In the intensive care unit (ICU), clinical decision making is fundamentally driven by predicting a patient outcome of sustained benefit in terms of quality and length of life. This project will demonstrate a suite of AI-driven methods that have been created in UQ’s Data Science discipline across eight years of research, spanning five PhD theses and more than a dozen top journal and international conference articles. 

The dataset such as ANZICS (Australian & New Zealand Intensive Care Society) APD (Adult Patient Database) and MIMIC III will be considered. The methods we will demonstrate are aimed at assisting clinicians’ decision making in ICUs across several clinically relevant tasks, including treatment recommendation, SOFA Indexing alarming, EEG prediction, patient similarity identification and automatic summary of temporal multimodality data for life-critical patient situations. In ICU, the most frequently asked question is about whether the patient is improving or deteriorating. The proposed AI-powered software will provide clinicians with an additional, data-driven perspective on the patient situation and their likely prognosis in a bid to assist clinicians in making better evidence-based decisions to reduce false-positive and false-negative errors. 

One of the critical components developed in this project is to make AI-as-a-Service (AaaS) for the stakeholders in an AI-Enabled ICU (Intensive Care Unit) systems. This project is to design and implement the AaaS component in the AI-ICU system. The four major AI services such as scenarios of “What-If”, “Why-Not”, “So-What”, and “How-about” will be implemented based on ICU data and machine learning algorithms.  

Project members

Professor Xue Li

Affiliate of ARC COE for Children and Families Over the Lifecourse
ARC COE for Children and Families Over the Lifecourse

Associate Professor Sen Wang

Associate Professor
School of Electrical Engineering and Computer Science

Associate Professor Robert Boots

Associate Professor, UQ Faculty of Medicine