AI gives targeted intel on heart failure risks

Thanks to artificial intelligence (AI), researchers have a better idea about what types of heart failure could be riskier than others, a study published in Lancet Digital Health reveals. This could pave the way for more accurate treatments and better knowledge to guide patients.

The scientists evaluated anonymous data from more than 300,000 people in the UK over 30 who were diagnosed with heart failure over a 20-year span. They used four machine learning methods, a type of artificial intelligence, or AI. They also used 87 factors to establish the subtypes — things like age, test results and medications patients took.

Then they applied machine learning methods to five subtypes: early onset, late onset, atrial fibrillation related, metabolic and cardiometabolic. They found differences between the subtype and the patient’s risk for dying a year after being diagnosed. Death from all causes risks based on subtype were: atrial fibrillation related (61%), late onset (46%), cardiometabolic (37%), early onset (20%), and metabolic (11%).

The researchers also created an app that healthcare professionals can use to find out which subtype a patient has. This can guide patient discussions and could improve predictions about future risks, they say.

“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,” said Amitava Banerjee of the University College London Institute of Health Informatics, emphasizing how hard it is to predict disease progression.

“Some people will be stable for many years, while others get worse quickly,” Banerjee said.

“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,” Banerjee added.

In the future, the researchers want to see if this method for classifying heart failure will improve insight about risks and patient treatment paths.

“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,” Banerjee said.