Developing explainable AI models for insights to support lung cancer management

Opportunity

Cancer often comes recurs or spreads to other parts of the body (metastasises), which is a major reason treatments don’t work as well and survival rates are lower. In Australia (2024), lung cancer represents 8.9% of new cancer diagnoses and 16.9% of cancer deaths. Predicting a person’s outlook is difficult – clinicians can underestimate or overestimate survival – so better ways to support decision-making are needed. This project will use “explainable” AI to find patterns linked to lung cancer outcomes, including whether cancer returns, spreads, and how long people survive. The goal is to build reliable data models that clinicians and researchers can understand and interrogate. The work will use linked health datasets (demographics, tumour details, spread of disease, treatments, follow-up and outcomes) to answer clinically relevant questions about lung cancer care.

Project Objectives

• Use AI methods that can clearly explain their results to help support data-driven clinical decisions.
• Build strong, reliable models to improve understanding of lung cancer outcomes – especially treatment results, cancer spread, cancer returning, and survival.
• Make the models easier to interpret by showing which factors matter most and how they influence the predictions.