Using AI to predict future stroke using routine investigations
Research type
Research Study
Full title
Using Artificial Intelligence to predict future stroke using routine historical investigations.
IRAS ID
306246
Contact name
Stephen Mullin
Contact email
Sponsor organisation
University of Plymouth
Duration of Study in the UK
3 years, 0 months, 1 days
Research summary
We seek permission to use routine computer tomography (CT), magnetic resonance spectroscopy (MRI), ECG and echocardiogram data to develop an artificial intelligence model to predict the risk of future stroke. Our hope is that, based on this data, we will be able to develop of a tool which will be able to quantify the risk of future stroke and haemorrhage, based on historical routine investigations.
Briefly, stroke cases will be identified from routine the Sentinel stroke National; Audit Programme, hospital electronic records and GP records. Relevant clinical data will also be extracted by the routine clinical care team. CT/MRI/ECG/echocardiogram data will be matched using pseudonymised identifiers (NHS number, year of birth, first two letters of postcode) by a member of the routine clinical care team who is not part of the research team. A number of matched control CT/MRI/ECG/echocardiogram data will also be obtained as a control group. Once matched, the complete dataset will be de-identified and released to the university of researchers for analysis. The anonymised dataset will remain only on the hard drive of the departments dedicated machine learning computer or on a cloud database under the control of the University of Plymouth. The UHPNT research team will not have access to the pseudonymised or identifiable datasets.
REC name
South Central - Oxford B Research Ethics Committee
REC reference
23/SC/0217
Date of REC Opinion
19 Jun 2023
REC opinion
Favourable Opinion