Introduction
Multimorbidity is a widely recognised global health priority, yet there is limited understanding of its different forms and how different conditions develop and change over time within an individual's life course. This project will combine very large electronic healthcare data with specialist statistical, data science, machine learning and clinical epidemiological skills in cardiovascular disease to investigate the underpinning methods required to investigate longitudinal multimorbidity phenotypes.
Explaining the science
One of the barriers to our understanding of multimorbidity is the lack of advanced analytical techniques which can account for complex multimorbidity disease interactions whilst providing output which is reproducible, interpretable and clinically relevant. Currently available and longstanding tools to measure multimorbidity include aggregated scores such as the Charlson and Elixhauser comorbidity index which fail to capture different forms of multimorbidity and do not consider a standardised timeframe for the accrual of individual conditions. Moreover, current epidemiological techniques for temporal analyses, such as multistate modelling, rely on prior knowledge of the order in which conditions occur and their complexity increases exponentially with the number of conditions included.
This project therefore aims to investigate statistical machine learning techniques to develop longitudinal multimorbidity phenotypes which strike a balance between handling the complexity in the data structure as well as the complexity of many disease interactions – yet provide outputs which are reproducible, interpretable and clinically useful.
Project aims
The proposed project provides the necessary step towards deepening our understanding of the complex causal pathways of multimorbidity following myocardial infarction (MI) alongside open access methodologies with potential for transfer to other clinical areas. The anticipated outputs will inform researchers of the most appropriate methods which can begin to tackle the complexities in multimorbidity research as well as inform patients, clinicians and commissioners of the long term prognosis and risk of post-MI multimorbidity.
Applications
The project focuses on multimorbidity development following myocardial infarction (MI). Major treatment advances over recent decades have led to substantial improvements in survival following MI. However, this increased survivorship is associated with an increased risk of recurrent events and development of other diseases, and an estimated 60% of patients with MI are multimorbid. Yet to date there is limited understanding of the nature, sequence and timing of the development of a range of conditions following MI - which has limited the development of appropriate management strategies to minimise adverse outcomes.
Organisers
Professor Niels Peek
Turing Fellow
Professor Ronan Lyons
Clinical Professor of Public Health, Medicine, Swansea University
Professor Chris Holmes
Programme Director for Health and Medical Sciences, and Turing Fellow
