Ainos, an artificial intelligence and biotechnology company, will collaborate with researchers from National Taiwan University (NTU) on a one-year research program starting in July to assess whether the company’s Smell AI platform can diagnose causes of dyspnoea by analysing volatile organic compounds (VOCs) in patients’ exhaled breath.
Dyspnoea (shortness of breath) is among the most common complaints in emergency departments and can signal a variety of conditions. The study will specifically examine whether the system can distinguish between acute exacerbation of chronic obstructive pulmonary disease (AECOPD) and acute decompensated heart failure (ADHF), two conditions that require different treatments.
Technology: Smell AI and AI Nose
The core of Ainos’s Smell AI platform is the AI Nose module, which contains multiple micro-electromechanical system (MEMS) sensors and an integrated digital processor. According to the company, the sensors’ electrical resistance rises in the presence of certain gases; that change is converted into a digital signal and interpreted by the system in a way the company likens to how a human nose processes odors.
Interpretation is performed by Ainos’s proprietary Smell Language Model, trained to learn, classify and contextualize complex scent patterns.
Eddy Tsai, chief executive officer of Ainos, said: “The AI Nose was originally developed for medical diagnostic applications where non-invasive sensing, accuracy and validation in realistic environments are critical. This research programme brings that experience into a high-value clinical environment and expands our Smell AI platform into digital breath analysis.”
Potential impact
Ainos and NTU aim to develop and validate a system that can identify the presence of AECOPD and/or ADHF in patients presenting with dyspnoea using VOC-based “breathprints.” If successful, the project could contribute to the creation of a breathprint database related to shortness of breath and support future research in emergency, outpatient and even home-monitoring settings.
Background testing
This study follows a previous programme in which the AI Nose system was trialled on an active emergency department at National Taiwan University Hospital. There the solution was used for tracking respiratory infections and monitoring crowding in waiting rooms, treatment areas and observation zones.
Limits and outlook
Scent-based diagnostics offer a promising non-invasive and rapid triage tool, but they require clinical validation, particularly for a symptom as heterogeneous as dyspnoea, which can stem from diverse underlying diseases. The Ainos–NTU collaboration represents an initial clinical step in exploring broader healthcare applications of digital breath intelligence.
