AI-enhanced dogs detect cancer markers in breath with over 90% accuracy
A Bengaluru startup, Dognosis, combined trained dogs with artificial intelligence to identify cancer-related odors in breath samples, achieving sensitivity and specificity above 90 percent in a study of more than 1,500 participants.
Researchers at Dognosis trained fifteen dogs to sniff cotton masks that volunteers breathed into for about ten minutes. Sensors recorded the dogs' breathing, brain activity and movements while AI interpreted their responses. The trial, which involved over 1,500 people, showed the method could correctly flag cancer cases more than 90 percent of the time. Dogs work only 30 to 45 minutes a day, making the approach potentially low-cost for initial screening before standard medical tests.
Why it matters
It shows a low-cost, AI-assisted method could help detect cancer early, expanding screening access worldwide.
In this story
