A non-invasive method for concurrent detection of multiple early-stage cancers in women

Abstract

Untargeted serum metabolomics was combined with machine learning-powered data analytics to develop a test for the concurrent detection of multiple cancers in women. A total of fifteen cancers were tested where the resulting metabolome data was sequentially analysed using two separate algorithms. The first algorithm successfully identified all the cancer-positive samples with an overall accuracy of > 99%. This result was particularly significant given that the samples tested were predominantly from early-stage cancers. Samples identified as cancer-positive were next analysed using a multi-class algorithm, which then enabled accurate discernment of the tissue of origin for the individual samples. Integration of serum metabolomics with appropriate data analytical tools, therefore, provides a powerful screening platform for early-stage cancers.