Cancer Breath Test Stuns Oncologists

Scientist holding a test tube with red liquid — Photo: Rapeepat Pornsipak / Shutterstock

A Bengaluru startup says a simple breath test guided by dogs and artificial intelligence flagged cancers with about 90% accuracy in a 1,502-person trial.

Story Snapshot

Dognosis reports 90%+ sensitivity and specificity across seven cancers in Phase 2 results.
Trained dogs sniff breath samples; sensors and software turn signals into a cancer score.
The company touts stable performance at early stages, when treatment works best.
Backers pitch a low-cost, at-home screening path if future trials confirm the results.

Company Reports Strong Accuracy In Large Breath-Test Trial

Dognosis, a Bengaluru company, announced Phase 2 trial results that claim over 90 percent sensitivity and specificity in detecting cancer from breath samples. The study, described as the largest of its kind with 1,502 participants, covered seven major cancers and reported steady sensitivity at Stage 1 and Stage 2 disease. The company said results were published in the Journal of Clinical Oncology, and framed the method as a potential multi-cancer early detection tool based on odor patterns in exhaled breath.

The firm’s posts highlight an area under the curve of 0.962 for its “canine-powered Bayesian model,” which blends dog responses with machine learning. While company materials present a confident outlook, the claims still come from the developer and its communications. The core finding is straightforward: in this trial set, the breath approach reached about 90 percent accuracy across cancer types, including in earlier stages where screening can most change outcomes.

How The Dogs-And-Data System Works

The system starts with a ten-minute breath collection using a custom mask. The mask captures volatile organic compounds thought to reflect disease. Trained dogs then sniff the sealed sample and show responses that are measured with sensors, including a noninvasive headset that tracks brain waves and behavior. Software converts those signals into a quantitative signature and a cancer likelihood score, which can be scaled across many samples over time.

Company pages describe this as teaching machines how to smell by first reading the dog’s olfactory “decision” and then mapping it into a model. The aim is to deliver a repeatable, low-cost screen without needing a dog in every clinic. Dognosis markets the approach as noninvasive, affordable, and suitable for at-home kits that can ship to a lab for analysis if later studies support real-world use.

Why Early Cancer Signals In Breath Matter

Researchers have chased breath-based cancer detection for years because catching disease early saves lives and money. Cancer cells release chemical compounds that can change a person’s breath profile. Dogs have long served as a sensitive front end for this search, with published work showing dogs can flag several diseases by smell. The new results suggest a path to turn that skill into a scalable screening tool with sensors and code guiding the process.

For patients, a simple breath test could remove barriers that keep people from screening, like cost, travel, and fear of invasive procedures. For health systems and insurers, a cheap, multi-cancer screen could shift spending from late-stage care to early treatment. That promise resonates across the political spectrum, where many say the health system is too expensive and too complex. A lower-cost screen that actually works would help families, not just big institutions.

What Comes Next For Access And Trust

Dognosis and partners will need further trials that test accuracy across clinics, ages, and real-world settings. Health authorities will also look at how a positive breath test should guide follow-up scans and biopsies. If later studies confirm these Phase 2 numbers, regulators and payers will face pressure to move faster on affordable tools that find cancer early. People want care that is simple, fair, and effective, not locked up by red tape or elite interests.