Impact of artificial intelligence (AI) on predicting marginal fit and aesthetic outcomes for custom implant abutments

TL;DR

This study shows that AI can accurately predict the fit and aesthetics of custom dental implants, improving clinical outcomes.

Contribution

A custom CNN model was developed to predict marginal fit and aesthetics of implant abutments with high accuracy.

Findings

The AI model achieved 93.5% accuracy for marginal fit within a 25 µ tolerance.

Strong correlations (r = 0.89 for fit, r = 0.82-0.85 for aesthetics) were observed between AI predictions and actual outcomes.

AI can reduce adjustments and remakes, improving clinical success rates in implant prosthodontics.

Abstract

The integration of artificial intelligence into implant prosthodontics enhances the precision of pre-fabrication predictions for
clinical success. In this study, a custom convolutional neural network (CNN) model achieved a prediction accuracy of 93.5% for marginal
fit within a 25 µ tolerance and 87.6% concordance with clinical evaluations of custom implant abutment aesthetics. The AI-estimated
mean spatial gap (78.6±18.2 µ) closely approximated the actual measurement (82.3 ± 21.5 µ), with strong
correlations observed between predicted and actual outcomes for both fit (r = 0.89) and aesthetic appeal (r = 0.82-0.85). Thus, we show
the potential of AI as a preventive quality assessment tool in implant prosthodontics, capable of minimizing adjustments and remakes
while improving overall clinical success rates.