Data fusion in haemodialysis
Research type
Research Study
Full title
Can data fusion techniques predict adverse physiological events during haemodialysis?
IRAS ID
130010
Contact name
Clare MacEwen
Contact email
Sponsor organisation
Clinical Trials and Research Governance
Research summary
Patients with kidney failure being treated by haemodialysis frequently experience symptoms and signs of reduced blood flow to major organs (heart, brain and gut) during their treatment. This not only affects their sense of wellbeing but has been associated with the development of heart failure, cognitive decline and other serious comorbidities. This phenomenon is often associated with falls in blood pressure (intradialytic hypotension or IDH). It usually occurs without warning, and may even go undetected, and so nurses are unable to provide preventative treatment. We intend to prospectively monitor continuous non-invasive blood pressure, standard vital signs (e.g. respiratory rate, heart rate and oxygen saturations) and dialysis machine parameters in a group of haemodialysis patients who are prone to instability or symptoms during treatment. We will then process this data using a technique known as data fusion, initially developed for use in the aerospace industry, and already successfully applied to other areas of medicine. Data fusion combines information from multiple sources to create a probabilistic model of normality in multiple dimensions: deviation from the predicted zone of normality allows for a more sensitive and specific early warning of deterioration. In this example, data fusion will be used to create patient-specific models which give an early warning of adverse events on dialysis, allowing timely and effective intervention. The data gathered will also help us to better understand the physiological stress of haemodialysis.
REC name
South Central - Berkshire B Research Ethics Committee
REC reference
13/SC/0397
Date of REC Opinion
2 Sep 2013
REC opinion
Further Information Favourable Opinion