That old methodology could soon get a fresh lease of life, courtesy of new research from the American Chemical Society. What researchers involved with the project have established are unique “breathprint” identifiers for a range of different diseases — including ones like kidney cancer and Parkinson’s disease that you might not immediately assume come with a signature smell.
“In this study, we tested the ability of a nano-based intelligent technology to detect and classify a large number of diseases by the analysis of exhaled breath,” Professor Hossam Haick, one of the researchers on the project, told Digital Trends. “We analyzed breath samples obtained from 1,404 subjects, either categorized as healthy controls or having one of 17 different diseases. The nano-based intelligent system was able, after a training phase, to diagnose and classify the samples according to the healthy condition [or] disease of the donor with 86 percent accuracy.”
The analysis was carried out using a technology called mass spectrometry, a technique in which chemical species are ionized and the ions sorted based on their mass-to-charge ratio. The team discovered that each disease produces a unique volatile chemical scent, based on differing quantities of 13 components.
“Our system is based on an array of nano-materials-based sensors, modified with a sensing organic layer,” Haick continued. “Bioinspired by the mammalian sense of smell, these sensors perform semi-selective analysis of the target volatile molecules in exhaled breath, and with artificial intelligence [are capable of learning and detecting] the unique patterns associated with each disease.”
Impressively, the AI tool is able to detect additional diseases even when one is already present.
Right now, the work exists in the form of a research paper, although Haick said that the plan is to develop a breathalyzer-type portable technology that could be used for early diagnosis in the field. “This is an ongoing process and hopefully soon could reach the daily use in the clinics,” he noted.
