Astrophysicists will swap galaxies for general practice in an innovative new study that aims to improve the early diagnosis of dementia.
Researchers at BSMS and the University of Sussex will analyse data from 96,000 GP records to identify common, early indicators of dementia, using statistical techniques developed to catalogue galaxies. The ASTRODEM study has been awarded £94,000 by the Wellcome Trust.
Researchers will observe a broad range of data available in the GP records, such as the number of appointments a patient has had, or whether they attended with a family member. These details will be combined with other clinical information known to be predictive of dementia, and could provide a wide range of indicators that may help GPs identify patients at high risk of developing the condition.
Project lead Dr Elizabeth Ford said: “Dementia is one of the greatest public health challenges of our era and more timely diagnosis may help many patients. Therefore, GPs could benefit from a ‘real-time’ risk stratification tool to use in their workflow, which could then be used sensitively to bring up the topic of memory problems and perhaps start a conversation about a referral for a memory assessment, if they feel it is appropriate.”
Dr Philip Rooney, a Research Fellow in Astrophysics at the University of Sussex, will conduct much of the data analysis.
Dr Rooney said: “Astrophysicists at the University of Sussex regularly apply statistical and machine-learning techniques to catalogues of galaxies and galaxy spectra. Researchers can apply these techniques to GP patient records, where the clinical condition is analogous to a galaxy, the GP database is similar to a galaxy catalogue and diagnosis of a condition is similar to the modelling of galaxy properties.
“Methods have been developed for large physics experiments to best account for uncertainties in a dataset and to efficiently extract information. These methods should translate well in analysing GP records where, similarly, we cannot repeat measurements and the data may be noisy, with recording errors and misdiagnosis.”
Timely diagnosis of dementia allows patients to maximise their quality of life, benefit from treatments and plan for the future. However, currently, only 50-60% of patients with dementia receive a diagnosis. Increasing diagnosis rates in general practice and diagnosing earlier in the course of the disease are strategic aims for the UK government and NHS.
ASTRODEM seeks to produce a statistical model that could be embedded into GPs’ standard computer software, to help identify patients with initial indications of dementia.
The GP records have been provided by the Clinical Practice Research Datalink, which curates GP patient records from around 5 million current patients. Following an application process, sets of anonymised records are released to researchers under licence.
Dr Ford, Research Fellow in Primary Care Epidemiology at BSMS, was awarded a Wellcome Trust Seed Award to fund the study. Co-applicants were Prof Jackie Cassell, Prof Helen Smith and Prof Sube Banerjee from BSMS, and Prof Seb Oliver and Dr Peter Hurley from the University of Sussex.
The application for the grant followed nine months of preliminary data analysis by Dr Rooney, under the supervision of Dr Ford and ASTRODEM’s physics lead Prof Seb Oliver, funded by the University of Sussex’s STFC Impact Acceleration Account. The project has also been supported by the Sussex Research Development Fund, for a parallel workstream testing out new analysis methods in synthetic data.