Accuracy of Navigation and Robot-Assisted Systems for Dental Implant Placement: A Systematic Review

<b>Background</b>: Computer-assisted implant surgery (CAIS) aims to improve placement accuracy versus freehand drilling. We compared the three-dimensional accuracy of robot-guided CAIS (r-CAIS), dynamic navigation (d-CAIS), static-template guidance (s-CAIS), and freehand (FH) in clinical and in vitro settings. <b>Methods</b>: We searched PubMed/MEDLINE, Scopus, and Web of Science (1 January 2019-2025). Eligible populations were adults receiving conventional or zygomatic implants in vivo, plus validated in vitro human-jaw models using plan-versus-placement workflows; studies had to report study-level means with dispersion for ≥1 primary outcome with ≥5 implants per arm. Interventions were r-CAIS, d-CAIS, or s-CAIS; with a baseline as the freehand technique. Risk of bias used RoB 2 (RCTs), ROBINS-I (non-randomized clinical), and QUIN (in vitro). Because of heterogeneity in definitions and workflows, we performed a descriptive synthesis by modality (no meta-analysis). <b>Registration</b>: OSF. <b>Results</b>: Forty-three studies (7 RCTs, 10 non-randomized clinical, 26 in vitro) reported more than 4000 implants. Across studies, typical study-level means for global linear deviation clustered around < 1 mm (r-CAIS), ~1 mm (d-CAIS), ~1.3 mm (s-CAIS), and ~1.8 mm (FH). In clinical contexts, d-CAIS often showed slightly lower angular deviation than s-CAIS. <b>Conclusions</b>: CAIS improves accuracy versus freehand. d-CAIS and s-CAIS show similar linear accuracy, with d-CAIS frequently yielding slightly lower angular deviation; r-CAIS exhibits tight error clusters in our dataset, but limited comparative clinical evidence precludes superiority claims. <b>Limitations</b>: non-uniform registration/measurement, variable operator experience, and absence of meta-analysis.