Can Machine Learning Make Fecal Testing Part Of CVD Screening?

Researchers claim that machine-learning analysis of fecal samples will be the first path to help mass screening for all types of cardiovascular disease (CVD) in patients.

Various machine learning algorithms provide data on the gut flora, and training can distinguish people with a high ROC curve of 0.70 with or without CVD, says Sachin Aryal, a university bioinformatics MS student. In Toledo, Ohio, reported by a group led.

“This shows the promising potential for applying microbiome-based ML. [machine learning] In the future, in addition to the intestinal microbiome, samples from various sources will be added to predict CVD, and certain types of CVD that incorporate combined features such as health records will be stratified. It will be important to further calibrate and improve the predictive capabilities of ML modeling, “the authors said.

Their research was presented as a virtual poster Hypertension meetingSponsored by the American Heart Association and published online at the same time November 2020 issue High blood pressure..

Researchers argued that they were the first to apply existing knowledge of gut flora symbiosis in CVD patients to a machine learning approach to CVD screening.

“These data point to an altered gut microbiota core set as a common feature of various clinical manifestations of CVD,” they said, citing hypertension, atherosclerosis, and heart failure as examples. I have.

“Multiple laboratory tests, such as electrocardiograms, chest radiographs, and blood tests, are usually needed to comprehensively assess cardiovascular health, so there are a variety of supervised ML approaches based on the intestinal microbiome. It is promising for early, regular cardiovascular health monitoring prior to advancing to clinical practice. Tests for the proper diagnosis of certain types of CVD, “said Aryal’s team.

Overall, the findings support the relationship between intestinal flora composition and CVD, but more importantly, more aggressive preventive efforts may benefit future cardiac adverse events. Identifying those at risk, Dr. Stanley Hazen commented, PhD at Cleveland Clinic.

According to Hayson, this study did not do that, and these signs are known to cause changes in the presence of CVD or drugs taken to prevent CVD (eg, statins or aspirin, the composition of intestinal microbes). It does not distinguish whether or not it is the cause.

Therefore, machine learning analysis of fecal samples is not currently a useful addition to clinical practice.

Rita Redberg, MD, of the University of California, San Francisco, said:

“Blood-based tests already exist for intestinal microbial metabolites that predict incident risk and provide insights into interventions that may be beneficial,” Hazen said. Today’s MedPage on mail. “There will be a future (already started) that will provide a cure for CVD prevention that targets the gut flora.”

For such interventions Dietary changes..

The study authors used fecal microbiome data from 478 CVD and 473 non-CVD participants in the American Gut project. Different bacterial taxa were identified between the groups.

The American Gut Project did not collect information about the patient’s diet and medication. Also, self-reported dependence on CVD status does not explain false alarms or undiagnosed cases, Arial et al.

It’s also unclear how researchers define CVD and how fecal screening is used clinically, Redberg said.

“Do we come to the office or ER with chest pain and ask to provide a sample of the stool? Wait for more data to show how this test leads to better diagnosis and results. I think we need it. We use sophisticated machine learning techniques, but there are no clinical applications at this time, “she added.

Last updated: September 10, 2020