AI and Data Analytics for Investigating the Surgical Treatment of Rectal Cancer Patients

Abstract

The aim of this work was to compare the short- and long-term outcomes of laparoscopic versus robot-assisted surgery, as well as to establish cancer recurrence prediction models for rectal cancer patients. To achieve this aim, this work employs propensity score balancing techniques for comparative analyses and tests multiple machine learning models for developing risk prediction models in large multicentre collaborative databases.
The propensity score balanced comparative analyses showed short-term benefits to the robotic surgical approach for low anterior resections (LAR) in terms of conversion rate, restorative procedure rate, and quality of mesorectal excision, but also a longer operative time. Moreover, there were potential benefits in the form of fewer postoperative complications, particularly in the abdominoperineal excision of the rectum (APR) cohort, and shorter length of stay. The long-term analyses showed comparable three-year outcomes between the surgical approaches for both LAR and APR, with potential benefits in five-year overall survival for the LAR cohort.
Furthermore, the work established two predictive risk models for local and systemic recurrence of rectal cancer, as well as identifying potential predictors. They showed that it is feasible to build a risk prediction model for predicting recurrence in rectal cancer patients, and that these could be implemented in clinical practice, although further testing and validation are required before clinical use.
Overall, this work shows that the application of robust statistics and ML models in medicine can lead to reliable, relevant comparative analyses and produce prediction models that can have a substantial effect on patient care if handled appropriately and carefully. The combination of these approaches requires a certain level of understanding of both the medical and technical backgrounds of the work, and multidisciplinary collaborative research can help significantly improve this growing field of study and have lasting real-world benefits for patient care.

Date of Award	5 Nov 2025
Original language	English
Awarding Institution	University of Portsmouth
Supervisor	Jim Briggs (Supervisor), Jim Khan (Supervisor) & Shamsul Masum (Supervisor)