AI nose uses ‘Smell Language Model’ to sniff out signs of disease

AI AND ML

Sampling patients’ breath may save lives and emergency room resources

Many people worry about what AI knows, but what about an AI Nose that can smell what disease you might have?

Ainos, an AI and biotech company that is developing smell technology, is working with National Taiwan University (NTU) to explore whether its platform can help diagnose patients by analyzing volatile organic compounds (VOCs) in exhaled breath.

The year-long research effort, which starts in July, will examine individuals who present with dyspnea, or shortness of breath, said to be one of the most common symptoms seen in emergency departments.

Dyspnea can be a symptom of many conditions, including acute exacerbation of chronic obstructive pulmonary disease (AECOPD) and acute decompensated heart failure (ADHF), each of which requires different treatments.

Ainos and NTU hope to develop and evaluate a system to analyze VOC-based breathprints to detect AECOPD and/or ADHF in patients.

Ainos’s Smell AI platform relies on an AI Nose module that features multiple micro-electro-mechanical system (MEMS) sensors and an integrated digital processor. Sensor resistance increases in the presence of detectable gases, and this is converted to a digital signal that is interpreted in much the way the human nose interprets scents, according to Ainos.

That interpretation is handled by by a proprietary Smell Language Model that has been developed to learn, classify, and contextualize complex scent patterns.

“AI Nose was originally developed with medical diagnostic applications in mind, where non-invasive sensing, accuracy, and real-world validation are essential,” said Ainos CEO Eddy Tsai.

“This research program brings that experience back into a high-value clinical setting and extends our Smell AI platform into digital breath intelligence.”

Not content with “digital breath intelligence,” a term we must confess to not being too familiar with, the the company frames the research as part of its broader vision of “building Smell ID data and Smell Language Model capabilities across healthcare, industrial, and physical AI environments.”

If successful, the research could help create a breathprint database for dyspnea and support future studies for emergency, outpatient, and even home-monitoring settings.

The research follows a separate program testing the AI Nose in an active emergency department at National Taiwan University Hospital. The system has been deployed to monitor respiratory infections and overcrowding in waiting areas, treatment areas, and observation zones. ®