Imperial College London partners with the Children's Hospital of Orange County (CHOC) to develop AI support systems for paediatric intensive care.
Together researchers and clinicians from Imperial and CHOC, part of Rady Children's Health and recognised as one of the leading paediatric healthcare systems in the United States, are training the AI Clinician model to help paediatric doctors make complex decisions and provide faster and more accurate treatment recommendations for critically ill children.
The research is led by Professor Aldo Faisal, Director for AI for Healthcare Centres and Imperial Global USA Academic Theme Lead who has brought together researchers and clinicians from the Department of Bioengineering, the Department of Computing and the Department of Surgery and Cancer, with a team of data scientists and clinicians at CHOC.
Prof Faisal is one of the few computer scientists worldwide leading clinical trials from algorithm to bedside and he shared the potential this research has for transforming patient care in his keynote address at CHOC's 2nd Annual International Paediatric and Lifespan Data Science Conference in April 2025.
World leading international paediatric dataset
Unlike earlier models that relied on simple rules, Imperial's AI Clinician uses advanced algorithms to provide personalised and informed recommendations that can consider the full complexity of a patient's health. Training this model on a rich and diverse dataset is fundamental to its accuracy.
The partnership has combined 36,000 anonymised records from CHOC with 20,000 anonymised NHS patient records to create the largest international paediatric dataset of its kind.
"Our aim is that this UK and US partnership will lead to an AI Clinician system which can help children with a much broader diversity of conditions and medical backgrounds." Professor Aldo Faisal Imperial Global USA Academic Theme Lead for Human and Artificial Intelligence
Professor Aldo Faisal said: "The gold-standard for developing and evaluating patient-ready AI technology are multi-centre international trials.
"This UK and US partnership and collaboration allow us to make our AI Clinician system smarter so that it can help children with a much broader diversity of conditions and medical backgrounds."
The need for AI in paediatric intensive care
In a typical Paediatric Intensive Care Unit (PICU), patient data flows continuously, generating up to 100 different data categories every hour for a single patient. These data points offer valuable insights into a patient's health and guide clinical decisions.
However, interpreting such large, complex datasets quickly and effectively poses a significant challenge for medical teams, especially in high pressure situations where lives may be at risk.
"By enabling earlier detection and prediction of critical clinical events, as well as identifying optimal treatment strategies, we aim to significantly improve outcomes and deliver more personalized, data-driven care to every child we serve." Dr Terence Sanger Chief Scientific Officer at CHOC
"The partnership between CHOC and Imperial College London seeks to advance the use of artificial intelligence in paediatrics. This initiative reflects CHOC's broader commitment to harnessing AI to transform children's health.
"By enabling earlier detection and prediction of critical clinical events, as well as identifying optimal treatment strategies, we aim to significantly improve outcomes and deliver more personalized, data-driven care to every child we serve." said Dr. Terence Sanger, Chief Scientific Officer at CHOC.
Combining human insight with AI efficiency
AI models excel at analysing complex, multi-dimensional data, and the aim of AI Clinician is that such analysis can be applied in a real-world scenario where each patient is unique and even slight variations in treatment can impact recovery.
Ensuring the data can be accurately interpreted by the AI Clinician is crucial, which is where the insight and experience of human clinicians is needed. One of the researchers' first tasks involves standardising drug dosages and adjusting data recorded at irregular intervals to fit the hourly format required for AI training. CHOC clinicians work closely with Imperial's data scientists to clarify the significance of certain data points, ensuring the AI models are reliable and robust in their individualised decision-making processes.
"By combining granular data from tens of thousands of children admitted to paediatric critical care units in the UK and USA, this effort has the potential to revolutionise care for sick children in the future." Dr Padmanabhan Ramnarayan Deputy Director of Imperial's Centre for Paediatrics and Child Health
Dr Padmanabhan Ramnarayan, Deputy Director of Imperial's Centre for Paediatrics and Child Health and Professor of Paediatric Critical Care expressed excitement about the collaboration: "We are excited about our transatlantic paediatric critical care data collaboration between Imperial and Children's Hospital of Orange County (CHOC).
"By combining granular data from tens of thousands of children admitted to paediatric critical care units in the UK and USA, this effort has the potential to revolutionise care for sick children in the future."
Legacy of AI Clinician research partnerships
The partnership with CHOC builds on previous Imperial-led successes in using the AI clinician model to support adult sepsis treatment and AI-guided ventilation management which has been trialled in London hospitals. This research involved partnerships with other US-based hospitals and healthcare systems, including Johns Hopkins University, Emory University, and Sentara Healthcare.
