Medical Xpress reports on a Science Translational Medicine study that pairs a machine-learning algorithm with ctDNA methylation signatures to diagnose colorectal cancer — reaching 87.5% sensitivity and 89.9% specificity.
Featured in Medical Xpress. By Bob Yirka.
Medical Xpress reported on a study, published in Science Translational Medicine, in which a research team in China combined a machine-learning algorithm with cancer methylation signatures to diagnose colorectal cancer — a less invasive alternative to colonoscopy, still the gold standard for detection.
The team identified colorectal-cancer-specific methylation signatures in circulating tumor DNA (ctDNA), then trained a machine-learning algorithm on samples from 801 people with colorectal cancer and 1,021 without. In testing, the system reached 87.5% sensitivity and 89.9% specificity. A companion prognostic model was useful for predicting risk of death for up to 26.6 months, and one methylation marker proved especially valuable during screening.
This ctDNA-methylation approach reflects the science behind Helio Genomics’ liquid-biopsy technology. The underlying study is one of the papers listed among our publications.
Source: Medical Xpress. Read the full article.