Validation of an AI-powered online search strategy for finding optimal biomarker combinations
Approved Research ID: 87991
Approval date: May 23rd 2022
Early detection of diseases plays an important role in improving treatment outcomes and reducing burden on the healthcare system. Unfortunately, many diseases are diagnosed at a late stage due to the lack of convenient and reliable diagnostic tests which can be used for regular monitoring in an at-home setting. As there are often multiple biomarkers associated with a disease, finding the optimal combination of biomarkers (which are affordable, accurate, and easy for people to measure) that can predict a disease can be cumbersome with traditional approaches.
Sanome, a UK- and Netherlands-based healthtech start-up, has developed a search strategy powered by artificial intelligence for finding the optimal combinations of biomarkers that predicts a disease (or a health condition), which can be packaged into a diagnostic test. This strategy efficiently investigates a huge number of possible combinations of biomarkers and requires fewer samples to be collected compared to traditional diagnostic biomarker studies. In this project, we will be validating this strategy on UK BioBank data for predicting colorectal cancer using the biomarkers specified in a previous paper "Circulating liver function markers and colorectal cancer risk: A prospective cohort study in the UK Biobank", which successfully found an association with colorectal cancer, and some additional biomarkers to make the search task more challenging.
The proposed project is expected to last 12 months. We will publish these findings as a short article and keep the original authors of the paper informed, while also using the findings to inform the search strategy work within Sanome. Ultimately, Sanome's search strategy will make the development of new diagnostic tests cheaper and faster, enabling earlier detection of diseases and easing burdens on healthcare systems.