AI is trained to spot warning signs in blood tests tophyper

This is the third feature in a six-part series examining how artificial intelligence is changing medical research and treatments.

Ovarian cancer is “rare, underfunded, and fatal,” says Audra Moran, president of the Ovarian Cancer Research Alliance (Ocra), a global charity based in New York.

Like all types of cancer, the earlier it is detected, the better.

Most cases of ovarian cancer start in the fallopian tube, so by the time it reaches the ovaries, it may have already spread elsewhere as well.

“Five years before you have any symptoms, you may have to detect ovarian cancer, to impact the mortality rate,” says Ms Moran.

But new blood tests have emerged that use the power of artificial intelligence (AI) to detect signs of cancer in its very early stages.

And it’s not just cancer, AI can also speed up other blood tests for potentially deadly infections such as pneumonia.

Dr. Daniel Heller is a biomedical engineer at Memorial Sloan Kettering Cancer Center in New York.

His team has developed a testing technique that uses nanotubes, which are tiny tubes of carbon about 50,000 times smaller than the diameter of a human hair.

About 20 years ago, scientists began discovering nanotubes that could emit fluorescent light.

In the past decade, researchers have learned how to change the properties of these nanotubes so that they respond to almost anything found in the blood.

It is now possible to place millions of nanotubes in a blood sample and have them emit different wavelengths of light depending on what sticks to them.

But that still leaves the question of interpreting the signal, which Dr. Heller likens to finding a match to a fingerprint.

In this case, the fingerprint is a pattern of molecules attached to sensors, with different sensitivities and binding strengths.

But the patterns are too subtle for humans to pick out.

“We can look at the data and not understand it at all,” he says. “We can only see different patterns with artificial intelligence.”

Decoding the nanotube data meant loading the data into a machine learning algorithm, and telling the algorithm which samples came from patients with ovarian cancer, and which ones came from people without it.

These included blood from people with other forms of cancer, or other gynecological diseases that could be confused with ovarian cancer.

A big challenge in using AI to develop blood tests for ovarian cancer research is that it is relatively scarce, which limits the data needed to train algorithms.

Much of that data is kept at the hospitals that treated them, with minimal data sharing to researchers.

Dr. Heller describes training the algorithm on data available from just a few 100 patients as “passing the Hail Mary.”

But he says the AI ​​was able to get better accuracy than the best cancer biomarkers available today, and this was just the first attempt.

The system is undergoing further studies to see if it can be improved using larger sets of sensors and samples from many more patients. More data can improve algorithms, just as self-driving car algorithms can improve with more testing on the street.

Dr. Heller has high hopes for the technology.

“What we would like to do is triage all the gynecological diseases – so when someone comes in with a complaint, can we give doctors a tool that quickly tells them whether it is more likely to be cancer or not, or more likely to be cancer than that.”

This could take “three to five years,” says Dr. Heller.

Not only is AI useful for early detection, it is also speeding up other blood tests.

For a cancer patient, pneumonia can be fatal, and since there are about 600 different organisms that can cause pneumonia, doctors have to perform multiple tests to identify the infection.

But new types of blood tests are simplifying and speeding up the process.

Carois, based in California, uses artificial intelligence to help accurately identify the pathogen of pneumonia within 24 hours and choose the appropriate antibiotic.

“Before our test, a patient with pneumonia would have 15 to 20 different tests to identify it in just their first week in the hospital — that’s about $20,000 in tests,” says Alec Ford, CEO of Carius.

Karius has a microbial DNA database containing tens of billions of data points. Test samples from patients can be compared against that database to identify the exact pathogen.

Ford says this would have been impossible without artificial intelligence.

One challenge is that researchers do not necessarily currently understand all the connections that AI might make between test biomarkers and diseases.

Over the past two years, Dr. Slav Petrovsky has developed an AI platform called Milton that uses biomarkers in UK Biobank data to identify 120 diseases with a success rate of more than 90%.

Finding patterns in such a huge amount of data is just something artificial intelligence can do.

“These are often complex patterns, where there may not be a single biomarker, but you have to take into account the whole pattern,” says Dr. Petrovsky, who works as a researcher at pharmaceutical giant AstraZeneca.

Dr. Heller uses a similar pattern-matching technique in his work on ovarian cancer.

“We know that the sensor binds to and responds to proteins and small molecules in the blood, but we don’t know which proteins or molecules are specific to cancer,” he says.

More broadly, data, or the lack of it, remains a barrier.

“People don’t share their data, or there’s no mechanism for doing so,” says Ms Moran.

Ocra funds a large-scale patient registry, with patients’ electronic medical records allowing researchers to train algorithms on their data.

“It’s early days, and we’re still in the Wild West of AI now,” says Ms. Moran.