About the Project
This project will investigate sepsis prediction algorithms for both children and adults. Recognising sepsis as a global threat, Queensland Health has recently created the Sepsis Breakthrough Collaborative. One focus of this initiative is to rely on the power of data and machine learning algorithms to develop a decision tool that supports early disease detection and treatment. While training and developing robust machine learning models are critical in detecting the disease with high accuracy, it is equally important that these models are effectively integrated into clinical workflows and effectively used by health practitioners such as doctors and nurses.
This project focuses on the effective use of sepsis detection algorithms and how the algorithmic challenges can be overcome. Overall, it will address the following research question: How can sepsis prediction algorithms be effectively integrated, used by health practitioners and adapted and diffused to different clinical settings?
The research will advance knowledge on the use of algorithms for sepsis detection in practice, and how algorithmic processes and tools should be designed, developed, and integrated to maximise value for patients, doctors and other stakeholders. The aim is to develop a theoretical model of effective use for sepsis prediction algorithms, as well as expert guidance on how algorithmic work processes should be designed and managed in practice.
This project commenced in February 2022 with the recruitment of PhD researcher Krishna Dermawan who is based at The University of Queensland. Krishna will have the opportunity to collaborate with leading experts in Queensland Health to understand how clinical teams can best leverage new AI risk prediction algorithms, and work with the Sepsis Breakthrough Collaborative, a new initiative at Queensland Health, aiming to utilize machine learning algorithms for early detection of sepsis in children and adults. The project will help and support the Queensland Health team to minimize the risks and maximize the value of algorithmic decision making. In addition to helping to understand and improve the rollout and use of new risk prediction tools for sepsis, the knowledge from this project will have implications for how clinicians use a range of new digital health tools, as the sepsis case is an instance of a general trend occurring across the clinical specialties. It is supervised by CIRES Research Director and Chief Investigator Professor Marta Indulska, Dr Ida Asadi Someh, Professor Andrew Burton-Jones, and Dr Adam Irwin (Queensland Health).
This project is one of three CIRES projects with Queensland Health related to paediatric sepsis management. Further details on the other projects can be found via the project pages below:
- Expanding Data Sets to Allow Improved Critical Care for Children – Inpatient Risk Prediction
- Expanding Data Sets to Allow Improved Critical Care for Children – Outpatient Risk Prediction
Queensland Health wants to constantly improve the health of Queenslanders and the care they receive and aims to capture all relevant longitudinal data on Queenslanders to understand about their health – past, current and anticipated, treatment decisions made, and outcomes experienced. However, it is complicated to capture the right information once, curate it and access the information in an ethical and consented way that protects the individual privacy of Queenslanders. A partnership with CIRES bridges a major existing gap and represents the glue that can enable the data scientists, the clinicians and consumer community to help get this right.
Recognising that sepsis is a leading cause of preventable harm in children, Queensland Health clinicians launched the Queensland Paediatric Sepsis Program. There are three CIRES projects with Queensland Health which focus on how clinicians can best implement tools derived from transparent technical solutions to improve the recognition of sepsis in children. In one of these we will apply the power of data and machine learning algorithms to develop a decision tool that supports early detection and management of sepsis in children.