Machine learning flags patients at risk of TB comeback

New diagnostic tools could better identify which patients with tuberculosis are at the highest risk of reactivation of the disease.
lung model on white (tuberculosis concept)

New diagnostic tools such as machine learning and precision medicine could help identify tuberculosis patients with the highest risk of reactivation of the disease, report researchers.

The researchers are showing that identifying multiple biomarkers can provide a more accurate diagnosis for patients.

“A multi-array test can provide a more detailed, disease specific glimpse into patient’s infection and likely outcome,” says Ryan Bailey, professor of chemistry at the University of Michigan and coauthor of the study, which appears in Integrative Biology.

“Using a precision medicine approach reveals previously obscured diagnostic signatures and reactivation risk potential.”

Latent tuberculosis infection (LTBI) affects nearly 2 billion people around the world and about 10 percent of those cases result in active tuberculosis. The reactivation from latency can happen anytime and the mechanism for it is not well-understood.

Currently, doctors test for LTBI using a skin scratch test or a blood test that identifies one biomarker but cannot distinguish between memory immune response, vaccine-initiated response, and non-tuberculous mycobacteria exposure. The possibility of correctly identifying the disease through these tests is less than 5 percent.

Doctors use an antibiotic regimen to treat tuberculosis, but that can increase the risk of antibiotic resistance.

The new diagnostic tools will help identify patients with the highest risk of reactivation and who will benefit from therapy, and reduce some of the side effects of overtreatment. By introducing multiple biomarker assays in blood tests with powerful analysis tools, the chances for correctly diagnosing TB increases dramatically.

The researchers used a precision normalization approach to correct for differences in individual basal immune function that revealed a high- and low-reactivation risk.

“This high-level multiplexing, high-assay performance can be cost-effective and scalable,” Bailey says, adding that it can also be used to detect other diseases like autoimmune diseases and cancer.

Additional researchers from the University of Michigan, the University of Illinois, and the Mayo Clinic in Rochester, Minnesota contributed to the study.

Source: University of Michigan

The post Machine learning flags patients at risk of TB comeback appeared first on Futurity.

More from Futurity

Futurity3 min read
Facebook-style Algorithms Hunt For Dark Matter In Space Maps
Machine learning algorithms can help computers teach themselves to extract information about dark matter and dark energy from maps of the universe, researchers report. Understanding how our universe came to be what it is today and what its final dest
Futurity3 min readSociety
Anti-psoriasis Compound Could Lead To New Drug For Malaria
Redesigning molecules originally developed to treat the skin disease psoriasis could lead to an effective new drug against malaria, according to new research. Researchers modified a class of molecules called pantothenamides to increase their stabilit
Futurity2 min readPolitics
RBG Explains How The Supreme Court Has (and Hasn’t) Changed
Supreme Court Justice Ruth Bader Ginsburg has seen major changes in the law and the highest court in the land, but there are some things that she says haven’t changed all that much about serving on the Supreme Court. “One way it hasn’t changed—the co