Cost-effectiveness of Artificial Intelligence for Proximal Caries Detection.

Area of Science:

• Dental diagnostics
• Artificial intelligence in healthcare
• Cost-effectiveness analysis

Background:

• Artificial intelligence (AI) shows potential in assisting dentists with image assessment, particularly for caries detection.
• The broader health and economic implications of integrating AI into dental diagnostics remain largely unevaluated.
• Evaluating the cost-effectiveness of AI for proximal caries detection on bitewing radiographs is crucial for understanding its clinical value.

Purpose of the Study:

• To compare the cost-effectiveness of proximal caries detection on bitewing radiographs with and without the assistance of artificial intelligence (AI).
• To assess the impact of AI on diagnostic accuracy, patient outcomes (tooth retention), and healthcare costs from a German mixed-payer perspective.

Main Methods:

• A U-Net convolutional neural network was trained, validated, and tested on 3,293, 252, and 141 bitewing radiographs, respectively.
• A Markov model simulated patient outcomes over a lifetime, considering true/false positives/negatives and subsequent treatment decisions.
• Cost-effectiveness was evaluated using incremental cost-effectiveness ratio (ICER) and Monte-Carlo microsimulations, with tooth retention years as the health outcome.

Main Results:

• AI demonstrated higher accuracy (0.80) compared to dentists (mean 0.71) and significantly greater sensitivity (0.75 vs. 0.36).
• AI-assisted detection was associated with improved tooth retention (64 years) and lower costs (298 euro) versus non-AI assessment (62 years; 322 euro).
• The ICER indicated that AI was dominant (cost-saving and more effective), with AI being less costly and more effective in over 77% of scenarios.

Conclusions:

• Implementing AI for dental caries detection on bitewing radiographs is likely cost-effective, primarily due to improved detection rates and reduced missed lesions.
• The cost-effectiveness is contingent on dentists managing detected early-stage lesions non-restoratively, aligning with AI's diagnostic capabilities.
• AI offers a valuable tool for enhancing diagnostic accuracy and optimizing resource allocation in dental care.