## Understanding Long-term Conditions and Cognition (UnLoCC) ## Given the close relationship between cognitive impairment and multiple long-term medical conditions (Lipnicki et al., 2019), new approaches to identify, predict and manage cognitive decline in this population are urgently required. Prediction of cognitive change in people with multiple long-term conditions would facilitate future care planning, selection for specific interventions, and stratification for clinical trials, but is hampered by heterogeneity between individuals and between different long-term conditions. Artificial intelligence (AI) methods offer an approach to tackle the complex data required to identify clusters of long-term conditions associated with specific cognitive profiles, and to make predictions about future cognitive change. Linking AI to defined clinical pathways will provide a clear path to translation. The work will be underpinned by established longitudinal datasets with clinical assessments, neuroimaging and blood-based biomarkers on a range of different patient groups with multiple long-term conditions, with and without neurodegenerative disease. This includes two NHS memory clinic-based datasets (South London and Maudsley Memory Clinic, and Quantitative MRI in the NHS - Memory Clinics dataset). To facilitate use of these datasets, we will establish methodologies to harmonise cognitive tests, neuroimaging pipelines. The DEMON network ([http://demondementia.com/][1]) was established in 2019 to bring together experts in AI with clinicians and scientists to tackle challenges in dementia using deep phenotyping and big data. There are currently members from the UK and other countries around the world. A key aspect of this project is the translational element. We will outline a path to specific interventions within NHS clinics to predict decline in people with multiple long-term conditions using machine learning tools, which are linked to clearly defined clinical pathways of support for cognitive impairment, and intervention for specific long-term conditions to prevent cognitive decline. These clinical pathways will not depend on a diagnosis of a specific neurodegenerative disease, but rather on predicted cognitive decline. The work packages for the main project will be: * Cognition and Neuroimaging: to develop predictive models that are informative about the link between multiple long-term conditions and cognition * Cardiovascular disease: to measuring cerebrovascular disease, and to assess the influence of cardiovascular risk factors on neuroimaging and cognition * Polygenic risk scores: to understand the influence of polygenic risk scores for cognition and dementia on the rate of cognitive change in multiple long-term conditions * Translation: developing a pathway to translate machine learning methods into clinical practice through regulatory approved tools and defined clinical pathways. We expect the outcome of the project to be a better understanding of the impact of long-term conditions on cognitive impairment, and artificial intelligence tools for prognosis and intervention that are trial-ready with a view to registration as medical devices for use in clinical practice. [1]: http://demondementia.com/