Background/Objectives: A novel tool based on artificial intelligence (AIT) to evaluate immediate loading feasibility was recently introduced. The aim of this study is to evaluate the correspondence between the AIT prediction and the operator's evaluation in a large sample of implants. Methods: 11 operators were asked to classify the recorded insertion curves of all the implants placed during the period between September 2022 and August 2023 as suitable or non-suitable for immediate loading. Next, the same curves were analyzed by the AIT, which classified them as belonging to YES (suitable for immediate loading) or NO (non-suitable for immediate loading) class. Results: 1320 dental implants were placed and a total of 21 different implant systems were used. According to the surgeons' evaluation, 999 curves were suitable and 321 were non-suitable for immediate loading; when evaluated by the AIT, 916 curves belonged to class YES, while 404 belonged to class NO. The resulted sensitivity was 90.49% (95% CI = 88.5% to 92.2), specificity was 96.26% (95% CI = 93.6% to 98.1%), PPV was 98.7% (95% CI = 97.7% to 99.3%), and NPV was 76.5% (95% CI = 72% to 80.5%). Conclusions: The AIT tested in the present clinical multicenter study demonstrated a high level of accuracy in the prediction of immediate loading feasibility.
Keywords: artificial intelligence; dental implants; immediate loading; insertion torque; primary stability.