that they are carried out properly, often by reviewing video recordings. However, reviewing surgical videos is slow, costly, and labour-intensive. As a result, many studies do not include QA processes.

Project aims

This project will explore whether artificial intelligence (AI) can help analyse surgical videos more quickly and efficiently. We will use videos of two operations for treating gastro-oesophageal reflux.

First, we will create a clear system for labelling three key steps of the operation, including which important anatomical structures should be visible. If the structure is clearly seen, this suggests the step has been completed properly. These labelled videos will then be used to train and test an AI model, working with computer science experts.

What we hope to achieve

We hope to show that AI can support QA in surgical trials, making the process faster and less expensive.

If our project is successful, we plan to extend our AI model to be able to analyse more complex videos, for example where something unexpected happens. Eventually we hope to run a larger study, to find out whether AI can be used for QA of other surgical procedures and in early phase studies, to help researchers understand how new surgical techniques develop over time and whether they are safe.

We will also create a free library of annotated surgical videos to support future research.

In the long term, this could improve surgical standards, strengthen research, and lead to better outcomes for patients.