Data-Driven Insight and Prediction
The fifth work package of the Advanced Care Research Centre.
This workpackage will develop, validate and disseminate a suite of new risk prediction models for a set of adverse outcomes such as mortality, increased care needs or hospitalisation. In the context of population ageing and resource constrained services, risk prediction tools have great potential to ensure the delivery of the right care to the right person in the most cost-effective way. However, existing tools are suboptimal, lacking either precision in identifying those at risk of adverse outcomes or sufficient validation for confident use in applied settings. Consequently, few risk prediction models have been deployed at scale in real-time as part of larger complex interventions to improve care.
The work package is interdisciplinary and focussed around data and quantitative methods, building on the rapid increase in routine data availability and improvements in available computing power. Analysis of these new routine linked data will allow for new insights into health, vulnerability and care in later life that will be validated and informed by complementary analysis of bespoke research survey data. The insights gained will enable better predictions of future individual trajectories and vulnerability and more effectively targeted interventions.
Dissemination Outputs and Impact
The overall aims of this workpackage are (1) to use bespoke research survey data to explore and understand how later life trajectories of frailty, wellbeing and social participation interrelate and are influenced by factors such as housing, wealth, income, care, neighbourhood context; and (2) to use linked routine data to develop and validate a suite of risk prediction tools, drawing on a range of correlates of adverse outcomes informed by the survey analysis, for use in health and social care delivery. This workpackage will work closely with workpackages 3, 4, and 6 in terms of sharing data, insight, and outcome.
Looking back after someone has died, it is relatively easy to see that there are several reasonably distinct routes or trajectories that individuals take on the road to death. It is much harder to predict in advance what an individual’s trajectory or outcomes will be. Risk stratification is an essential element of any complex intervention to maximise function, quality of life and independence in later life. Existing prediction tools, however, are suboptimal since many have not been adequately validated and few have been deployed at scale in real-time care as part of larger complex interventions to improve care.
We plan an interdisciplinary programme of data-focused work drawing on theoretical perspectives and modelling methods from social science, epidemiology, and machine learning. There will be two strands of related work, drawing on the complementary strengths of survey and routine data as well as multivariate statistics and machine learning. Survey data, such as the English Longitudinal Study for Ageing, provides rich detail in terms of health measures and socio-economic circumstances, while routine data has sample size, timeliness and coverage that should be used in risk prediction models to identify those individuals who require some form of intervention. Similarly, multivariate statistical techniques offer theoretically informed models, while machine learning offers flexible data-driven insight. We will adopt machine learning techniques to uncover new associations and relationships whilst validating substantive findings using theoretically-informed multivariate statistical models. Our expectation is that (1) the complementary strengths and weaknesses of data, method and theory will bring new insights to improve effectiveness of existing risk insight models that have so far been found lacking in performance and validation; and (2) we make a significant methodological contribution by combining different data and different approaches to quantitative modelling to yield stronger risk prediction models.