Development of a machine learning-based predictive model for maxillary sinus cysts and exploration of clustering...

Area of Science:

• Dentistry
• Medical Imaging
• Artificial Intelligence

Background:

• Maxillary sinus cysts present ongoing controversies regarding influencing factors and clinical management.
• Accurate prediction and risk stratification are crucial for effective patient care.

Purpose of the Study:

• To develop a predictive model for maxillary sinus cysts using cone beam computerized tomography (CBCT) and machine learning (ML).
• To explore the clustering patterns of maxillary sinus cysts to identify high-risk groups.
• To provide a theoretical foundation for the prevention and clinical management of maxillary sinus cysts.

Main Methods:

• Evaluated 6000 CBCT images from 3093 patients to identify influencing factors.
• Employed statistical methods and ML algorithms, including eXtreme Gradient Boosting (XGBoost), to construct a prediction model.
• Utilized SHapley Additive exPlanations (SHAP) for model interpretation and K-mean clustering for risk factor identification.

Main Results:

• The XGBoost model demonstrated excellent classification ability with an average area under the curve (AUC) of 0.921 in the test set.
• Cluster analysis identified apical lesions, severe periodontitis, and age ≥ 53 as significant high-risk factors for maxillary sinus cysts.
• A web-based calculator was developed for predicting maxillary sinus cyst risk.

Conclusions:

• The study provides valuable insights into the etiology and risk stratification of maxillary sinus cysts.
• Integration of CBCT imaging and ML techniques offers potential for personalized prevention and treatment strategies.
• Findings support evidence-based approaches for managing maxillary sinus cysts.