Improving lesion recognition using routinely collected endoscopic and pathology data Guy’s and St Thomas’ NHS Foundation Trust
- Run by Guy’s and St Thomas’ NHS Foundation Trust.
- Aimed to improve lesion recognition by gastroenterology department endoscopists and therefore increase the early detection of cancer and other diseases.
- Developed a framework that allows data linkage between routinely collected endoscopy and pathology reports for individual endoscopists, with suggestions for improvement in endoscopic performance.
- Ran from August 2019 to November 2020.
To improve early detection of cancer and other diseases in gastroenterology, the quality of endoscopic investigations needs to be standardised and optimised. This requires effective training and continuous feedback on endoscopist performance.
Various performance metrics, such as colonic polyp detection, have been shown to directly reflect missed cancer rates for individual endoscopists. However, current processes for analysing these manually are laborious, poorly executed and provide no feedback on how to improve endoscopic performance.
This project by the Guy’s and St Thomas’ NHS Foundation Trust gastroenterology department aimed to introduce automated feedback on endoscopic performance to improve lesion recognition and therefore early detection.
Using the R package EndoMineR, a framework was developed to allow data linkage between endoscopy specimens, images and reports, and provide regular feedback to individual endoscopists. A number of endoscopic conditions (eg inflammatory bowel disease and eosinophilic oesophagitis) performance metrics for colonoscopy and machine learning features were added to expand the EndoMineR code base.
Key successes included integrating an analytics tool into the clinician workflow and existing IT infrastructure through close collaboration with Trust clinical informatics, and rapidly accelerating audit procedures that previously took many months. EndoMineR is now embedded into the gastroenterology department, with lesion recognition reports forming a standard part of endoscopist training.
The next steps will involve data validation and analysis to demonstrate an increase in lesion recognition amongst trainees, following an extended period of individual feedback on their endoscopies.
This project has laid the foundations for the Trust’s analytical infrastructure by demonstrating the benefits of applying data science to solve complex clinical problems, representing a valuable training tool to further expand the use of R locally and nationally.
Contact information
For more information about this project, please contact Sebastian Zeki, Consultant Gastroenterologist, Guy’s and St Thomas’ NHS Foundation Trust.
Work with us
We look for talented and passionate individuals as everyone at the Health Foundation has an important role to play.
View current vacanciesThe Q community
Q is an initiative connecting people with improvement expertise across the UK.
Find out more