Predicting difficult airway intubation in thyroid surgery using multiple machine learning and deep learning...

Area of Science:

• Anesthesiology
• Medical Informatics
• Artificial Intelligence

Background:

• Predicting difficult airway intubation is crucial for patient safety during thyroid surgery.
• Machine learning (ML) and deep learning (DL) offer potential tools for improving airway management prediction.

Purpose of the Study:

• To evaluate the efficacy of ML and DL algorithms in predicting difficult airway intubation in patients undergoing thyroid surgery.
• To identify key patient factors contributing to difficult airway prediction.

Main Methods:

• Ten ML and DL algorithms were trained and validated to predict difficult airways.
• Statistical analysis was performed using R, with ML models developed in Python.
• Gradient Boosting, Convolutional Neural Networks (CNN), and Light Gradient Boosting Machine (LGBM) were among the algorithms tested.

Main Results:

• Gradient Boosting achieved the highest performance in the test group, with an Area Under the Curve (AUC) of 0.848, accuracy of 0.913, and precision of 100%.
• Key predictors for difficult airways identified by ML included age, sex, weight, height, and Body Mass Index (BMI).
• In the training group, Gradient Boosting showed an AUC of 0.932 and accuracy of 0.929.

Conclusions:

• ML and DL, particularly Gradient Boosting, are effective tools for predicting difficult airway intubation in thyroid surgery patients.
• Patient demographics such as age, sex, weight, height, and BMI are significant factors in predicting difficult airways.