The VOCALISE project
The COVID-19 pandemic has highlighted the importance of having rapid diagnostic tools available for general population screening of viral respiratory infections (VRIs). The VOCALISE project explores the real-time analysis of exhaled air to detect infection markers and develop a new diagnostic method that is non-invasive, reliable, and suitable for large-scale use.
The project builds on the COVID-AIR clinical study conducted in Lyon, which recruited nearly 5,000 patients and measured the composition of exhaled air in real time using mass spectrometry. Based on this data, we are developing an analysis pipeline that combines biostatistics and artificial intelligence. To this end, we are comparing various data analysis approaches, including both unsupervised and supervised methods. The different phases of the breath are also being analyzed in detail to extract biomarkers relevant to the diagnosis of these infections. Through these combined approaches, we are evaluating the method’s ability to identify atypical biomarker profiles that could help detect emerging pathogens.
To achieve these objectives, the multidisciplinary VOCALISE consortium brings together teams from Lyon with complementary expertise: virology and clinical medicine (CIRI/CARVI – Université Lyon 1 & Hospices Civils de Lyon), signal processing and artificial intelligence (Physics Laboratory, SiSyphe team – ENS de Lyon, CNRS), biostatistics and health data (Laboratory of Biometry and Evolutionary Biology & HCL Public Health Department), and analytical and atmospheric chemistry (IRCELYON – CARE team, CNRS).
Ultimately, the VOCALISE project aims to establish a robust, explainable, and reproducible procedure that ensures a “breath → measurement → data → models → clinical validation” continuum, thereby accelerating the clinical implementation of breath analysis and feeding into the VORTEX project (PEPR-MIE).
Project leaders:
Alexandre GAYMARD
Alexandre Gaymard is a virologist at the Hospices Civils de Lyon and the International Center for Infectiology Research (CIRI, équipe CARVI-VirPath). An Associate Professor and Hospital Practitioner (MCU-PH) at Lyon 1 Université, he practices at the Institute of Infectious Agents (affiliated with the CNR des virus des infections respiratoires). He coordinates the VOCALISE consortium and leads the development and evaluation of new diagnostic tools and data processing; he also coordinates the VORTEX project (PEPR-MIE), which aims to accelerate the clinical implementation of breath analysis for respiratory infections.
Pierre BORGNAT
Pierre Borgnat is a CNRS Research Director specializing in signal processing and data analysis at the ENS de Lyon Physics Laboratory (UMR CNRS 5672), within the Signals, Systems, and Physics (Sisyphe) team, which he leads. He teaches at the ENS and heads the “Complex Systems Modeling” Master 2 program. Within VOCALISE, he leads the methodological development in signal processing and artificial intelligence—focusing particularly on graphs and anomaly detection—to transform raw breath data into robust, explainable biomarkers.
