Machine learning to assist oesophageal cancer surgeons

New research in Southampton is using a branch of artificial intelligence (AI) to help doctors make complex decisions around patient care.

The project, led by Professor Tim Underwood, is seeking to create a digital model that can assist doctors treating cancers of the food pipe (oesophagus) and stomach.

It forms the basis for Navamayooran Thavanesan’s PhD studentship, jointly funded by University Hospital Southampton (UHS) and the University of Southampton’s Institute for Life Sciences.

Predicting doctors’ decisions

Machine learning tools can find complex patterns in large amounts of data that humans might struggle to see.

This has the potential to be helpful for surgeons, who need to consider large amounts of information when making decisions around patient care, such as whether to operate to remove a tumour.

The researchers intend to use machine learning to create a digital model that can ‘learn’ from the decisions made by surgeons in the Upper Gastrointestinal Surgery Multidisciplinary Team at UHS.

The hope is this model will then be able to correctly predict the decisions they go on to make.

Spotting patterns

The model will use information on patients from their weekly meetings to analyse patterns in how they make decisions, such as whether to operate.

It will use details such as the patient’s age, where the cancer is and how advanced it is, images from scans, if they smoke or drink alcohol, and how far away they live from their nearest hospital.

The project is supervised by Prof Tim Underwood, Professor of Gastrointestinal Surgery at UHS, and Dr Ganesh Vigneswaran, NIHR Clinical Lecturer in Interventional Radiology.

Prof Underwood said: “Machine learning has a huge potential to help us work more efficiently, allowing us to treat more patients and ensure those who need surgery get it at the right time. This is especially important right now, as we are seeing a rise in new oesophageal cancer diagnoses.”