Enhancing predictive analytics in mandibular third molar extraction using artificial intelligence: A CBCT-Based...

Area of Science:

• Oral and Maxillofacial Surgery
• Artificial Intelligence in Dentistry
• Radiology and Imaging

Background:

• Predicting mandibular third molar extraction difficulty is vital for surgical planning and complication avoidance.
• Cone-beam computed tomography (CBCT) reports contain valuable data for assessing extraction complexity.
• Current methods for assessing difficulty can be subjective and time-consuming.

Purpose of the Study:

• To develop and validate an AI-driven predictive model for forecasting the difficulty of mandibular third molar extractions.
• To utilize CBCT report data for training a deep learning model to classify extraction difficulty.
• To enhance clinical decision-making in third molar surgery through accurate AI-based predictions.

Main Methods:

• A retrospective study analyzed 738 CBCT reports of mandibular third molars.
• Natural Language Processing (NLP) extracted features like angulation, root morphology, and root-nerve proximity.
• A deep learning neural network was trained on these features to classify extraction difficulty into four categories.

Main Results:

• The AI classification model achieved 95% accuracy on both training and validation datasets.
• High precision (0.97) and recall (0.95) were observed in the training set.
• The validation set showed strong performance with precision of 0.97 and recall of 0.89.

Conclusions:

• AI-based models reliably forecast the complexity of mandibular third molar extractions from CBCT reports.
• The developed AI model accurately predicts extraction difficulty, supporting informed clinical decisions.
• Utilizing AI can potentially lead to improved patient outcomes in third molar surgery.