Abstract
This review aims to explore the application of AI (artificial intelligence) in fixed prosthodontics and implant-supported fixed restorations, with a specific focus on the accuracy, effectiveness, and clinical applicability of AI models for optimizing treatment planning and predicting clinical outcomes. This review followed the PCC (Population, Concept, Context) framework. A systematic search was conducted using PubMed, Scopus, and Embase for studies published between January 2010 and July 2025. Keywords included "artificial intelligence,", "deep learning", "digital dentistry", "prosthodontic treatment planning," "clinical decision support," and "outcome prediction." The initial databases search yielded 834 results. After selection, 20 studies were included for analysis in this review. AI applications were grouped into four domains: implant planning, crown design, full-arch framework optimization, and prognostic modeling. Convolutional neural networks (CNNs), generative adversarial networks (GANs), regression models, and optimization algorithms were most frequently employed. In implant planning, AI achieved high accuracies (90-99.5%) for site detection, drilling protocols, and bone assessment. Crown design studies demonstrated occlusal and morphological deviations within clinically acceptable thresholds (0.18-0.30 mm) and mean internal gaps of 59-83 μm, while reducing design time by up to fourfold compared with conventional workflows. For full-arch prosthetics, optimization methods such as particle swarm optimization (PSO) and bi-evolutionary structural optimization (BESO) enhanced efficiency and reduced stress concentration in simulated workflows. Prognostic modeling showed promising performance, with models achieving over 90% accuracy in predicting implant survival and treatment outcomes. AI applications in prosthodontics are most developed in implant site planning and crown design, with fewer studies on full-arch optimization and prognosis. Models generally achieve high accuracy and efficiency, but most evidence is early-stage and simulation-based. Future research should focus on prospective validation, multimodal integration, and patient-centered outcomes to ensure clinical reliability.