Researchers have developed a new machine learning platform for identifying patients having colorectal cancer.
The platform also helps in the prediction of disease severity and survival chances of the patients.
The study was published in the journal Science Translational Medicine.
The method which is a non-invasive one has added on to the technological advances that are used for the analysis of circulating tumour DNA or the ctDNA and can also be used for spotting colorectal cancer at earlier stages in patients.
Colorectal cancer is one of the very few cancers that can be treated if detected before they have metastasized to other tissues.
Researchers leveraged machine learning techniques for the development of a less invasive diagnostic method that is capable of detecting colorectal cancer.
The scientists carried out the research by creating a diagnostic model based on nine different methylation markers that were associated with colorectal cancer. They used a plasma sample of 801 colorectal cancer patients for identifying them as well as 1,021 controls.
This model was accurate in distinguishing people with colorectal cancer and healthy individuals patients from healthy individuals that have a sensitivity and specificity of 87.5 per cent and 89.9 per cent, respectively.
A modified prognostic model further helped predict the risk of patients' death during a follow-up period of 26.6 months.
( With inputs from ANI )