AI-enabled electronic nose detects ovarian cancer

Electronic nose technology has been around for about six decades. The prototype the researchers used is equipped with 32 sensors that respond to different volatile compounds released by the sample. Because different types of cancer produce different sets of volatile substances, each has its own unique “chemical profile.” The sensors themselves are of simple design and are available on the market. However, thanks to the rapid development of machine learning and artificial intelligence in recent years, this technology has been applied in new ways and its capabilities have been significantly increased.

Currently, cancer screening using blood tests is usually based on the search for specific biomarkers characteristic of a particular tumor type. Such analysis takes time and is not always highly accurate. As the researchers note, unlike breast cancer, a reliable method of screening for ovarian cancer does not yet exist. Current tests often rely on a single biomarker and do not provide sufficient accuracy for early detection of the disease.

The technology created by scientists is not focused on finding one specific biomarker. Instead, the “electronic nose” detects a wide range of volatile compounds that are released from blood plasma samples. The resulting data is then processed using advanced machine learning algorithms that can identify patterns specific to ovarian cancer. Models are trained on pre-studied samples stored in a biobank. According to test results, the accuracy of the method reaches 97%.