The person is Grace Wade.

Close Up Photo of Doctor Hands in Orange Gloves Taking Throat Swab Test from Mature Man in the Hospital; Shutterstock ID 2109822863; purchase_order: -; job: -; client: -; other: -

Current methods can be hard to diagnose oral and throat cancer because of the large size of the tumors.

Tijana Simic has a stock photo.

A new diagnostic tool in the US uses artificial intelligence to detect oral and throat cancer using saliva samples.

There will be tens of thousands of new cases of oral cancer and oesophageal cancer in the US this year, according to estimates. When detected early, the five-year survival rates for these cancers jump to more than 86 and 47 percent.

Most oral and throat cancer aren't detected early. Screening methods rely on visual exams. Until they can be seen by the eye, they must grow large.

People who are cancer-free have different levels of oralbacteria than people with cancer. In order to see if he and his colleagues could create a more accurate diagnostic tool by looking at changes to the microbiome, Guruduth Banavar and his colleagues went to New York City to do some research.

They collected saliva samples from 1175 people who were 50 years old or older and had a history of tobacco use. They took genetic material from the samples.

In order to train a machine-learning model, the team used genetic data from people with oral and throat cancer. There are 88 distinct changes to gene expression in people with oral and throat cancer, as well as 182 genetic features unique to thebacteria found in samples from those people.

Some of the samples came from people with cancer. There were very low rates of false negatives and false positives because it accurately identified 90 percent of samples from people with cancer and 95 percent of samples from people without.

The saliva test they developed with this work was given breakthrough device designation by the FDA in April 2021. Products that can improve treatment or diagnosis of life threatening illnesses are given expedited review.

Under rules from the US Centers for Medicare & Medicaid Services, the test is now available for purchase in the US: people at high risk for oral or throat cancer can fill out a questionnaire and get results in around two weeks. The test would be covered by health insurance providers if the FDA approves the test.

It's not clear how useful this test will be in the future. The Northwell Health Cancer Institute in New York has an oral and throat cancer specialist. When we have all these technologies, it's archaic to visually inspect someone and wait until a large lump of tissue is visible.

He points out that the diagnostic tool was tested in a small number of people and that it didn't find cancer. There is a question of cause and effect. Is it possible that the cancer is causing thesebacteria to change? Is it possible that some people will develop cancer later in life?

There is a practical issue as well. What are doctors supposed to do with a test that's 100 percent accurate? If a cancer is too small to be seen by the eye, it cannot be biopsied, and if it is cancer, insurance companies won't approve treatments.

The technology will get better over time according to Banavar. The more data we collect, the better machine learning will be.

Cancer’s evolutionary journey Kat Arney at New Scientist Live this October

There are more on this topic.