Heart failure is a heterogeneous syndrome reflecting multiple underlying causes. Heart failure subtypes have been analyzed using machine learning, but not across vast, distinct population-based datasets, across the whole range of causes and manifestations, or with clinical and non-clinical validation using several machine learning techniques.
In a new study, UCL researchers used their published framework to identify and validate heart failure subtypes upon population-representative data. Using AI tools, they identified five subtypes of heart failure that could be used to predict future risks for individual patients.
The tools were trained on data segments, and once they had selected the most robust subtypes, they validated these groupings using a separate dataset. The subtypes were determined using 87 (of a total of 635) variables, such as age, symptoms, the presence of additional illnesses, the drugs the patient was taking, and the outcomes of tests (such as blood pressure measurements) and evaluations (such as kidney function measurements).
Researchers examined detailed anonymized patient data from over 300,000 people aged 30 years or older diagnosed with heart failure in the UK over 20 years. They used multiple machine learning methods and came up with five subtypes: early onset, late onset, atrial fibrillation related (atrial fibrillation is a condition causing an irregular heart rhythm), metabolic (linked to obesity but with a low rate of cardiovascular disease), and cardiometabolic (linked to obesity and cardiovascular disease).
Researchers also noted significant differences between these subtypes n patients’ risk of dying the year after diagnosis. Early onset (20%), late-onset (46%), atrial fibrillation-related (61%), metabolic (11%), and cardiometabolic (37%) death rates were the highest at one year.
To determine which subtype a person has, the researchers also developed an app for routine clinical use, which could enable the evaluation of effectiveness and cost-effectiveness. This app is expected to improve future risk predictions and inform patient discussions.
Lead author Professor Amitava Banerjee (UCL Institute of Health Informatics) said: “We sought to improve how we classify heart failure, with the aim of better understanding the likely course of the disease and communicating this to patients. Currently, how the disease progresses is hard to predict for individual patients. Some people will be stable for many years, while others get worse quickly.”
“Better distinctions between types of heart failure may also lead to more targeted treatments and may help us to think in a different way about potential therapies.”
“The next step is to see if this way of classifying heart failure can make a practical difference to patients – whether it improves predictions of risk and the quality of information clinicians provide, and whether it changes patients’ treatment. We also need to know if it would be cost-effective. The app we have designed needs to be evaluated in a clinical trial or further research, but could help in routine care.”