Dental Composite Performance Prediction Using Artificial Intelligence.

Area of Science:

• Materials Science
• Biomaterials Engineering
• Computational Science

Background:

• Dental composites require enhanced performance and longevity.
• Accelerating the market translation of novel dental composites is crucial.
• Predictive modeling can optimize composite development.

Purpose of the Study:

• To employ artificial intelligence (AI), specifically machine learning (ML), for predicting dental composite performance outcomes (POs).
• To evaluate the efficacy of various ML models in predicting discrete and continuous POs.
• To identify key composite attributes influencing material performance.

Main Methods:

• Curated a comprehensive dataset from over 200 publications.
• Trained nine ML models for discrete PO prediction and five for continuous PO regression.
• Evaluated model performance using metrics like accuracy and receiver-operating characteristic area under the curve.

Main Results:

• Different ML models demonstrated varying strengths in predicting specific POs (e.g., KNN for flexural modulus, Decision Tree for flexural strength).
• Random Forest showed high efficacy for flexural strength and volumetric shrinkage.
• Feature importance analysis identified key chemical components and physical properties influencing composite performance.

Conclusions:

• AI and ML models show significant potential for predicting dental composite properties.
• Utilizing diverse ML models and large datasets is essential for robust predictions.
• This approach can facilitate the optimization of composite properties and accelerate new material development.