A startup that wants to revolutionize cancer treatment by using artificial intelligence and rapid lab tests to find the most promising medicine for each individual patient has shown promise in initial clinical studies and is attracting interest from investors and Big Pharma.

This kind of precision medicine has been the dream of cancer treatment ever since the decoding of the human genome two decades ago. But it turns out genetic sequencing a patient’s tumors is frequently insufficient to pinpoint the best treatment. That’s partly because most of these tests only look for a few select genetic characteristics, or biomarkers, while the process that governs whether a particular cancer will respond to any given drug are highly complex.

In a 2020 study published in Nature, up to 90% of lung cancer patients who were heavy smokers had tumors that didn’t respond to therapies that genetic sequencing suggested were likely to work.

Allcyte, a three-year old company in Vienna, Austria, wants to solve this problem, and it’s come up with a radically simple approach. Instead of using genetic testing, it takes live cancer tissue samples from a patient and then tries a large number of cancer medicines on them in the lab to see which work best.

“Instead of trying to extrapolate from DNA, here we take the actual tumor, and we ask the tumor the question of what will work,” Nikolas Krall, Allcyte’s co-founder and chief executive, told Fortune. “This is about selecting the right drug for the right patient on an individual basis.”

Courtesy of Allcyte

In order to assess so many micro-experiments quickly, the company uses computer-vision technology, a kind of A.I. that can classify images, to observe the experiments and measure how the cells respond to each drug. It says it has screened hundreds of billions of cells in this way.

The approach, called “pharmacoscopy,” has already yielded some impressive results. In a clinical trial conducted by the Medical University of Vienna and presented at the American Society of Hematology’s annual conference in December, doctors used Allcyte’s technology to look for new treatments for 143 patients with aggressive, late-stage blood cancers. The patients had already exhausted all known treatment options. Allcyte screened tissue samples from these patients to see how they would respond to 136 different drugs, many of them not typically used for chemotherapy for these cancer types or even, in some cases, for cancer at all. In many cases, the patients were too ill to receive the treatment Allcyte’s system recommended. But in the 56 cases in which patients were treatable, 55% of the patients responded, with many surviving far longer than patients who did not receive recommended treatments.

“This could give patients a new lease on life,” Krall said. He said the company had already seen in cases of “a couple hundred patients” instances where “drugs that you would never expect to work are working and we haven’t even had time to scratch the surface.”

He said the company had found potential new uses for older drugs, which can be an important finding for pharmaceutical companies looking to extend their intellectual property rights for drugs that are coming to the end of their patent life. In addition to screening existing drugs, the company also has partnerships with 10 large pharmaceutical companies, including German pharma giant Boehringer Ingelheim, to use the same technology to assist in the discovery of new drugs.

Today, Allcyte announced it has raised an additional $6 million from a venture capital firms including Air Street Capital, from London; 42cap and Amino Collective from Germany; VP Venture Partners from Switzerland; and PUSH Ventures, from Austria.

While Allcyte’s pharmacoscopy approach is fairly unique, it is far from the only company trying to use A.I. methods in some manner to bring more precision medicines to cancer patients. Among these are Biodesix, a company in Boulder, Colorado, and Lantern Pharma in Dallas. And there are hundreds of startups trying to use various A.I. methods to accelerate new drug discovery.

Building the A.I. Allcyte uses wasn’t easy. Training a computer-vision system to analyze cancer cells requires a fairly large, specialized data set that has to be labeled by experts. “We went through thousands of images of cells,” Krall says, adding that the company used PhD. researchers to annotate the images. “It was a huge engineering challenge and a huge annotation challenge,” he says.

Krall, who is a chemist by training, founded Allcyte in 2017 along with Gregory Vladimer, an immunologist, who serves as the company’s chief science officer, and Berend Snijder and Giulio Superti-Furga, who are both molecular biologists. The team had been working together at The CeMM Research Center for Molecular Medicine of the Austrian Academy of Sciences in Vienna when they hit upon the technology and later spun the company out of the research center.

The company now employs about 30 people in Vienna. Krall says the company will use some of its new funding to hire more researchers and machine-learning experts, as well as to hire business employees. He said the company would likely open its own lab in the U.S. to offer its testing there. The company needs to seek regulatory approval for its methods in both the U.S. and Europe.

Krall also said Allcyte wants to expand its technique from blood cancers, where it has mostly been used so far, to solid tumors. But this is challenging, he says, because rather than analyzing a two-dimensional image, as its system can do with blood cancers, the company will need to create an A.I. system that can analyze three-dimensional imagery. One way to do this, Krall says, is to take tiny slices of tumor to study how drugs are interacting with cells in the tumor’s interior, not just on its surface. Some chemotherapy and immune therapies have trouble penetrating the interior of solid tumors so their effectiveness cannot be judged from imagery that only captures cells on the tumor’s surface.

This story has been updated to correct the number of pharmaceutical companies with which Allcyte currently has partnerships. That figure is 10 not 15. It has also been updated to include VP Venture Partners among the list of investors involved in the company’s latest funding round.