Researchers have used new artificial intelligence (AI) techniques to identify which patients with heart failure do, or do not, benefit from beta blockers. Their approach interrogates data more deeply than was previously possible, identifying clusters of patients “invisible” to standard research methods. Even more significantly, the researchers believe their approach is widely applicable across many diseases.
Beta blockers are widely used to treat patients with heart failure and reduced left ventricular ejection fraction (LVEF). Despite their value for many patients, death rates remain unacceptably high. Anecdotal evidence suggested that they may not be effective in preventing death among patients with co-morbidities such as atrial fibrillation (AF). However, conventional research methods failed to isolate which cohorts of patients did or didn’t benefit.
The multidisciplinary cardAIc group (supported by HDR UK) applied a novel combination of machine learning methods better able to take account of patients’ other conditions.
This approach (which uses neural network-based autoencoders to isolate key features in the data) assesses multiple and higher-dimension interactions of comorbidities and can cluster patients in new ways. Specifically they could identify whether beta blockers affected mortality rates differently in patients with normal heart (or so-called sinus) rhythms compared to those with AF – also accounting for factors such as age and sex.
A resulting paper (published in The Lancet) describes a deep interrogation of a meta-study of 15,669 heart failure patients with reduced LVEF. Some 12,823 had normal heart rhythm and 2,837 had AF.
Impact and outcomes
First author Dr Andreas Karwath, a research fellow in the University of Birmingham Institute of Cancer and Genomic Sciences and HDR UK Rutherford Fellow, says the team found that beta blockers reduced deaths among most patients with normal heart rhythms – except older patients with less severe symptoms. Conversely mortality was not reduced among patients with AF, except in the cluster of lower risk younger patients.
These results had not been found previously because novel approaches were needed to identify patterns hidden beyond the linear dependencies.
The research (which was presented at the 2021 European Society of Cardiology congress) could result in tailored treatments for individual patients and reduced NHS drugs spending. Looking ahead, it’s hoped that it could spur the development of new drugs for those who do not benefit from beta blockers.
Corresponding author Georgios Gkoutos, Professor of Clinical Bioinformatics at the University of Birmingham and co-lead for the CardAIc group, said:
“I would anticipate that this approach would help shape healthcare policy and improve treatment and outcomes for patients with heart failure aiding the development of personalised treatments for each individual patient, taking account of their particular health circumstances to improve their well-being”
He added that the new approach is:
“hugely exciting due to its potential to be applied across other cardiovascular and non-cardiovascular conditions.”
The Impact Committee
The HDR UK Impact Committee described the work as of international excellence in originality, significance and rigour. It was commended for its multidisciplinary approach and potential to be applied across conditions. The study also contributes to HDR UK’s National Priorities of Applied Analytics and Better Care.
Dr Karwath firstname.lastname@example.org
Can we accurately forecast non-elective bed occupancy and admissions in the NHS? A time-series MSARIMA analysis of longitudinal data from an NHS Trust
1 July 2022
Hospitals need to be able to predict their capacity for admitting patients when planning elective surgeries. Researchers funded by HDR UK developed a new model for making forecasts that were more...
A population-based cohort study of obesity, ethnicity and COVID-19 mortality in 12.6 million adults in England
21 June 2022
Obesity dramatically increases the risk of death from COVID-19 but, the extent of this risk across different body weights and ethnic groups was not clear. Researchers using health and Census...
Patient-centric characterization of multimorbidity trajectories in patients with severe mental illnesses: A temporal bipartite network modeling approach
21 June 2022
People with severe mental illness have a lower life expectancy and a higher risk of physical conditions. To improve how these comorbidities can be detected and predicted, researchers have used...