Evaluating clinical AI in real-time

The proposed research project is a component of the NASCENT Medical Research Future Fund project to develop and validate a research infrastructure that enables researchers at Australian hospitals to trial and evaluate AI-based clinical systems that provide diagnostic, treatment and prognostic support to clinicians. A key hurdle for implementing AI within Australian hospitals is a real-time trial using real patient data to ensure the AI is fit for clinical use. The proposed project will focus on research that helps to overcome this hurdle. In particular, the PhD student will support the evaluation of two use cases for predicting clinical deterioration and sepsis in hospital wards.  The evaluation trials will be conducted at multiple hospital sites around Australia with the objective of demonstrating suitability for implementation into clinical practice.

The student will have the opportunity of identifying the final scope of their research at the beginning of their candidature. During that phase, they will research current state-of-the-art methods for evaluating clinical AI in real-time environments and can propose a suitable research topic to tie in with the use-case evaluations over the life of their candidature. Examples of such topics could include real-time evaluation methods that incorporate clinical and patient feedback, drift monitoring and updating, automated and real-time AI improvement modelling or specific AI technique evaluation, such as clinical LLM use and evaluation.  In the second stage of the project the student will develop methods to use during the trials and in the final phase of the project the student will trial their methods and report on the findings.

You will work out of our RBWH (Herston) campus and will have access to our large computing centre.