Machine Learning methods for cancer risk prediction of upper gastro-intestinal cancers

Project Details

This project aims to develop machine learning and advanced statistical methods for the prediction of upper gastro-intestinal (GI) cancers in Australian primary care patients.

We will start by reviewing existing literature on risk prediction methodologies that use machine learning models applied to electronic health records data sources and identify which are best suited to our data. The primary care data in PATRON is extremely rich but in most studies, usually only a small subset of information is used to carry out predictive analyses, for example specific blood tests or given symptoms. Furthermore, these analyses are often done using information at a single point in time to make predictions.

In this project we aim to make more complete use of the trends in clinical information in the lead up to a cancer diagnosis in order to improve on existing approaches to risk prediction for upper GI cancers. A secondary outcome of this study will therefore be to provide further evidence on the suitability of machine learning for risk prediction using electronic health records.

Patron ID: PAT1044_5

Project Lead: Dr Alex Lee