Inspiring Health
Transforming Care

Data Mining & Machine Learning

Currently only 2-3% of cancer patients receiving radiotherapy are involved in clinical trials, hence only this proportion of patient data is used to improve care of future patients. However cancer centres record lots of electronic-data on all patient treatments. This project’s aim is to use anonymised electronic medical records and images to learn mathematical models that predict treatment outcomes. This can then help doctors and patients to select the best approach. In addition, this project also explores new methodologies for mathematically extracting descriptive features of a tumour from images, termed ‘radiomics’, which can also be used to help clinical decisions in the same way as the clinical data.

Key investigators

  • A/Prof Lois Holloway
  • Prof David Thwaites
  • Prof Andre Dekker
  • Dr Matthew Field
  • Dr Mohamed Barakat

Current Research Opportunities

Rapid Learning Proof-of-Concept

Contact:Matthew Field and Mohamed Samir Barakat

Aim:To develop a NSW radiotherapy network for training outcome prediction models, pertaining to survival from non-small cell lung cancer, in a distributed manner.

The Effect of Imputing Missing Clinical Attribute Values on Decision Support Systems

Contact:Matthew Field and Mohamed Samir Barakat.

Aim:To examine the impact of accounting for missing data items when constructing clinical decision aids for radiotherapy treatment.

Developing and validating a survival prediction model for early death in NSCLC patients following curative-intent (chemo) radiotherapy

Contact:Matthew Field and Mohamed Samir Barakat.

Aim:To develop and test models that predict survival from non-small small cell cancer after different (chemo) radiotherapy treatment regimes in local and international clinics.

Contact Us to Get Involved

Matthew Field

02 8738 9220