AI is trained to identify warning signs in blood tests.

This is the third feature in a six-part series looking at how AI is changing medical research and treatment.

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

As with all cancers, the earlier it is detected, the better.

Most ovarian cancer starts in the fallopian tubes, so by the time it reaches the ovaries, it has already spread elsewhere.

“Five years before any symptoms appear is when you need to detect ovarian cancer, to affect mortality,” says Ms Moran.

But new blood tests are emerging that use the power of artificial intelligence (AI) to detect signs of cancer in its earliest stages.

And it’s not just cancer, AI could speed up other blood tests for potentially deadly infections like pneumonia.

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

His team has developed a testing technology that uses nanotubes – tiny tubes of carbon that are 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 last decade, researchers learned how to change the properties of these nanotubes so that they respond to almost anything in the blood.

It is now possible to insert millions of nanotubes into a blood sample and have them emit different wavelengths of light based on what sticks to them.

But that still left the question of signal interpretation, which Dr. Heller likens to finding a match for a fingerprint.

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

But the patterns are so subtle that a human cannot be selected.

“We could look at the data and it wouldn’t make sense to us,” he says. “We can only see patterns that are different with AI.”

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 from people without it.

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

A major challenge in using AI to develop blood tests for ovarian cancer research is that it is relatively rare, which limits the data for training algorithms.

And even then, much of the data is hushed up at the hospitals that treated them, with minimal data sharing for researchers.

Dr. Heller describes training the algorithm on only a few 100 patients’ available data as a “Hail Mary pass.”

But he says AI was able to achieve better accuracy than the best cancer biomarkers available today — and that was just the first attempt.

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

Dr. Heller has high hopes for this technology.

“What we’d like to do is treat all gynecological diseases — so when someone comes in with a complaint, can we give doctors a device that tells them quickly whether it’s cancer or not, or This cancer is more than that.”

Dr. Heller says that could be “three to five years” away.

It’s not just early detection that AI is potentially useful, it’s also speeding up other blood tests.

For a cancer patient, getting pneumonia can be life-threatening, and since there are about 600 different organisms that can cause pneumonia, doctors have to run multiple tests to identify the infection.

But new blood tests are making the process easier and faster.

California-based Karuis uses artificial intelligence (AI) to help identify the exact cause of pneumonia within 24 hours, and select the right antibiotic for it.

“Prior to our test, a patient with pneumonia had an infection in their first week in hospital,” says Alec Ford, chief executive of Karius. 15 to 20 different tests will be performed to identify — that’s about $20,000 to check.”

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

Mr Ford says this would have been impossible without AI.

One challenge is that researchers don’t necessarily currently understand all the connections an AI test can make between biomarkers and diseases.

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

Finding patterns in such a large scale of data is just something that AI can do.

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

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

“We know that the sensor binds 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 extensive data, or the lack thereof, is still a drawback.

“People aren’t sharing their data, or there’s no mechanism to do so,” says Ms Moran.

Ocra is funding a large-scale patient registry with patients’ electronic medical records that have allowed researchers to train algorithms on their data.

“It’s early days – we’re still in the wild west of AI,” says Ms Moran.