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

AIM-CISC is a research programme that aims to produce a detailed picture of multiple long-term health conditions (MLTCs) in the UK, and use the results of this research to inform future strategies in dealing with this issue.

Summary (Research in a nutshell)

Long-term conditions are health issues which persist over years. Many people have more than one long-term condition (e.g. having both diabetes and asthma), which is usually called multimorbidity. Multimorbidity often seriously affects how well people feel and what they are able to do. However, health professionals like doctors or nurses often focus on one condition at a time. This can make it hard to ensure that care is properly joined-up and coordinated for people with multimorbidity. Research also tends to focus on one condition at a time. This is partly because people with multimorbidity can have very complicated mixes of different conditions and treatments. Artificial Intelligence (AI) uses powerful computers to analyse and understand this kind of complex information, which is carefully gathered by the NHS and others under strict rules to keep it safe and confidential. New AI methods can help us better understand what patterns of multimorbidity are common and which most affect people’s lives, and help improve the quality and safety of care. We propose a programme of research which will use AI methods to:

1. Identify the most common combinations of long-term conditions that people have, and examine whether people inherit a tendency to get particular combinations of conditions from their parents. This will help us understand what causes multimorbidity, and identify potential treatments.

2. Examine whether particular combinations of long-term conditions are more common in some areas or communities. This will help us understand how where people live affects their health (for example, through differences in air pollution, or in access to safe open-space to play/exercise, or in having local shops that sell affordable healthy food).

3. Develop new methods for identifying people who are likely to have unexpected health problems like falls or bleeding. This will help us to work out what changes to their treatment or care could prevent these problems. One example of a possible change is medicines review to make sure that people take the right medicines for them as an individual. Another is rapid, comprehensive assessment of people admitted to hospital to ensure that their care suits all of their needs rather than just the problem that brought them into hospital.

Our team combines expertise in AI methods and clinical researchers with experience of delivering and researching healthcare for people with multiple long-term conditions. We will work closely with patients and members of the public to help us focus and interpret our research, and to help us make our findings widely known. We will work closely with other research teams to share learning and methods, and with the NHS and government to ensure that the research turns into practical improvements in care.

Key people

Contact

 aim-cisc@ed.ac.uk

Key publications

Imane Guellil, Jinge Wu, Honghan Wu, Tony Sun, and Beatrice Alex

In 'Proceedings of The Seventh Workshop on Social Media Mining for Health Applications, Workshop & Shared Task', pages 148–152, Gyeongju, Republic of Korea. Association for Computational Linguistics.

DOI: to be confirmed.

https://aclanthology.org/2022.smm4h-1.40.pdf

Key Collaborations

University College London

Partners and Funders

Funder: National Institute for Health and Care Research

Timeline

Start date: 1^st August 2021

End date: 31^st July 2024

How long is the project: 36 months

Scientific themes (keywords)

MLTC, multimorbidity, multiple long term conditions, clusters

Methodology keywords

Multilayer networks, polygenic models, geo-profiling, spatial analysis, natural language processing, knowledge graphs, quantitative modelling.