Data-driven modeling of dose-dependent chlorhexidine release from glass ionomer cements using gradient boosting...

Area of Science:

• Biomaterials Science
• Dental Materials Science
• Computational Chemistry

Background:

• Long-term prediction of antimicrobial release from bioactive dental materials is complex due to nonlinear kinetics.
• Accurate forecasting is crucial for the design and clinical application of effective antimicrobial dental materials.

Purpose of the Study:

• To apply the extreme gradient boosting (XGBoost) predictor for modeling sustained chlorhexidine (CHX) release from glass ionomer cements (GICs).
• To enable accurate long-term predictions of CHX release beyond experimental observation periods.
• To develop an accessible and uncertainty-aware tool for researchers and clinicians.

Main Methods:

• Trained XGBoost models using cumulative CHX release data from CHX-hexametaphosphate (CHX-HMP) functionalized GICs over 663 days.
• Utilized Bayesian hyperparameter optimization to maximize model performance.
• Developed closed-form predictive equations and an online graphical user interface (GUI) on the Streamlit platform.

Main Results:

• Optimized XGBoost models showed high predictive accuracy for cumulative CHX release and daily release rates across various dose levels.
• Exponential forecasting equations achieved R² scores greater than 0.92 for all dose levels.
• Conformal prediction provided reliable prediction intervals for the daily CHX release rate.

Conclusions:

• Accessible, uncertainty-aware models for predicting long-term CHX release from antimicrobial GICs have been developed.
• The approach facilitates reliable forecasting beyond experimental timeframes, supporting bioactive dental material design.
• The online tool aids practical adoption, addressing challenges in long-term antimicrobial release prediction in restorative dentistry.