About the Project
In collaboration with Queensland Health, this project aims to provide a platform-independent decision support framework using an interpretable machine learning approach for making effective risk predictions for paediatric patients at risk of sepsis. These predictions will be based on multi-source, readily available, critical care data. They are intended to be a practical electronic clinician assistant and provide new insights into clinical decision-making.
The challenges of the project include effective linkage of relevant data together from different medical data sources, computational analysis of the fused representations in a real-time manner, medical interpretability formulation, uncertainty regarding the reference standard diagnosis of sepsis and integration of interpretability and accurate prediction in a joint training scheme. The significance of this project centres on the novel linkage-analysis and usage of novel transparent algorithmic development to open the black-box ML algorithms in the medical domain.
The research aims to showcase medical analysis in a data-rich environment, with the proposed model and research tasks adapted and applied to other medical predictive tasks, for instance, sepsis prediction/monitor, linking genomics with risk prediction, and so on.
About the Candidate
This is an opportunity for a highly motivated student to join the CIRES project team, collaborating closely with leading experts in Queensland Health on investigating algorithm use and organizational implications in inpatient settings. This scholarship is one of three CIRES projects with Queensland Health related to pediatric sepsis management. The successful candidate will collaborate with Queensland Health to develop a platform-independent decision support framework using an interpretable machine learning approach to make effective risk predictions for pediatric patients at risk of sepsis.
This project seeks to develop a strong and capable future leader who can undertake medical analysis in a data-rich environment. The developed model will demonstrate the flexibility to be adapted and applied to other medical predictive tasks, e.g., sepsis prediction/monitor, linking genomics with risk prediction, etc.
For this position, CIRES is seeking a candidate with an understanding of concepts from applied statistics/probability, numerical linear algebra, machine learning, algorithms and complexity. This project also requires proficiency in Python programming language, and machine learning software packages such as Pytorch. CIRES particularly encourages applicants with a medical background and relevant knowledge in the research domain.