The Machine Healthcare Intelligence Unit is a multi-disciplinary initiative comprising UCLH Analytics, the UCL Queen Square Institute of Neurology and UCL’s AI Centre. The Unit is embedded within CRIU and underpinned by the core values of UCLH as a research hospital. One of the primary aims of the research hospital is to transform the hospital by applying artificial intelligence (AI) across its entire fabric, bringing the benefits of machine learning to the NHS.
The objective is to develop a pioneering hospital-embedded framework for complex, real-time operational modelling, and to deliver - fully implemented within UCLH operations - a suite of projects with transformational impact on hospital operations. Central to the approach is the idea that the operation of a hospital is the result of a complex field of interacting factors best captured by high-dimensional, highly-expressive mathematical models.
Introducing such models potentially enables prediction of individual care outcomes, more sensitive inferences about causal factors, and prescription of pathway alterations that maximise quality of care. This kind of models are data-driven, capable of distilling more intelligence from information than conventional statistics, and less reliant on potentially biased expert assumptions about underlying causes. They have the potential to enhance mechanisms for guiding activity currently based on simpler models (e.g. for scheduling), but also to introduce wholly new mechanisms of operation (e.g. flexible resourcing guided by forecast service volumes and characteristics).
Current areas of work are:
- high-dimensional modelling of patient flow
- high-dimensional modelling of scheduled hospital activity
- a computational framework for predicting rare events
- motion categorisation for monitoring and optimising care
- spatial modelling of urban health with novel geostatistics.
