269 - Artificial Intelligence as a Decision-Support Tool for Vital Pulp Therapy in Primary Molars

Abstract: In pediatric dentistry, the diagnostic “gray zone” between reversible and irreversible pulpitis remains a major clinical challenge. Decisions between Indirect Pulp Treatment (IPT), Direct Pulp Capping, and Pulpotomy are often based on subjective interpretation of radiographs and inconsistent clinical indicators. With the shift toward minimally invasive, biologically driven endodontic approaches, Artificial Intelligence (AI) has emerged as a promising tool to support objective and data-driven decision-making. This systematic review evaluates the current evidence on AI-based models for predicting treatment outcomes in pediatric vital pulp therapy (VPT).



A comprehensive search of PubMed, Scopus, and Embase (2018–2025) identified studies employing Machine Learning (ML) and Deep Learning (DL) for VPT planning in primary teeth. From 142 initial records, 18 studies met the inclusion criteria. These focused primarily on radiographic features such as caries-to-pulp distance, dentin density, and the integration of multimodal clinical variables.



The results indicate that Convolutional Neural Networks (CNNs) achieved pooled diagnostic accuracies exceeding 88% in predicting pulp exposure, outperforming less-experienced clinicians. More advanced multimodal models, combining radiographic data with patient age and symptom history, demonstrated improved prognostic capability for predicting 12-month success of bioceramic pulp therapies. However, a persistent limitation is the lack of explainability, as many models do not provide biologically interpretable outputs necessary for clinician trust.



In conclusion, AI currently functions best as a decision-support system rather than a standalone diagnostic tool. When integrated thoughtfully, AI may reduce overtreatment, standardize care across varying experience levels, and enhance long-term tooth preservation in pediatric patients.

Identify Supporting Agency and Grant Number: Reseach supported by Azad university of Tehran