At one part per trillion, a beagle smells early cancer

Doges Editorial · 2026-06-30 · 6 min read

The Rainbow Study — more than 1,400 participants, published in Scientific Reports in 2024 — found SpotitEarly's beagles detected breast, lung, colorectal, and prostate cancer with 95% sensitivity in early-stage cases. A new AI platform means they can now run tests at the scale of 1.7 million per year.

What a beagle's nose can do that no machine yet can

A dog's nose contains approximately 300 million olfactory receptors — 50 times more than a human's — and roughly 30 percent of a dog's brain function is dedicated to processing scent, compared with about 5 percent in humans. The result is a detection threshold estimated at somewhere between 10,000 and 100,000 times more sensitive than our own: dogs can identify individual odor compounds present at concentrations of one part per trillion. When cancer cells are active in the body, they produce volatile organic compounds — VOCs — that are released into the breath at exactly those concentrations. A beagle, it turns out, is extraordinarily well-matched to this problem.

SpotitEarly, a cancer-detection company operating across the US and Israel, has built a clinical pipeline around this fact. The company's CEO, Shlomi Madar, PhD, describes the biology plainly.

Canines have 300 million olfactory receptors. Like humans, they also have an organ called an olfactory bulb that is much more evolved than the olfactory bulb in humans. The olfactory bulb helps dogs separate the signal from the noise. With their extraordinary sense of smell, they experience an onslaught of scent, and the olfactory bulb allows them to focus on important scents.

Breathing into a mask, then waiting for the dogs

The test process is deliberately low-friction. A patient breathes into a face mask for three minutes. The mask is sealed in a container and sent to SpotitEarly's lab. There, a team of trained beagles works through a set of sniffing ports—four dogs to a team, three teams rotating through four dedicated rooms. Each dog follows its nose through the samples. When a dog identifies the VOC signature associated with cancer, it signals: typically by sitting down and pausing at that port.

The results from what SpotitEarly calls the Rainbow Study — the largest clinical trial to date for this approach — were published in Scientific Reports, a Nature group journal, in 2024 (PMC11568277). Researchers analyzed breath samples from more than 1,400 individuals between the ages of 22 and 94. Across breast, lung, colorectal, and prostate cancers, the system achieved 93.9% sensitivity and 94.3% specificity. But the number with the most clinical weight is the early-stage figure: for cancers at stages 0 through 2, sensitivity climbed to 95% — higher than the overall average, and at a detection window where treatment options are widest.

What LUCID 2.0 actually changes

The news this month is LUCID 2.0 — the updated AI platform that SpotitEarly launched to sit alongside its dogs. In earlier versions of the system, human handlers observed the beagles as they worked, watching for behavioral signals that indicated a positive detection. LUCID 2.0 replaces and augments that with a multimodal sensor array: video cameras tracking facial expressions and posture, microphones monitoring breathing rhythms, motion sensors measuring how quickly each dog moves between the sniffing ports. All of it is fed into a proprietary deep learning model.

The AI doesn't have a nose. What it has is a detailed, continuous data stream about how the beagle's body is behaving in the moments after it encounters each sample — and a trained model for distinguishing routine sniffing from the behavioral pattern a dog produces when it finds what it's looking for.

We have proprietary deep learning AI models that analyze the data such as breaking down video into frames and pixels to see whether the dogs are displaying behaviors that are different from routine gestures. Eventually, there is enough data to exclude anything that may skew the results. The AI generates a binary decision on whether there is a signal from a sample that indicates cancer.

Why a happy beagle is also a more accurate one

SpotitEarly's dogs work two hours a day. That limit is not incidental — it is structural to the protocol. A tired dog loses acuity and motivation. An anxious or disengaged dog produces unreliable signals. The dogs need to want to engage with the work, which means the work needs to feel like something they enjoy. Happy beagles, in this context, are not an animal-welfare nicety. They are a clinical requirement.

That constraint is also the basis of the scalability math behind LUCID 2.0. With four rooms and three teams of four dogs each, all working within a two-hour daily window, SpotitEarly says the new platform allows them to process the equivalent of 1.7 million breath tests per year — a throughput that exceeds many standard diagnostic labs.

The P.I.N.K. Study and what comes next

The most significant near-term test for LUCID 2.0 is the P.I.N.K. Study — a multi-site clinical trial focused on breast cancer detection, running simultaneously across the US and Israel, and led in the United States by Elias Obeid, MD, at Hackensack Meridian Health's Hennessy Institute for Cancer Prevention and Penn Medicine. Women already undergoing mammograms or biopsies are invited to also breathe into a SpotitEarly mask; researchers then cross-reference the dog's result against the actual imaging or biopsy outcome. Interim results are expected by the end of 2026. A separate multi-year collaboration with Fox Chase Cancer Center, announced in February 2026, will extend the same approach to lung cancer detection.

The technology studied in the P.I.N.K. study is an opportunity to add another tool in our toolbox for early breast cancer detection and diagnosis. The fact that it is a possible detection mode utilizing dogs' inherent sniffing ability alongside AI adds to our enthusiasm.

A different kind of working dog shift

Most working dogs people are familiar with — search and rescue, explosive detection, narcotics work — operate in physically demanding environments. SpotitEarly's beagles have a more unusual brief: they come to a climate-controlled lab, sniff a controlled set of ports, and clock out after two hours. They do not know they are looking for cancer. They know they are looking for a particular smell, and that finding it reliably is rewarding.

Whether any of the beagles has a sense of the stakes on the other side of the test is unknowable. But the humans who receive the results do. If the P.I.N.K. Study data holds, the practical implication is significant: a team of well-cared-for dogs working a short daily shift in a purpose-built lab could represent a meaningful expansion in who gets access to early cancer screening, and when. That is an unusual sentence to write about beagles. The data will determine whether it is also a true one.

This story was reported by Christopher Cheney for HealthLeaders Media and published June 23, 2026.