Artificial Intelligence and Multimorbidity: Clustering in Individuals, Space and Clinical Context

Objective 4

Developing AI tools to reduce adverse events in those with MLTCs.

As people age they become likelier to experience unexpected health problems, known by researchers as ‘serious adverse events’. Objective 4 aims to improve how we deal with these problems. The first step is to improve the recording of such events. Electronic health records use a combination of codes and free text – notes written by the medical professionals treating the patient. We will use a computerised analysis technique called Natural Language Processing to develop the information we can get from free text. This will provide data on adverse events e.g. falls, delirium, and the resulting outcomes, such as loss of independence and care home admission. It will also give information on other important details such as where people live, and who they rely on for care, which influence adverse events. This enhanced data will allow us to better predict adverse events. We know roughly which groups of people will have unexpected health problems (older people, single people), but if we can predict which individuals will develop them, we can apply more effective care.

The next step is to understand the routes patients take through the healthcare system, referred to as “care pathways”. A patient with a hip fracture might typically start in A&E for rapid surgery, move to an orthopaedic ward, and finally transfer to medicine for the elderly for rehabilitation. But adverse events can change this pathway, so we will use various Artificial Intelligence techniques to develop an in-depth understanding of care pathways, and use this understanding to develop new pathways of care for those vulnerable to adverse events.



Natural Language Programming

Honghan Wu is a lecturer and group lead ( at Institute of Health Informatics, UCL and holds an MRC/HDR-UK Rutherford Fellowship. His research uses text analytics and Knowledge Graph techniques for health data research. His text-analytics tool (CogStack-SemEHR) is routinely run at 5 NHS Trusts on > 3.5m patient records and used by > 60 clinicians. He will lead the Natural Language Processing work in UCL, and lead external validation/transferability work using large-scale data at London NHS Trusts. Find out more about Honghan at the link below:

Artificial Intelligence