Dogs have detected cancer for decades. Now a machine learns from them.

Doges Editorial · 2026-05-31 · 5 min read

SpotitEarly's LUCID 2.0, unveiled May 28, uses trained dogs as biological sensors and AI as the decision-maker — scaling a dog's uncanny nose into a clinical cancer screening platform capable of 1.73 million tests per year.

The room is quiet except for the movement of a dog working its way along a row of small sealed ports. Each one holds a breath sample, collected earlier through a simple face mask. The dog pauses at one port, lingers a fraction longer than the others, and shifts her weight almost imperceptibly. To the camera above her, that shift is data. To the AI reading the camera, it is a signal. To the person whose breath is in that tube, it could be the earliest warning they will ever receive.

On May 28, 2026, SpotitEarly — a biotech startup based in Englewood, New Jersey — unveiled LUCID 2.0, the next generation of its Bio-AI Hybrid platform for breath-based cancer screening. Built to process up to 1.73 million tests per year per screening center, the platform pairs trained detection dogs with a real-time AI system that reads the dogs' micro-behavioral responses and converts each sniff into a clinical confidence score. Source: BusinessWire (https://www.businesswire.com/news/home/20260528372891/en/AI-Meets-Olfaction-SpotitEarly-Unveils-LUCID-2.0-Scaling-Early-Cancer-Detection-to-Millions), May 28, 2026.

A Nose That No Machine Has Matched

The science behind canine cancer detection is not new. Researchers have known for more than two decades that trained dogs can detect the chemical signature of certain cancers in breath or urine samples. Multiple peer-reviewed studies document dogs identifying lung cancer, breast cancer, colorectal cancer, and prostate cancer — in some trials with sensitivity exceeding 94 percent across all four types. The mechanism is well established: cancer cells produce volatile organic compounds (VOCs) that healthy cells don't. A dog's olfactory system, equipped with 250 to 300 million scent receptors, is between 10,000 and 100,000 times more sensitive than a human's. Roughly 30 percent of a dog's brain is dedicated to processing smell.

The problem has always been what comes next. A dog who smells something isn't automatically producing a result that a clinical system can certify, replicate, or submit to regulatory review. The subjective read — a trainer who notices a dog's body language and makes a judgment call — doesn't survive the scrutiny that medical diagnostics require. So the science stayed in academic journals, while clinical diagnostics moved forward with machines that, at best, approximated what the dogs were already doing.

The Translation Machine

LUCID 2.0 is built to close that gap. As a detection dog works through a sample at a SpotitEarly station, the platform captures hundreds of data points every second from cameras, audio sensors, and motion arrays built into the screening port. The behavioral micro-signals — shifts in posture, changes in breathing rate, the particular stillness that precedes a positive detection — are tracked continuously. Crucially, many of these signals occur before the dog consciously registers the scent: the body responds first, at a speed below the animal's own awareness, in the same way a human's pulse changes before they realize they're frightened.

Every sniff resolves into a 0-to-100 confidence score. The dog provides the biological detection; the AI provides the final clinical decision. The architecture is deliberately designed for replicability — each LUCID Station is built to be installed in certified labs worldwide, starting with SpotitEarly's own CLIA-registered facility.

LUCID 2.0 is what allows our science to operate as a real medical platform at clinical scale. There is a growing body of peer-reviewed literature demonstrating the ability of dogs to sniff cancer. Our platform is the first one to operationalize that science into a reproducible end-to-end process that patients and clinicians can trust.

— Shlomi Madar, PhD, CEO of SpotitEarly

Dogs Working in Committee

One architectural detail is worth pausing on. SpotitEarly doesn't rely on a single dog to make a call. Multiple dogs work the same sample, and the platform synthesizes their responses — effectively building a committee consensus, the way a tumor board might approach a difficult diagnosis. The company notes, with understandable understatement, that the dogs experience this as play. A dog who thinks she's playing is probably more consistent than one who senses she's being evaluated. Variability in individual detection dogs has historically been one of the field's credibility problems; running multiple dogs on each sample and aggregating the results addresses it directly.

What Early Actually Means

The word "early" in oncology carries enormous weight. For most major cancer types, the difference between a stage-one diagnosis and a stage-four one is measured not only in treatment options but in survival probability. A non-invasive breath test that catches a chemical signal before a patient has symptoms — before they would ever think to schedule an imaging scan — could shift those outcomes for millions of people annually. SpotitEarly is still subject to regulatory approval, but the platform's clinical infrastructure is already running.

The seamless integration of natural scent detection principles with advanced AI is a true breakthrough in the field. Utilizing Volatile Organic Compounds for cancer detection is an innovative approach that marks a major leap forward.

— Dr. Reef Einoch Amor, PhD, Director of Chemical Analysis, SpotitEarly

The Nose You Already Know

There is something clarifying about this technology if you have ever watched your own dog work through a trail of scent — the pause at a specific spot on the sidewalk, the way she circles back and approaches something from a different angle, the full 30-second stop at a patch of grass that looks identical to the patch of grass 10 feet back. She isn't performing curiosity. She is processing information at a resolution that human instruments are still learning to measure. The LUCID platform is, in a sense, an attempt to put instruments around something that was always happening — to hear what the dogs have been saying all along.

Dogs have been reading chemical signals out of the air for as long as they have been dogs. The question has always been whether we could build a system sensitive enough to translate what they find.