Prediction of Difficult Mask Ventilation Using 3D-Facescan and Machine Learning

Proof-of-principle Study for the Prediction of Difficult Mask Ventilation Using 3D-Facescan and Machine Learning

The aim of this study is to prove feasibility and assess the diagnostic performance of a machine learning algorithm that relies on data from 3D-face scans with predefined motion-sequences and scenes (MASCAN algorithm), together with patient-specific meta-data for the prediction of difficult mask ventilation. A secondary aim of the study is to verify whether voice and breathing scans improve the performance of the algorithm. From the clinical point of view, we believe that an automated assessment would be beneficial, as it preserves time and health-care resources while acting observer-independent, thus providing a rational, reproducible risk estimation.

Inclusion Criteria:

• Patients scheduling for ENT or OMS surgery in general anaesthesia, who require facemask ventilation and tracheal intubation after induction of anesthesia
• Patients aged at least 18 years
• Ability to understand the patient information and to personally sign and date the informed consent to participate in the study
• The patient is co-operative and available for the entire study
• Provided informed consent/patient representative

Exclusion Criteria:

• Pregnant or breastfeeding woman
• Rapid sequence induction or other contraindications for facemask ventilation
• Planned awake tracheal intubation