Machine Learning Predictions and Identifying Key Predictors for Safer Intubation: A Study on Video Laryngoscopy Views

Area of Science:

• Medical Informatics
• Anesthesiology
• Machine Learning

Background:

• Effective airway management is crucial for patient safety during anesthesia.
• Video laryngoscopy offers improved visualization but requires skill and can be challenging.
• Predictive models can aid in optimizing intubation success rates.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting video laryngoscopic view quality.
• To identify key patient factors influencing the percentage of glottic opening (POGO) score.
• To enhance the efficiency and safety of airway management through predictive analytics.

Main Methods:

• Utilized a dataset of 212 participants (169 training, 43 testing).
• Applied various machine learning algorithms: Random Forest, Light Gradient Boosting Machine, K-Nearest Neighbors, Support Vector Regression, Ridge Regression, and Lasso Regression.
• Assessed model performance using Root Mean Squared Error (RMSE) and SHapley Additive exPlanations (SHAP) for feature importance.

Main Results:

• Machine learning models achieved RMSE values between 20.4 and 21.9.
• SHAP analysis consistently identified age as a significant predictor of POGO score across all tested models.
• The models demonstrated the potential for accurate prediction of laryngoscopic view quality.

Conclusions:

• Machine learning models show promise in predicting video laryngoscopic view quality.
• Patient age is a critical factor influencing the percentage of glottic opening (POGO) score.
• These predictive insights can contribute to safer and more efficient airway management strategies.