NHLBI IN THE PRESS

Machine-learning model identifies gut bacteria associated with cardiovascular health

Beneficial healthy intestinal bacterium in the micro flora.

A study supported by the National Heart, Lung, and Blood Institute finds machine learning, a type of artificial intelligence, can scan fecal samples for gut bacteria associated with cardiovascular health. Researchers from the University of Toledo School of Medicine presented the abstract at the American Heart Association’s virtual Hypertension 2020 Scientific Sessions on Sept. 10-13, 2020. The study published in Hypertension on Sept. 10, 2020.

Out of 951 fecal samples analyzed from the American Gut Project, 478 were from people with cardiovascular disease and 473 were from people without cardiovascular disease. There wasn’t a distinction among cardiovascular disease types, such as elevated blood pressure, previous heart failure, or coronary artery disease. The researchers viewed a general cardiovascular disease classification as beneficial for broader risk calculations for this study, as well as a variable to manipulate in future research to further personalize cardiovascular disease-risk assessments.

Eight bacterium were abundant in the cardiovascular disease samples: Bacteroides, Subdoligranulum, Clostridium, Megasphaera, Eubacterium, Veillonella, Acidaminococcus, and Listeria. Seven bacterium were common in non-cardiovascular disease samples: Faecalibacterium, Ruminococcus, Proteus, Lachnospira, Brevundimonas, Alistipes, and Neisseria.

The researchers confirmed their hypothesis, which was the ability to train machines to detect dozens of bacterium associated with heart and vascular disease. Machine learning has been used to identify cancer, diabetes, and inflammatory bowel disease, but this is the first model to analyze a machine’s ability to sort gut bacteria associated with cardiovascular health.

This model isn’t ready for clinical use, but the researchers envision microbiota sampling could complement future health records and cardiovascular health screenings – such as analyzing age, family history of heart disease, blood pressure, blood sugar, and cholesterol. They note fecal sample analyses also have the potential to preempt or provide a preliminary step to advanced heart screenings, including chest x-rays, echocardiograms, and electrocardiograms.