The beagles at Kibbutz Hama'apil and the cancer they can smell

Doges Editorial · 2026-06-13 · 6 min read

At Kibbutz Hama'apil in central Israel, twelve beagles are learning to detect lung, breast, prostate, and colorectal cancers from breath samples — with 94 percent accuracy in a peer-reviewed double-blind study. The startup behind them wants to turn that ability into a $250 mail-in screening test.

The beagle enters the testing room at a trot. She swings toward the row of sample containers — six of them, evenly spaced, each holding a breath sample collected from a different person. She works systematically, nose down, pausing at each container for exactly the amount of time she needs, then moving on. In a separate room, technicians and staff watch through glass. The handler stands close but doesn't know which container holds the cancer sample. The dog does, or will shortly.

This is the daily routine at Kibbutz Hama'apil, a small agricultural community in central Israel, where SpotitEarly — a medtech company with laboratories in New Jersey — is training twelve beagles as part of a cancer detection system that achieved 93.9 percent sensitivity and 94.3 percent specificity across 1,386 participants in a peer-reviewed double-blind trial. That result was published in Nature's Scientific Reports in 2024. The story, reported by JNS journalist Etgar Lefkovits on June 11, 2026, begins with a question the data makes harder to dismiss: what if the most sophisticated biosensor on the planet has been living in our houses for ten thousand years?

250 million against 5 million

The numbers first. Human beings have roughly five million scent receptors. Beagles have approximately 250 million. That fifty-fold difference isn't just a matter of scale; it reflects a qualitatively different relationship with chemical information. Where a human nose registers a smell as a general impression — coffee, rain, smoke — a beagle's nose receives it as a detailed chemical profile, parsing individual compounds within a complex mixture at concentrations as low as parts per trillion.

Cancer cells have a metabolic signature. As they grow, they produce volatile organic compounds — VOCs — that enter the bloodstream, travel to the lungs, and are exhaled in trace amounts with every breath. A trained beagle can reliably distinguish the exhalation of a person with an early-stage tumor from the exhalation of someone without one. The question SpotitEarly's founders asked was whether that ability could be packaged into something scalable.

How it works

The proposed screening process is disarmingly simple. A person orders a mask online — similar in form to an N95 — breathes into it according to the provided instructions, reseals it, and mails it back to SpotitEarly's laboratory. There, a trained beagle screens the sample alongside five others, none of which are known to the handler. If the dog indicates a positive, the sample is flagged for follow-up.

The company was founded by tech entrepreneurs and former commanders from the Israeli Defense Forces' elite canine unit — people who spent careers understanding what trained dogs are capable of and decided to apply that knowledge to a medical problem. Their R&D operation is at the kibbutz, where the dozen beagles — ages two to four, a mix described as Floridians, Europeans, and native Israelis — live and work.

What happens on the analysis side is more complex than the dog's alert alone. SpotitEarly's LUCID platform collects thousands of data points per second from the dogs' behavioral responses — body posture, nose movement, hesitation patterns — and runs them through proprietary AI algorithms to determine whether a sample should be flagged. This bio-hybrid design, combining olfactory detection with computational interpretation, is what allows the system to be standardized and reproduced rather than depending on a single handler's judgment.

We took on a mission to use the ability of dogs and transform it into a marketable product that can save many lives through early detection.

— Udi Bobrovsky, co-founder and COO, SpotitEarly

The Rainbow Study

The trial behind the 94 percent figure was called the Rainbow Study. Published in Nature's Scientific Reports in 2024, it enrolled 1,386 participants across three Israeli medical centers: Tel Aviv Sourasky Medical Center, Rambam Health Care Campus, and Hadassah Medical Center. The study was fully double-blind — neither dog handlers, sample processors, nor AI analysts knew which samples came from cancer patients. All 338 cancer-positive samples were randomized, anonymized, and processed independently.

The breakdown by cancer type: breast and lung both hit 95.0 percent sensitivity; prostate reached 93.0 percent; colorectal came in at 90.0 percent. For a pre-screening tool, the most telling figure is early-stage performance — stages 0 through 2 were detected at 94.8 percent sensitivity. All nine stage-0 samples in the study were identified correctly.

To put those numbers in context: digital mammography detects breast cancer with sensitivity typically ranging from 70 to 87 percent depending on population and equipment. SpotitEarly's beagles reached 95.0 percent sensitivity for breast cancer, including at its earliest stages — when the American Cancer Society notes five-year survival rates approach 99 percent. The dogs aren't replacing mammography. They are designed as a first-pass screen, prompting earlier clinical follow-up, catching tumors when treatment is most effective.

The team behind the noses

Irit Gazit is SpotitEarly's canine production manager. She has worked with dogs for three decades, across military, law enforcement, and now biomedical contexts. She watches the beagles in the laboratory with the eye of someone who knows what dogs can and cannot do, and takes no particular satisfaction in overstating either.

The scent of dogs is absolutely extraordinary and is unmatched by all the AI technology in the world. If anyone can detect it, they can.

— Irit Gazit, canine production manager, SpotitEarly

Ehud Cappon is one of the dog handlers, a former soldier from the IDF's elite canine unit. He works double-blind: when a beagle moves through the testing room, the handler genuinely does not know which of the six samples contains the positive specimen. This prevents the subtle unconscious cues — a tension in posture, a shift in attention — that can inadvertently guide a dog toward an expected answer.

This is a type of dream job.

— Ehud Cappon, dog handler, SpotitEarly

Why beagles

The choice of breed is deliberate. Beagles were bred as scent hounds — their entire evolutionary purpose, for centuries, was following a smell through complicated terrain until it resolved into something findable. Their noses are among the most sensitive in the domestic dog world, but what distinguishes them for this kind of work is temperament as much as anatomy. Beagles are sociable, focused, and resistant to distraction. They are not easily spooked by unfamiliar environments, which matters when a laboratory setting is their office.

They are also, on first glance, entirely ordinary. The dozen dogs at Kibbutz Hama'apil rush into the testing room like dogs who are happy to be at work — tail movement, eager attention, the particular canine expression of being in the place they are supposed to be. Nothing about them announces what they can do.

The broader significance

SpotitEarly is not the first organization to train dogs to detect cancer from biological samples. Medical Detection Dogs in the United Kingdom has been doing similar work for years, and in April 2026 published what it described as the world's largest study of its kind, reporting greater than 90 percent accuracy across multiple cancer types in breath samples. What SpotitEarly is attempting is to industrialize the process — turn what has so far been a proof-of-concept into something accessible at scale, for $250, through the mail, without requiring a clinic visit or a referral.

The company has raised $20.3 million to date and is building toward US clinical validation, including a partnership with Hackensack Meridian Health on a breast cancer trial and validation work with the University of Pennsylvania. One fully operational lab running 18 to 20 dogs, the company calculates, can process approximately one million tests per year. Two follow-up studies are underway before the 2027 commercial launch — which has not yet received FDA approval, adding regulatory time to the timeline. But the peer-reviewed results are, by any measure, striking enough to take seriously.

There is something worth noticing in what Irit Gazit said about the limits of artificial intelligence. She did not say it with defensiveness or sentimentality. She said it as a statement of professional assessment: the nose is extraordinary and unmatched. In an era when AI's capabilities in medicine are reliably described as revolutionary, a beagle at Kibbutz Hama'apil is doing something no algorithm has yet replicated at the same accuracy, at the same cost, with the same calm focus. The beagle finishes her last sample, signals, and waits for her reward. She doesn't know what a tumor is. She just knows what it smells like.