Canine-AI Collaboration Detects Cancer Odors in Breath with 94% Accuracy

A research study published in Scientific Reports unveils an innovative screening method that combines dogs and artificial intelligence (AI) to detect cancer odors in patients' breath. The novel partnership proves highly precise and sensitive, successfully identifying breast cancer, lung cancer, colorectal cancer, and prostate cancer with an accuracy rate as high as 94%.

Of particular note is the method's exceptional performance in both early and late-stage cancer detection. According to SpotitEarly, the Israeli company behind the technology's development, early detection is crucial for improving cancer survival rates.

During the study, the team trained Labrador retrievers to sniff breath samples. When the dogs detected the scent of cancer, they would sit down as a signal. While this cue may seem straightforward, continuously and accurately interpreting a dog's body language is no easy feat for humans. This is where artificial intelligence plays a pivotal role. Researchers devised an AI model based on machine learning and computer vision to identify and interpret the dogs' behavior.

Collaborating with Israeli medical centers, the team examined breath samples from nearly 1,400 participants, including 261 individuals diagnosed with one of the four types of cancer. With the assistance of AI, the dogs accurately identified 245 positive cases while maintaining an impressively low false positive rate for negative samples—out of 1,048 negative cases, only 60 were mistakenly classified as positive, resulting in a false positive rate of approximately 6%.