Prediction of the extubation outcome through Electrical Impedance Tomography measurements

Vincent Janiak (1) , Vincent Jousselin , Gwendoline Tallec , Christophe Marsala (2) , Umar Saleem , Martin Dres , Andrea Pinna (1)

1 SYEL - Systèmes Electroniques
2 LFI - Learning, Fuzzy and Intelligent systems

Extubation failure can occur in 10 to 25% of patients who were successfully separated from their ventilator. In which case, patients need to be re-intubated. To reduce the extubation failure rate, we monitor patients during 48h after extubation using Electrical Impedance Tomography (EIT). In total, we recruited 34 patients from which 8 failed their extubation in the ICU ward. Prediction of extubation success or failure using only non-invasive EIT data show a sensitivity of 0.80 and a specificity of 0.73. The prediction for the 8 extubation failures was accurate for most right after the extubation and achieved 100% accuracy for the measurement set preceding the failure.

Mots clés: Electrical impedance tomography; Mechanical ventilation; Weaning; Failure prediction

Domaines: Traitement du signal et de l'image [eess.SP]; Pneumologie et système respiratoire

Vincent Janiak, Vincent Jousselin, Gwendoline Tallec, Christophe Marsala, Umar Saleem, et al.. Prediction of the extubation outcome through Electrical Impedance Tomography measurements. IEEE Biomedical Circuits and Systems (BIOCAS), Oct 2022, Tapei, Taiwan. ⟨hal-03821945⟩