Children’s Intensive Care Research Program

MELODY: Machine Learning in the Paediatric Intensive Care Unit, Development of Risk Prediction Models

Tailoring medical treatments to account for a patient’s individual variability is the future of healthcare. Clinical trials have demonstrated that such “precision medicine” approaches can improve clinical outcomes for patients. Recently, breakthroughs in applying computational science to healthcare have shown the promise of using computer algorithms to analyse health data and assist in clinical decision-making. For the most critically unwell children admitted to hospital, there is a lack of high-quality research for the treatments they receive—the sickest children are often excluded from high-quality trials, and thus may not receive treatments that are optimal for them. A large amount of health-related data is generated in paediatric intensive care units (PICUs), but to date, it has not been fully utilised. Using extensive, high-resolution data from >32,500 patient admissions across three large PICUs (Brisbane, Melbourne, and Zurich), this project will employ advanced machine learning techniques to develop a model that assists clinicians in detecting early deterioration in a child’s condition, enabling prompt intervention. This will result in the right treatments being delivered at the right time — reducing mortality, decreasing neurodevelopmental complications, and ultimately increasing quality of life for children admitted to intensive care.

Eqnuiries: k.gibbons@uq.edu.au 

Funding: NHMRC Investigator Grants, Ramaciotti Health Investment Partnership

Key Contacts: Prof Kristen Gibbons | Dr Moloud Abdar

Status: In Progress

LYRICAL: Evaluating an assisted decision-making tool on clinical outcomes in critically ill children

There is a paucity of high-quality trials to inform the life-saving care of children admitted to the PICU. >75% of all PICU trials have ≤100 participants, >80% recruit from a single centre, and only 3% use an innovative trial design, severely hampering the delivery of robust and generalisable results. The COVID-19 pandemic highlighted the need for agile trial infrastructure and international networks to rapidly generate gold-standard, personalised evidence to inform practice. Adult intensive care researchers demonstrated the feasibility of such an approach in a critical care setting by enrolling thousands of patients in global interventional trials using innovative designs. These learnings need to be translated to the paediatric setting to trial precision medicine approaches in a robust, efficient manner, delivering the best care for children. This project will lay the foundations for translating a machine-learning model into clinical practice. Components of the project will include conducting a systematic review; developing and deploying simulation models in clinical care; collecting and evaluating stakeholder feedback on the implementation of these models; and using implementation science to design best-practice tools to promote the highest level of successful model translation.

Enquiries: k.gibbons@uq.edu.au (Lead, Chief Investigator)

Key Contacts: Prof Kristen Gibbons | Mr Eugene Slaughter

Status: In Progress

Educational Outcomes of PICU Sepsis Survivors in Queensland

Sepsis is a dangerous condition in children caused by the body's response to infection, which can lead to organ problems and even death if not treated quickly. Children who survive sepsis may face ongoing developmental and educational difficulties and poorer long-term quality of life. To better understand these long-term effects, this project will use data from a 20-year period, linking information about sepsis survivors in pediatric intensive care units (PICU) with developmental, educational and health administrative datasets in Queensland. The results will help inform support programs for this vulnerable group, ensuring they receive the necessary help to succeed in their education. 

Enquiries: p.gilhom@uq.edu.au 

Key Contacts: Dr Trish Gilhom | Prof Kristen Gibbons

Status: In Progress

PaeNut ECMO: Nutrition Therapy in Paediatric Patients Receiving ECMO: A Prospective Multi-Site Observational Study

The PaeNut ECMO Study is a prospective, multi-site observational study investigating nutrition therapy in infants and children receiving extracorporeal membrane oxygenation (ECMO) across Australia and New Zealand.

Children requiring ECMO are among the most critically unwell patients in paediatric intensive care. However, there is limited evidence to guide optimal nutrition support during ECMO treatment. The study will examine how nutrition is prescribed and delivered throughout hospitalisation and explore the relationship between nutrition intake, feeding outcomes, growth, organ function, and recovery.

By benchmarking current practices across paediatric ECMO centres, the study aims to inform future evidence-based models of nutrition care and improve outcomes for children supported on ECMO.

Enquiries: j.woodgate@uq.edu.au 

Key Contacts: Jemma Woodgate

Status: In Progress

Building Together

Building Together is a collaborative research program focused on improving consent practices for critically ill patients and their families. In intensive care, treatments are often time-critical, and patients or families may not be able to consider research participation before urgent care begins.

The program is exploring the use of consent-to-continue, an approach that allows eligible patients to be enrolled in certain studies before consent is obtained, with consent sought as soon as possible once the immediate emergency has passed.

By working with patients, families, clinicians, researchers, and ethics committees, Building Together aims to better understand experiences of consent-to-continue and develop practical resources to support clearer communication, ethical decision-making, and meaningful family involvement in intensive care research.

Enquiries: r.lemarsney@uq.edu.au 

Key Contacts: Ms Renate Le Marsney 

Status: In Progress

SPARK-KIDS: Synergistic Partnership Advancing Research Knowledge in Paediatric Intensive Care

SPARK-KIDS  is a consumer and community partnership program that brings together researchers and people with lived experience of paediatric intensive care. By working together, the team aims to shape meaningful research that reflects the needs, priorities, and experiences of children and families. The program will also explore how to strengthen consumer partnerships in paediatric intensive care research. 

Enquiries:  Alyssa.Serratore@rch.org.au 

Key Contacts: Ms Alyssa Serratore

Status: In Progress

Project Title

Optimising Beta‑Lactam Infusion Strategies in Paediatric Sepsis: Pharmacokinetics, Resistance and Clinical Outcomes

Project Summary

PhD project evaluating pharmacokinetics, antimicrobial resistance and microbiome effects of intermittent, extended and continuous beta‑lactam infusion in children with sepsis to improve dosing and outcomes.

Project Description

This PhD candidate will work within the wider trial team in the multicentre, multinational RCT. The project will examine pharmacokinetic, microbiological and clinical effects of intermittent, extended and continuous beta‑lactam infusion in critically ill children with sepsis. Pharmacokinetic outcomes will be assessed using blood samples collected on day 3 post‑randomisation, alongside measurement of drug concentrations in sterile compartments including cerebrospinal fluid and pleural fluid. Microbiological sampling from skin swabs, stool and endotracheal aspirates or sputum will assess microbial colonisation, antimicrobial resistance development and changes in antibiograms compared with the wider PICU cohort. For patients with positive cultures, minimum inhibitory concentrations will be measured and compared between infusion modes. Associations between administration strategy, site of infection, causative microorganisms and outcomes will be explored. A metagenomics sub‑study will evaluate microbiome changes linked to infusion strategy.

Project type: PhD

Contact: Dr Sainath Raman 

Project Title

Machine Learning in the Paediatric Intensive Care Unit: Development of Risk Prediction Models

Project Summary

This project will develop interpretable machine learning models to support clinical decision‑making in paediatric intensive care units. Using high‑resolution data from over 32,500 admissions across three international PICUs, it will focus on early detection of patient deterioration. A strong emphasis on explainability will ensure predictions are transparent and clinically meaningful. By aligning decision support tools with clinicians’ reasoning and workflows, the project aims to enable timely interventions and improve outcomes for critically ill children.

Project Description

This PhD project will develop interpretable machine learning models to support clinical decision-making for critically ill children in paediatric intensive care units (PICUs). Using high-resolution data from over 32,500 patient admissions across three international PICUs, the research will focus on early detection of patient deterioration. Emphasis will be placed on explainability and reasoning, ensuring that predictions and recommendations are transparent and clinically meaningful. The project will explore how to design and implement decision support tools that align with clinicians’ workflow and reasoning processes in high-pressure environments. The project aims to improve outcomes of Australia’s most critically unwell children by supported enhanced clinical decision making and, in turn, timely interventions, ultimately supporting precision medicine in critical care.

Project type: PhD

Contact: Professor Kristen Gibbons

Project Title

Methodological Advances in Paediatric Critical Care Trials: Novel Approaches for Cluster Crossover and Registry‑Enabled Designs

Project Summary

This PhD project focuses on developing and applying innovative statistical approaches for cluster crossover and registry-enabled randomised trials, using the landmark RESONANCE-PICU (Buffered Solutions versus 0.9% Sodium Chloride for Intravenous Fluid Therapy to Improve Outcomes for Critically ill Children: The RESONANCE-PICU Multicentre, Binational Randomised Controlled Trial) study as a real-world application setting.

Project Description

This PhD candidate will work within the wider trial team in the multicentre, bi national RCT; RESONANCE PICU.  RESONANCE PICU compares the safety and efficacy of balanced/buffered solutions against standard 0.9% sodium chloride for intravenous fluid therapy. The successful candidate will explore cutting‑edge methods for trial design and data integration within large clinical registries, working closely with biostatisticians, clinicians, and national research networks. This is an exceptional opportunity for a statistics/mathematics graduate to help shape how future large‑scale trials are conducted, while gaining hands‑on experience with rich, high‑quality pediatric ICU data.

Project type: PhD

Contact: Dr Sainath Raman 

Precision medicine tailors therapies for an individual by taking into account the underlying heterogeneity in a population. Many models have been developed, but translation has only been successful in few disciplines. To date, the scarce precision medicine models in paediatric critical care have been limited by small sample sizes, lack of external validation and have not been trialled in clinical practice. However, PICU is optimised for precision medicine due to the large quantity of biomarkers measured, widespread implementation of electronic health records (EHRs), and extensive digitised physiological monitoring. Machine learning models derived from these granular, high-quality PICU datasets, coupled with omics data, will explain the biological heterogeneity. The models will identify children who are more likely to benefit from certain interventions and establish personalised care pathways for consumer-prioritised outcomes. Using the extensive, highly curated databank from the NITRIC clinical trial (largest international trial of infants undergoing heart surgery) and the follow-up study, and the RAPIDS observational cohort investigating better identification of children with sepsis, as sandpits to develop models to establish personalised trajectories, this project will:

1.              Derive and validate phenotypes to quantify the heterogeneity of the patient cohort (including infants undergoing heart surgery, and septix patients).
2.              Develop prediction models for prioritised short- and long-term outcomes incorporating clinical and omics data.

Project type: Multiple Masters / PhD

Contact: Professor Kristen Gibbons

There is a paucity of high-quality trials to inform the life-saving care of children admitted to PICU. >75% of all PICU trials have ≤100 participants, >80% only recruit from a single centre and only 3% have used an innovative trial design, severely hampering the delivery of robust and generalisable results. The COVID-19 pandemic highlighted the need for agile trial infrastructure and international networks to rapidly generate gold-standard, personalised evidence to inform practice. Adult intensive care researchers demonstrated the feasibility of such an approach in a critical care setting, enrolling thousands of patients into global interventional trials using innovative designs. These learnings need to be translated to the paediatric setting to trial precision medicine approaches in a robust, efficient manner, delivering best care for children. This project will lay the foundations for translating a machine-learning model into clinical practice. Components of the project will include undertaking a systematic review, developing and deploying silent models into clinical care, and collecting and evaluating stakeholder feedback in relation to implementation of such models, and, using implementation science, designing best practice tools to promote the highest level of successful model translation.

Project type: Masters / PhD

Contact: Professor Kristen Gibbons

Sepsis, defined as life-threatening organ dysfunction caused by dysregulated host response to infection, affects an estimated 48.9 million cases every year. Sepsis is not only a leading cause of global mortality, but can result in long-term morbidity related to physical, cognitive, and psychosocial sequelae. This translates into a major health and economic burden to societies. In 2017, the World Health Organisation (WHO) declared sepsis a global health priority. The WHO resolution stated an urgent need to implement measures for the prevention, diagnosis, management, and follow-up support of sepsis. This project will use existing MRFF and NHMRC funded cohorts to identify and explore patterns of responses to long-term follow-up assessments and questionnaires from children and families after an intensive care admission, co-design with stakeholders enhanced methods for conducting long-term follow-up, and implement such strategies through a prospective project.

Project type: Masters / PhD

Contact: Professor Kristen Gibbons