Apr 2023 - May 2024
Apr 2023 - May 2024 The Upper Gastrointestinal (UGI) multidisciplinary Multidisciplinary team (MDT) makes critical treatment decisions in every oesophageal cancer patient’s journey (e.g., neoadjuvant therapies, expedited surgery or palliation). This has a profound impact on survival and quality of life. Recent studies have highlighted that rising caseloads and time pressures can lead to inconsistent and suboptimal decisions. Machine learning (ML) offers the potential to standardise and deliver data-driven decisions. We will mirror the decision-making process by incorporating human-selected critical variables and work with practitioners to assess regulatory and practical challenges (as human-AI partnerships) to ensure that the solution is trustworthy for routine clinical use.
1. We set out to improve physical and mental well-being by focusing on improvements in the consistency of cancer care decision-making for all Oesophageal adenocarcinoma (OAC) patients; treatment decisions confer significant implications for patient and relatives’ quality of life (mental and physical).
2. We seek to ensure that pivotal treatment decisions are data-driven and consistent so that all patients in equivalent circumstances are treated fairly and thereby promote justice and health equality.
3. Finally, an explainable AI approach provides a unique opportunity to uncover subconscious biases within the way we currently make cancer treatment decisions. This in turn allows us to challenge preconceived biases or subsequent inequality that may result and allows for the creation of an inclusive, fair and just world. Once implemented, our solution will seek to automate and streamline workflows with a direct impact on the efficiency of the OAC MDT, and in a wider sense, the health economy.
To do this we will analyse the UGI cancer MDT at UHS (a busy tertiary centre) to collect decision-making factors. This includes clinical and patient factors (age, comorbidities, cTNM, performance status) and human factors (e.g., types of MDT members present, MDT lead). Using retrospective data we will develop a prototype mirror ML model which integrates clinical, radiological, histological, and social variables to predict outcomes. We will utilise our prototype mirror ML model to conduct semi-structured interviews with physicians (chosen from AUGIS/UAOS/BSG/UKIOG) and present decision scenarios to discuss AI acceptability. This will show how users engage with our model and reveal crucial technical, cultural, psychological, and socio-economic elements affecting human-AI teaming.
Find out more
You can read more about this project through two blogs:
2.Enhancing Decision Support in Oesophageal Cancer Treatment
NIHR Clinical Lecturer in Interventional Radiology, University of Southampton
Professor of Gastrointestinal Surgery, University of Southampton
Professor of Artificial Intelligence, Director of TAs Hub, University of Southampton
Associate Professor in Translational Epigenomics, University of Southampton
Lecturer of AI and Data Science, University of Bournemouth
Lecturer in Electronics and Computer Science, University of Southampton
Research fellow at Agents, Interaction and Control (AIC), University of Southampton
Professorial Fellow in Information Technology, University of Southampton
Consultant GI Radiologist, University Hospital Southampton
Professor of Mental Health and Digital Technology, University of Nottingham
