AI-Powered Breath Detector Diagnoses 17 Different Diseases


Pioneer Founding member

Click for video and pictures:

Our breath contains a slew of information about our health in the form of molecules whose existence and concentration can serve as biomarkers for disease. Typically breath sensors focus on a single biomarker and therefore are limited in their scope and screening ability. A worldwide scientific collaboration headed by a team from Technion−Israel Institute of Technology has developed a breath sensor capable of detecting many different molecules and correlated these biomarkers to 17 different diseases.

The device consists of an array of specially prepared gold nanoparticle sensors and ones based on a random network of single-walled carbon nanotubes. While it is impressive on its own, what gave it special powers was to use it to collect breath samples from thousands of patients with different diseases and to use artificial intelligence software to find correlations in the data. Thanks to the AI component of this research, the average diagnostic accuracy of the system was demonstrated at 86% for 17 different diseases, including such disparate conditions as a number of cancers, Crohn’s disease, two types of Parkinson’s, preeclampsia, and pulmonary hypertension.