The Dognosis AI cancer breath test has recorded accuracy above 90 percent across seven cancer types in a study of more than 1,500 participants, according to the Bengaluru-based startup behind it. The trial, conducted across six hospitals in Karnataka, India, has not yet been independently verified, and researchers say much larger studies will be needed before the method can be considered clinically reliable.
The underlying concept is straightforward, even if the engineering behind it is not. Patients breathe into a cotton mask for around 10 minutes, and the sealed sample is sent to the lab. They never interact with a dog.
How the Dognosis AI cancer breath test actually works
At Dognosis, 15 dogs are fitted with helmets and harnesses carrying sensors that track their breathing, brain activity, and movement as they sniff sealed breath samples. The dogs are not making the diagnosis. The AI processes the biometric data and identifies patterns in how the animals respond to odour compounds associated with cancer in human breath. Those involuntary, physical reactions are the raw material the system trains on.
The goal is to make something dogs can already do consistent and repeatable at scale. Dognosis’s current product is called BreathEasy, a non-invasive, breath-based health monitoring tool that, according to Dognosis, generates a volatile organic compound (VOC) risk score for clinicians, rather than a standalone diagnosis.
The study behind the 90 percent figure was conducted in collaboration with Medical Detection Dogs, a UK-based charity described by Dognosis as a world leader in canine bio-detection research. Medical Detection Dogs has previously demonstrated that dogs can detect the odour of diseases including prostate and bladder cancer, Parkinson’s disease, malaria, and COVID-19, according to Dognosis’s published account of the trial.
The six hospitals and what the data shows
Six hospitals in Karnataka participated: RadOn Cancer Centre, Karnataka Cancer Therapy and Research Institute, and Karnataka Medical College and Research Institute, all in Hubli, plus three others in the state. The study enrolled more than 1,500 participants and returned accuracy above 90 percent across seven cancer types. Those figures come from Dognosis and have not been independently verified.
Oncologist Santhosh Kumar Devadas was measured in his response. ‘The study is promising, as an early signal,’ he said. ‘But they need to do the study with a much larger group before we make any definitive conclusions.’
That caveat matters. A single company-reported study, however well-designed, is not the same as peer-reviewed, independently replicated evidence. Dognosis has published its phase 2 trial in the Journal of Clinical Oncology, which places it in the scientific record, but external researchers have been careful to frame it as preliminary.
The context that makes this worth watching is the state of cancer detection in India. A large share of diagnoses come only after the disease has already advanced, partly because accessible, non-invasive early-screening options are limited. A breath test that flags people who may need further investigation, without requiring imaging, blood draws, or specialist access at the first step, would address a real gap. Dognosis is pitching BreathEasy as a first-pass filter, not a replacement for conventional diagnostics, and says it is pursuing further trials toward potential commercial approval.
The 90 percent accuracy figure is the number that will draw attention, and it should also be read alongside what it represents: a promising phase 2 result from the company building the product, covering seven cancer types, validated across six hospitals in one Indian state. Independent, larger-scale trials are the next necessary step, and Dognosis has said it intends to pursue them.
