The article discusses the effectiveness of an AI-enhanced breath analysis method in detecting COVID-19. This method was presented at the European Respiratory Society International Congress. The AI-selected clinical data, combined with breath analysis data, showed a notable improvement in distinguishing COVID-19 in patients, boasting a 98% sensitivity and a 98% negative predictive value.
The real-time mass spectrometry breath analysis, enhanced by artificial intelligence, is being considered as a potential screening method for COVID-19. The test takes less than a minute to perform, making it a high-throughput method. It could be particularly useful for quickly identifying COVID-19-negative individuals, with confirmatory tests then being used for those who test positive.
In the study, 173 adults were assessed who needed COVID-19 screening to evaluate the efficacy of this AI-enhanced method. The breath analysis used real-time proton transfer reaction time-of-flight mass spectrometry to analyze volatile organic compounds in exhaled breath. This data was then either used alone or combined with patient metadata and applied to AI and machine learning techniques to build mathematical models. The results were compared with those from the reverse transcription polymerase chain reaction (RT-PCR) tests.
The inventor of this method, Grassin Delyle, is also reported to be an inventor on a patent application covering breath analysis in COVID-19.