AI found five heart failure categories 2023
A recent study conducted by UCL researchers discovered five forms of heart failure that may be used to forecast patient risk using AI techniques.
“We sought to improve how we classify heart failure, with the aim of better understanding the likely course of disease and communicating this to patients,” stated lead author Professor Amitava Banerjee (UCL Institute of Health Informatics). Patients can’t forecast their illness progression. Some people are steady for years, while others decline quickly.”
Heart failure occurs when the heart cannot pump blood properly. Heart failure categorization schemes cannot accurately predict progression.
Better differences between forms of heart failure may lead to more tailored treatments and help us rethink prospective remedies.
Banerjee stated, “In this new study, we identified five robust subtypes using multiple machine learning methods and multiple datasets.
In the year after diagnosis, subgroups had different mortality rates.
The Lancet Digital Health research evaluated anonymized patient data from over 300,000 UK people aged 30 or older who were diagnosed with heart failure over 20 years.
The next stage is to see if this method of categorizing heart failure improves risk estimates, physicians’ knowledge, and patients’ care. Is it cost-effective? Banerjee continued, “Our app needs a clinical trial or further research, but it could help in routine care.”
Early onset, late onset, atrial fibrillation-related (an irregular heart rhythm caused by atrial fibrillation), metabolic (linked to obesity but with a low rate of cardiovascular disease), and cardiometabolic (linked to obesity and cardiovascular disease) were identified using machine learning methods.
Early (20%), late (46%), atrial fibrillation (61%), metabolic (11%), and cardiometabolic (37%) were the one-year all-cause death risks.
The study team also built an app to help physicians identify a heart failure patient’s subtype, which might enhance risk forecasts and patient talks.
Researchers grouped heart failure patients using four machine learning algorithms to reduce bias. They used these strategies to two large UK primary care datasets that were typical of the UK population and connected to hospital admissions and mortality records.
CPRD and THIN covered 1998–2018.
The study team trained machine learning methods on parts of the data and confirmed the strongest subtypes using a second dataset.
Age, symptoms, other conditions, medicines, and test and evaluation findings (e.g., blood pressure, renal function) were used to determine the subtypes.
The scientists examined genetic data from 9,573 UK Biobank heart failure patients. Certain heart failure subtypes were associated with increased polygenic risk scores for hypertension and atrial fibrillation.