Digitized canine olfaction and multimodal biosensing for breath-based multi-cancer detection: a hypothesis-driven perspective

Multi-cancer early detection (MCED) is critical for reducing cancer mortality, however current screening technologies have limitations in accessibility, cost, and early stage sensitivity. Breath-based detection using volatile organic compounds (VOCs) offers a non-invasive and scalable alternative, with trained detection dogs demonstrating exceptional olfactory sensitivity in clinical studies. However, the widespread deployment of canine scent detection has been hindered by subjectivity and lack of standardization. This perspective article proposes a novel canine olfaction based MCED platform that analyzes olfactory recognition using electroencephalography (EEG) signals from trained dogs, combined with behavioral and physiological biomarkers (respiration, heart rate, vision tracking), and integrates these modalities via machine learning to produce a cancer risk score. We review the biological and computational foundations of this approach, present preliminary validation strategies, and outline a phased experimental roadmap from EEG signal capture to clinical deployment. This bio-digital framework exemplified by the Dognosis platform offers a pathway toward reproducible, portable, scalable and low-cost cancer detection systems, particularly suited for deployment in low-resource settings. The proposed system may also extend to detection of infections, neurological disorders, and other VOC-associated diseases.