Table 7.

Advanced AI applications in prosthodontics.

Authors	Summarized Abstract	Methods Used	Results	Conclusions
Lee [57]	-Deep CNN algorithm accurately identified and classifies dental implant systems.	-Image preprocessing and transfer learning techniques were applied. -Utilized fine-tuned and pre-trained deep CNN architecture (GoogLeNet Inception-v3).	-Inception-v3 architecture demonstrated the best performance for classification tasks. -Straumann BLT implant system had the highest accuracy among implant types.	-Deep CNN architecture effective for dental implant system identification and classification.
Bayrakdar [58]	-AI system evaluated for dental implant planning in CBCT images.	-Manual assessment with InvivoDental 6.0 for bone height and thickness. -Deep convolutional neural network (Diagnocat, Inc., San Francisco, USA) for evaluations.	-AI system showed no significant differences in bone height measurements.	-AI systems aid in implant planning, supporting physicians in practice. -AI demonstrates high detection percentages for canals, sinuses, and missing teeth.
Lerner [59]	-AI used to fabricate implant-supported zirconia crowns with high success. -Study protocol included CAD design, milling, sintering, and clinical application.	-Intraoral scan of implant position, CAD design of abutment, milling of zirconia abutment, clinical application of hybrid abutment.	-Quality of fabrication of hybrid abutments had a mean deviation of 44 μm. -Three-year cumulative survival and success rates for MZCs were 99.0% and 91.3%.	-MZCs show excellent marginal adaptation, interproximal, and occlusal contacts.
Takahashi [60]	-Deep learning identified dental implants from radiographic images effectively. -Object detection algorithm used to identify six implant systems accurately.	-Deep learning with Yolov3 algorithm for implant system identification. -Utilized TensorFlow and Keras deep learning libraries for implementation.	-True positive ratio and average precision of implant systems evaluated.	-Implants can be identified from panoramic radiographic images using deep learning. -Identification system could assist dentists and patients with implant related problems.
Yamaguchi [61]	-AI predicted debonding probability of CAD/CAM CR crowns from 2D images.	-Deep learning with a CNN method to predict debonding probability. -Utilized STL models of die, adjacent teeth, and antagonist.	-Mean calculation time for test images was 2 ms/step. -Area under the curve (AUC) value was 0.998.	-Deep learning with CNN predicts CAD/CAM CR crown debonding probability accurately.