Head-to-Head: AI and Human Workflows for Single-Unit Crown Design—Systematic Review

Résumé fourni par la source

Objectives: To compare artificial intelligence (AI) crown design with expert or non-AI computer-aided (CAD) design for single-unit tooth and implant-supported crowns across efficiency, marginal and internal fit, morphology and occlusion, and mechanical performance. Materials and Methods: This systematic review was conducted and reported in accordance with PRISMA 2020. PubMed MEDLINE, Scopus, Web of Science, IEEE Xplore, and Dentistry and Oral Sciences Source were searched from 2016 to 2025 with citation chasing. Eligible studies directly contrasted artificial intelligence-generated or artificial intelligence-assisted crown designs with human design in clinical, ex vivo, or in silico settings. Primary outcomes were design time, marginal and internal fit, morphology and occlusion, and mechanical performance. Risk of bias was assessed with ROBINS-I for non-randomized clinical studies, QUIN for bench studies, and PROBAST + AI for computational investigations, with TRIPOD + AI items mapped descriptively. Given heterogeneity in settings and endpoints, a narrative synthesis was used. Results: A total of 14 studies met inclusion criteria, including a clinical patient study, multiple ex vivo experiments, and in silico evaluations. Artificial intelligence design reduced design time by between 40% and 90% relative to expert computer-aided design or manual workflows. Marginal and internal fit for artificial intelligence and human designs were statistically equivalent in multiple comparisons. Mechanical performance matched technician designs in load-to-fracture testing, and modeling indicated stress distributions similar to natural teeth. Overall risk of bias was judged as some concerns across tiers. Conclusions: Artificial intelligence crown design delivers efficiency gains while showing short-term technical comparability across fit, morphology, occlusion, and strength for single-unit crowns in predominantly bench and in silico evidence, with limited patient-level feasibility data. Prospective clinical trials with standardized, preregistered endpoints are needed to confirm durability, generalizability, and patient-relevant outcomes, and to establish whether short-term technical advantages translate into clinical benefit.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.