Video-Based Posture Analysis for Evaluating Endotracheal Intubation Performance.

Area of Science:

• Medical simulation
• Machine learning in healthcare
• Ergonomics and biomechanics

Background:

• Endotracheal intubation (ETI) is a critical airway procedure requiring effective training.
• Traditional ETI assessment methods are resource-intensive and require expert supervision.
• Objective and scalable training assessment tools are needed, especially in resource-constrained settings.

Purpose of the Study:

• To evaluate provider posture during endotracheal intubation (ETI).
• To explore the use of video-based posture analysis and machine learning for classifying successful ETI attempts.
• To develop an objective and scalable method for ETI performance assessment.

Main Methods:

• Novice participants performed ETI on a manikin.
• Videos were analyzed using pose detection to extract posture points.
• Machine learning (XGBoost) was employed with 10-fold cross-validation to classify ETI success based on posture features.
• Key spatial relationships, like elbow-wrist positions, were identified as predictive features.

Main Results:

• The machine learning model achieved 77.8% accuracy, 76.9% sensitivity, and 78.7% specificity in classifying successful intubations.
• Analysis identified specific provider postures and spatial relationships as significant predictors of ETI success.
• The model demonstrated the potential for objective performance evaluation.

Conclusions:

• Video-based posture analysis combined with machine learning offers a practical and scalable framework for assessing ETI performance.
• This approach can objectively evaluate ETI skills, minimizing the need for expert oversight.
• Further validation and integration into training programs can enhance ETI education, particularly in challenging environments.