Consultants within their intensive care units are presented with rich and diverse information that can inform on the immediate health status as well as provide the foundation for longer-term prediction of patient health outcomes. However, the data itself is often siloed within a variety of digital platforms. Furthermore, it takes a broad range of different forms: time-series data from real-time vital sign monitors, readings from blood tests, and patient notes recorded in natural language.

The research project will focus on collecting and combining the data from the diverse internal systems and then applying state-of-the-art machine learning algorithms to provide consultants with valuable insights into the effects of a range of interventions.

Research questions in focus:

• Impact of multi-organ support (MOS) – is it likely to be successful?
• Would 48 hours of MOS make it easier to identify those who will survive intensive care?
• If a patient is given MOS are they likely to survive with a reasonable quality of life?
• What is a patient’s likely length of stay in critical care?