Dogs Team Up with AI to Aid in Cancer Detection
A new screening method combining the unmatched olfactory abilities of dogs with cutting-edge AI is making waves in cancer detection. Developed by SpotitEarly, an Israel-based company, the method was recently found to accurately detect four types of cancer — breast, lung, colorectal, and prostate — with an impressive 94% success rate, according to a study published in Scientific Reports on November 15.
The process hinges on dogs’ remarkable sense of smell to identify cancer signatures from patient breath samples. “The new screening method taps dogs’ amazing olfactory capabilities,” said Assaf Rabinowicz, chief technology officer at SpotitEarly.
Early detection is pivotal in improving cancer survival rates, and this approach has shown equal effectiveness in identifying early-stage cancers as it has with advanced cases.
For the study, Labrador retrievers — Mars, Moon, and Pluto — were trained to sit when detecting cancer in breath samples. However, interpreting their signals consistently presented challenges for human handlers. To overcome this, SpotitEarly integrated machine learning and computer vision into their AI model, enabling it to accurately interpret the dogs’ cues.
In collaboration with medical centers in Israel, the team tested nearly 1,400 breath samples, 261 of which belonged to individuals already diagnosed with one of the cancers being studied. The canine-AI system successfully identified 245 of these cases and demonstrated high specificity, with only 60 false positives among 1,048 negative samples.
“The combination of dogs’ skills and AI’s precision offers a game-changing approach to early cancer detection,” Rabinowicz noted.
SpotitEarly has ambitious plans for the future, including a larger clinical trial in the United States, with preliminary results expected by 2026. The company is now incorporating beagles into the research due to their smaller size and ease of training.
However, the Labrador retrievers who pioneered the project, including Mars, Moon, and Pluto, continue to contribute to development efforts. “The Labs are very beautiful and very friendly,” Rabinowicz said, adding, “They did a very good job.”
This groundbreaking method not only underscores the potential of cross-species collaboration but also highlights the promise of AI in enhancing medical diagnostics. As the approach moves toward broader clinical trials, it could herald a new era in non-invasive and highly accurate cancer screening methods.
With early detection being a cornerstone of cancer survival, the partnership between dogs and AI might just prove to be a lifesaving innovation.