Ventilator-associated pneumonia (VAP) is the most common nosocomial infection in intensive care units (ICUs), primarily affecting patients requiring invasive mechanical ventilation. This condition not only increases potential complications but also extends hospital stays and significantly raises healthcare costs. Early detection is crucial for initiating timely treatment, reducing complications, and promoting faster recovery. However, despite its significant clinical impact, accurately and promptly identifying VAP remains an unsolved challenge.
To tackle this complex issue, the project Early Identification of Ventilator-Associated Pneumonia Using Machine Learning Techniques, funded by the Swiss Innovation Agency (Innosuisse), leverages Artificial Intelligence (AI) to support clinicians in the early detection of VAP.
“By analyzing data collected from ventilators, we have been able to identify very small changes in respiratory pattern that are predictive of the onset of VAP,” explains Laura Azzimonti, SUPSI Senior-researcher and lecturer. “These early changes are typically too subtle to be perceived by humans and often become evident only when the condition has already progressed.”
The novelty of the project lies in using data directly extracted from mechanical ventilators, recorded continuously and automatically without human intervention. These data, combined with the clinical characteristics of patients, enable the AI system to provide a reliable diagnostic tool. The system integrates seamlessly with bedside ventilators, making it practical and suitable for everyday clinical use.
The potential of the AI system goes beyond clinical practice. By facilitating the early and accurate detection of VAP, it ensures timely and effective treatments, improving patient outcomes and generating significant savings in healthcare costs. It also serves as a valuable tool in addressing the global challenge of antibiotic resistance, limiting excessive use and promoting a more sustainable therapeutic approach.
The project, led by the SUPSI research group in Knowledge-based AI at the Dalle Molle Institute for Artificial Intelligence (IDSIA USI-SUPSI), spearheaded by Laura Azzimonti, is conducted in collaboration with the Intensive Care Unit, the Medical education and Research Area (AFRi), and the Information and Communication Technology Area (ICT) of the Ente Ospedaliero Cantonale (EOC).
This collaborative effort between clinicians and data scientists sets a new standard in tackling a complex and pervasive ICU challenge, demonstrating the transformative potential of AI in healthcare.