Non-invasive multiple cancer screening using trained detection canines and artificial intelligence: a prospective...

Area of Science:

• Oncology
• Biotechnology
• Artificial Intelligence

Background:

• Early cancer detection significantly improves patient outcomes and survival rates.
• Current screening methods can be invasive, costly, or lack sensitivity for certain cancer types.
• Breath analysis offers a promising avenue for non-invasive, accessible cancer screening.

Purpose of the Study:

• To evaluate the sensitivity and specificity of a novel non-invasive multi-cancer screening method.
• To assess the performance of a bio-hybrid platform combining trained canines and artificial intelligence (AI) in detecting specific cancers from breath samples.
• To determine the platform's effectiveness in early-stage cancer detection.

Main Methods:

• A double-blind study was conducted with 1386 participants undergoing standard cancer screening or biopsy.
• Breath samples were collected and analyzed using a bio-hybrid platform integrating trained detection canines and AI algorithms.
• Performance metrics including sensitivity and specificity were calculated based on confirmed cancer diagnoses.

Main Results:

• The bio-hybrid platform achieved an overall sensitivity of 93.9% and specificity of 94.3% for detecting four trained cancer types (breast, lung, prostate, colorectal).
• Individual cancer detection sensitivities were high: breast (95.0%), lung (95.0%), prostate (93.0%), and colorectal (90.0%).
• The platform also demonstrated a sensitivity of 81.8% for detecting 14 other malignant tumors not specifically trained for, and 94.8% for early-stage cancers (0-2).

Conclusions:

• The bio-hybrid platform is a highly sensitive and specific non-invasive method for multi-cancer screening.
• This technology shows significant potential for early cancer detection across various types.
• The integration of canine olfaction and AI offers a powerful tool for future cancer diagnostics.