This project aims to assess the feasibility and safety of using large language models (LLMs) to assist clinicians in the care of paediatric patients in the perioperative setting. We will investigate the ability of LLMs to classify and predict the incidence of perioperative complications from pre-operative textual data. This includes perioperative respiratory adverse events (PRAE defined as major adverse events such as laryngospasm or bronchospasm and minor adverse events such as severe persistent coughing, desaturation, airway obstruction, stridor), pain, nausea and vomiting and other complications from multi-modal input. We hypothesise that LLMs will be able to
perform automatic classification and prediction of perioperative paediatric adverse events using preoperative information from the digital medical record.
Dr Liam O’Doherty, Dr Harry Smallbone, Perth Children’s Hospital, A/Professor Wei Liu, The University of Western Australia.
The project was awarded A$70,000 funding through the ANZCA research grants program for 2026.