Accuracy of autonomous robotic computer-assisted implant surgery in edentulous patients: a retrospective exploratory case series

Abstract

Introduction

This study aimed to evaluate the accuracy of robotic computer-assisted implant surgery in edentulous patients and to compare differences in accuracy between implants placed in the maxilla and mandible, as well as between anterior and posterior regions.

Materials and methods

Six patients who voluntarily underwent autonomous robotic computer-assisted implant surgery (R-CAIS) using the YaKebot system and met the eligibility criteria were included in this study. Preoperative planning data and postoperative CBCT images were imported into the robotic software system for registration to calculate deviations between the designed and actual implant positions. The primary outcomes included global deviations at the implant platform and apex, angular deviation, as well as horizontal and vertical deviations at the platform and the apex. Deviations were compared among implants placed at different anatomical locations, with p < 0.05 considered statistically significant.

Results

Six edentulous patients underwent R-CAIS with a total of 42 implants placed. The mean deviation at the implant platform was 0.519 ± 0.216 mm, while the mean apex deviation was 0.508 ± 0.224 mm. The mean angular deviation was 0.634 ± 0.402°, and the mean vertical deviation was 0.298 ± 0.248 mm. The results showed that global deviations at the apex, global angular deviations, and platform vertical deviations tended to be lower in the anterior region than in the posterior region. However, none of these differences reached statistical significance (all p-values > 0.05). Implants in the maxilla exhibited overall smaller deviations than those in the mandible. Notably, the buccolingual deviation at the apex was significantly lower in the maxilla than in the mandible (p = 0.021). No other significant differences were found between the two jaws (all p > 0.05). In terms of deviation direction, the implants tended to deviate toward the buccal and distal directions and were positioned slightly more superficially than designed.

Conclusions

R-CAIS can achieve high accuracy in implant placement in edentulous patients, supporting its broader clinical application in complex implant rehabilitation.

Introduction

Edentulism is a common condition among the elderly, primarily caused by periodontal disease and dental caries [1, 2], although it may also result from trauma, congenital anomalies, and tumors. With the significant increase in life expectancy and the acceleration of population aging, the prevalence of edentulism has shown an increasing trend [3]. Edentulous jaws are often associated with moderate to severe alveolar bone resorption [4, 5]. Long-term tooth loss may also lead to degeneration of the oral mucosa, the temporomandibular joints, and other related structures, thereby impairing fundamental oral functions such as mastication, swallowing, and speech. Due to the loss of dental arch support, patients may experience alterations in facial morphology, often presenting with an aged appearance, which may further negatively affect their social interactions, psychological well-being, and overall quality of life [6, 7].

Conventional complete dentures provide limited retention and support, making it difficult to meet the functional demands of most edentulous patients, and often require a prolonged period of adaptation and adjustment. Long-term use may lead to progressive alveolar bone resorption, resulting in decreased denture retention and complications such as mucosal soreness [8]. Approximately 30%-50% of patients experience discomfort, including soreness, instability, and reduced masticatory efficiency, within 1–3 years of use [9]. Implant-supported rehabilitation has become an important treatment option for edentulous patients due to its superior stability, improved comfort, minimal foreign-body sensation, high long-term success rate, and its ability to effectively reduce alveolar bone resorption [10, 11]. Both implant-supported overdentures and fixed implant prostheses can significantly enhance prosthetic retention and stability, thereby improving oral function and quality of life. However, implant rehabilitation in edentulous patients remains technically challenging. Severe alveolar bone resorption often results in a knife-edge ridge, which may increase the risk of drill deviation during implant placement and injury to adjacent anatomical structures, such as the maxillary sinus and inferior alveolar nerve, thus increasing intraoperative risks [12].

With continuous advances in medical imaging, computer-aided design/computer-aided manufacturing (CAD/CAM), and information processing technologies, digital workflows have been increasingly integrated into implant dentistry. These technologies provide valuable tools for surgical planning and intraoperative guidance [13]. Digital impression technology enables the acquisition of highly accurate oral and skeletal data [14, 15], and surgical guides are commonly used to assist in implant placement [16]. Robotic computer-assisted implant surgery (R-CAIS) integrates real-time intraoperative navigation with automated robotic execution, thereby overcoming the limitations of static computer-assisted implant surgery (S-CAIS), including a lack of dynamic adjustment, limited control of implantation depth, and insufficient irrigation during osteotomies [17, 18]. It also addresses the shortcomings of dynamic computer-assisted implant surgery (D-CAIS), which may be influenced by the surgeon’s experience, hand-eye coordination, and fatigue [19]. Furthermore, it enables a transition from “intraoperative guidance” to “precise execution,” thereby improving surgical accuracy and procedural efficiency, while enhancing minimally invasive outcomes [20, 21].

The YaKebot autonomous dental implant robot (YaKebot Technology Co., Ltd., Beijing, China) consists of a hardware system that includes a robotic unit, a vision system, and related surgical accessories. The software system comprises a patient data management system and functional modules for image visualization, implant design, surgical guide design, implantation workflow planning, robotic motion control, and postoperative evaluation. The force-feedback control system of the implant robot enables real-time sensing, quantification, and regulation of mechanical changes during drilling and implant placement, thereby reducing trajectory deviation and depth errors caused by variations in bone quality and operational instability. This system serves as a critical foundation for achieving high-precision and predictable implant placement. Chen [22] conducted an in vitro study to investigate the relationship between force sensing and the primary stability of implants placed using a robotic system. The results demonstrated that force feedback is a key factor in ensuring predictable primary stability of implant placement.

R-CAIS has developed rapidly in recent years, and its clinical accuracy is of paramount importance. Previous in vitro and animal studies have demonstrated that R-CAIS can achieve high accuracy [23, 24], and recent case reports have also shown promising outcomes [25–27]. However, clinical evidence regarding the accuracy of R-CAIS in edentulous patients remains limited, and few studies have compared accuracy differences between the maxilla and mandible or between anterior and posterior implant sites. Furthermore, the factors influencing the accuracy of robotic systems have not yet been fully elucidated. Therefore, this retrospective exploratory study aimed to evaluate the accuracy of robotic computer-assisted implant placement in edentulous patients and to compare differences in accuracy between implants placed in the maxilla and mandible, as well as between anterior and posterior regions. The findings of this study may provide additional clinical evidence to support further development and optimization of robotic implant systems.

Materials and methods

Study design and patient selection

This study was a retrospective exploratory case series. The study population comprised edentulous patients who underwent R-CAIS at the Department of Stomatology, the First Affiliated Hospital of the University of Science and Technology of China, between June 2024 and December 2025. Prior to surgery, all patients were fully informed of the proposed treatment plan, as well as the risks associated with both conventional implant surgery and R-CAIS, and provided written informed consent. After initiation of the present study, eligible cases were retrospectively identified from the hospital database and enrolled according to predefined inclusion and exclusion criteria. Clinical records, surgical records (including video recordings), radiographic data, and available follow-up data were collected and analyzed.

Six patients who voluntarily underwent R-CAIS using the YaKebot system and met the eligibility criteria were included in this study. The cohort consisted of two females and four males, with a mean age of 55 ± 8.15 years. A total of seven edentulous arches from six patients were included. One patient was completely edentulous and five patients each had a single edentulous arch. Consequently, 42 implants were included in the final analysis. All patients provided written informed consent for both treatment and the use of their clinical data for research purposes. The study protocol was approved by the Ethics Committee of the First Affiliated Hospital of the University of Science and Technology of China (No. 2024-re4).

Inclusion and exclusion criteria

Inclusion criteria: (a) edentulous patients who voluntarily underwent R-CAIS using the YaKebot system; (b) availability of complete cone-beam computed tomography (CBCT) data with image quality sufficient for implant accuracy analysis; (c) availability of complete clinical and surgical records that could be collected and analyzed; (d) willingness and ability to attend scheduled postoperative follow-up visits.

Exclusion criteria: (a) incomplete clinical or radiographic records; (b) poor-quality CBCT images or severe imaging artifacts that precluded accurate measurement; (c) refusal or inability to participate in postoperative follow-up; (d) requirement for additional surgical intervention before completion of the 3-month postoperative follow-up period.

Preoperative preparation

All patients underwent routine preoperative assessments, including CBCT, examination of hemostasis and coagulation, electrocardiography (ECG), and serological testing for infectious diseases before surgery.

At the initial visit, a preliminary impression of the edentulous arch was obtained using alginate impression material. A custom tray was then fabricated using light-cured resin, followed by sectional border molding to obtain the final impression. After tooth arrangement on the articulator, a denture-type radiographic guide with radiopaque markers was fabricated (Fig. 1A). If the patient had an existing denture, radiopaque marker material was bonded to the denture, which was then used as a radiographic template.

Fig. 1.

Procedure for the cortical bone screw guide. A Radiographic guide; B Desktop scanning of the radiographic template; C-D Design and fabrication of the cortical bone screw guide; E-F Placement of cortical bone screws

After fabrication of the radiographic template, each patient was recalled for a second visit and underwent a CBCT scan while wearing the radiographic template. The radiographic template was scanned separately using a desktop scanner (Hefei Meyer Optoelectronic Technology Inc., Hefei, China). Digital scanning (Fig. 1B) was performed under natural lighting conditions [28].

CBCT data were saved in Digital Imaging and Communications in Medicine (DICOM) format, and digital scan data were saved in Standard Tessellation Language (STL) format. Both datasets were imported into the DentalNavi robotic software system (YaKebot Technology Co., Ltd., Beijing, China) for registration. Based on the fused CBCT and STL datasets, the positions of the cortical bone screws and the preliminary positions of the implants were virtually planned. The cortical bone screws were positioned away from the designed implant sites to avoid interference with implant placement. Subsequently, a cortical bone screw guide was designed and then fabricated using a NextDent 5100 3D printer (3D Systems, Rock Hill, SC, USA) (Fig. 1C-D).

All CBCT scans were acquired by the same technician using an SS-X9010DPro-3D CBCT system (Meyer Optoelectronic Technology Inc., Hefei, China) with identical imaging parameters. CBCT scans were required to cover the entire oral and maxillofacial region, including the height and thickness of the alveolar bone at the implant sites, as well as the location of critical anatomical structures such as the maxillary sinus and the mandibular canal.

Implant surgery

After fabrication of the cortical bone screw guide, the patient was recalled for a third visit, during which implant placement was performed using the YaKebot autonomous dental implant robotic system. Following routine disinfection and local infiltration anesthesia with articaine, the clinician manually inserted three cortical bone screws (2 × 10 mm) through the cortical bone screw guide, utilizing the self-tapping property of the screws (Fig. 1E -F).

The guide was then removed, and a CBCT scan and an intraoral scan were performed. The CBCT and intraoral scan datasets containing the cortical bone screws were imported into the DentalNavi robotic software system for image registration. Based on the registered datasets, the designed implant positions and angulations were determined, and a patient-specific surgical calibration guide was subsequently designed and fabricated using a NextDent 5100 3D printer (Fig. 2).

Fig. 2.

Design, fabrication, and intraoral fixation of the patient-specific surgical calibration guide. A Digital design of the guide. B Fabrication of the guide using 3D printing (C) Intraoral fixation of the guide

The patient was instructed to rinse with chlorhexidine solution. The dental chair position was adjusted according to the implant site. The clinician performed calibration of the robotic arm end-effector. Subsequently, the intraoral connector and the patient-specific surgical calibration guide were secured in place, followed by registration of the calibration guide (Fig. 3A). After completion of system calibration and spatial registration, the probe-tip mode was used to further verify registration accuracy, thereby ensuring the reliability of the registration outcome. The robotic arm automatically adjusted the position and orientation of its end-effector according to the preoperative implant plan, and the implant placement procedure was initiated (Fig. 3B).

Fig. 3.

Implant surgery. A Marker matching. B Drilling by the robotic system. C-D Preoperative and postoperative photographs

Following the Astra implant system protocol, a round bur was used to accurately mark the implant entry point. Osteotomy preparation was automatically performed by the robotic system. After each step, the system returned to the initial position, and the clinician manually replaced the drill with the next-sized drill. After sequential osteotomy preparation, the implant was automatically inserted by the robotic system. Implant placement was completed according to the preoperative design (Fig. 3C-D). Throughout the procedure, the infrared optical tracking system (FusionTrack 250, Atracsys, Switzerland) was used to monitor the spatial positions of the patient and the robotic end-effector in real time.

Measurement of implant deviations

After surgery, the calibration guide was removed, and a postoperative CBCT scan was obtained by the same technician using identical scanning parameters. The preoperative planning data and postoperative CBCT images were then imported into the DentalNavi robotic software system and registered to determine the actual implant positions.

The deviation measurements were reviewed and confirmed by two additional implant specialists from the department, and the final deviation values were determined by consensus. To assess measurement reproducibility and reduce potential observer-related variability, the two implant specialists reviewed the standardized measurement protocol together before independent remeasurement to ensure consistency in image registration and deviation assessment. Subsequently, they independently remeasured each deviation parameter using the same standardized protocol. Interobserver agreement for each deviation parameter was evaluated using the intraclass correlation coefficient (ICC) and 95% confidence intervals.

After image registration and landmark confirmation, the software automatically extracted the coordinates of the planned and placed implants and calculated the corresponding positional deviations (Fig. 4).

Fig. 4.

Accuracy analysis of implant placement. A Schematic illustration of the designed and actual implant positions, including ① platform deviation, ② apex deviation, ③ angular deviation, ④ platform horizontal deviation, ⑤ apex horizontal deviation, ⑥ platform vertical deviation, ⑦ apex vertical deviation. B-D Postoperative implant positions were registered with the corresponding preoperative implant plans in DentalNavi software. E Three-dimensional registration model used for accuracy analysis

The primary outcomes included global deviations at the implant platform and the apex, angular deviation, and horizontal and vertical deviations at both the platform and the apex. Global deviation at the platform and apex (mm) was defined as the three-dimensional displacement between the designed and actual implant positions at the center of the implant platform and the implant apex, respectively. Angular deviation (degrees) was defined as the angle between the central axes of the designed and actual implants. Vertical deviation (mm) was defined as the difference in implant depth along the long axis of the implant. Horizontal deviation (mm) was defined as the two-dimensional deviation in the mesiodistal and buccolingual directions, irrespective of depth deviation.

Statistical analysis

Statistical analyses were performed using IBM SPSS Statistics, version 26.0 (IBM Corp., Armonk, NY, USA). The normality of data distribution was assessed using the Shapiro–Wilk test. Data were presented as the mean ± standard deviation (SD) for normally distributed variables and as the median (interquartile range) for non-normally distributed variables. Comparisons were performed between the anterior and posterior regions and between the maxilla and mandible. When data were normally distributed with homogeneity of variance, comparisons between groups were performed using the independent-samples t-test. When the assumption of equal variances was violated, Welch’s t-test was applied. For non-normally distributed data, the Mann–Whitney U test was used for group comparisons. All reported p-values were two-tailed, and a significance level of 0.05 was adopted. Given the potential within-patient clustering of multiple implants, implant-level subgroup comparisons were considered exploratory.

Results

Study population and surgical outcomes

Six edentulous patients underwent R-CAIS, including two females and four males, with a mean age of 55 ± 8.15 years. A total of 42 implants were placed. Patient characteristics and implant distribution are shown in Table 1. All procedures were completed according to the preoperative design using robotic assistance, and all implants achieved an insertion torque greater than 30 N·cm. No severe intraoperative or postoperative complications were observed, including excessive bleeding and injury to adjacent anatomical structures. In addition, no early implant failure, postoperative bleeding, wound dehiscence, or infection was reported. At the 3-month follow-up after the last patient’s surgery, all implants remained in place, corresponding to a short-term implant survival rate of 100%. Follow-up is ongoing to assess long-term implant success and peri-implant tissue stability.

Table 1.

Patient Characteristics and Distribution of Implant Locations (n)

Characteristic
Age, years(Mean (SD))	55(8.15)
Gender
Male	4
Female	2
Jaw
Maxilla	20
Mandible	22
Site
Anterior region	18
Molar region	24

Implant positional accuracy

The inter-observer ICC values for all deviation parameters ranged from 0.876 to 0.959, indicating good to excellent measurement reproducibility. Specifically, platform buccolingual deviation showed good inter-observer agreement (ICC = 0.876; 95% CI: 0.782–0.931), whereas all other parameters demonstrated excellent agreement, with ICC values greater than 0.90.

Descriptive statistics of overall implant deviations

The mean deviation at the implant platform was 0.519 ± 0.216 mm (range, 0.141–0.961 mm; median, 0.459 mm), and the mean deviation at the implant apex was 0.508 ± 0.224 mm (range, 0.103–1.055 mm; median, 0.432 mm). The mean angular deviation was 0.634 ± 0.402° (range, 0.121–1.867°; median, 0.556°). The mean vertical deviations at both the implant platform and the apex were 0.298 ± 0.248 mm. Detailed results are presented in Table 2.

Table 2.

Descriptive statistics of implant deviation parameters (n = 42)

	Mean	Standard deviation	Minimum	Maximum	Median
Global deviations at platform (mm)	0.519	0.216	0.141	0.961	0.459
Global deviations at apex (mm)	0.508	0.224	0.103	1.055	0.432
Global angular deviations(°)	0.634	0.402	0.121	1.867	0.556
Platform horizontal deviations (buccolingual, mm)	0.247	0.152	0.000	0.665	0.233
Platform horizontal deviations (mesiodistal, mm)	0.225	0.184	0.003	0.796	0.203
Apex horizontal deviations (buccolingual, mm)	0.245	0.153	0.008	0.689	0.237
Apex horizontal deviations (mesiodistal, mm)	0.220	0.165	0.013	0.744	0.189
Platform vertical deviations(apicocoronal, mm)	0.298	0.248	0.001	0.938	0.280
Apex vertical deviations(apicocoronal, mm)	0.298	0.248	0.000	0.942	0.279

Comparison of implant accuracy between anterior and posterior regions

A total of 18 implants were placed in the anterior region and 24 in the posterior region. The results showed that global deviations at the apex, global angular deviations, and platform vertical deviations tended to be lower in the anterior region than in the posterior region. However, none of these differences reached statistical significance (all p -values > 0.05; Table 3).

Table 3.

Comparison of implant deviations between anterior and posterior regions

	Anterior (n = 18)	Posterior (n = 24)	t/Z	p value
Global deviations at platform (mm)	0.459(0.401,0.670)	0.449(0.348,0.707)	-0.470	0.638
Global deviations at apex (mm)	0.432(0.396,0.602)	0.443(0.337,0.705)	-0.419	0.675
Global angular deviations(°)	0.495(0.340,0.611)	0.631(0.326,0.767)	0.941	0.347
Platform horizontal deviations (buccolingual, mm)	0.263 ± 0.189	0.235 ± 0.120	0.572	0.571
Platform horizontal deviations (mesiodistal, mm)	0.210(0.087,0.420)	0.195(0.063,0.261)	0.712	0.477
Apex horizontal deviations (buccolingual, mm)	0.256(0.098,0.389)	0.234(0.100,0.310)	-0.623	0.533
Apex horizontal deviations (mesiodistal, mm)	0.202(0.120,0.366)	0.175(0.079,0.259)	-1.309	0.191
Platform vertical deviations(apicocoronal, mm)	0.206(0.122,0.349)	0.316(0.038,0.478)	-0.623	0.533
Apex vertical deviations(apicocoronal, mm)	0.205(0.121,0.379)	0.317(0.038,0.501)	-0.585	0.559

Comparison of implant accuracy between the maxilla and mandible

A total of 20 implants were placed in the maxilla and 22 were placed in the mandible. In this exploratory analysis, maxillary implants showed smaller deviations than mandibular implants. Among all measured parameters, only the buccolingual deviation at the apex differed significantly between the jaws, with a smaller deviation observed in the maxilla (p = 0.021). No statistically significant differences were found for the remaining parameters (all p > 0.05) (Table 4).

Table 4.

Comparison of implant deviations between the maxillary and mandibular implants

	Maxilla(n = 20)	Mandible(n = 22)	t/Z	p value
Global deviations at platform (mm)	0.455 ± 0.182	0.577 ± 0.232	-1.884	0.067
Global deviations at apex (mm)	0.419 (0.300,0.521)	0.513 (0.385,0.842)	-1.776	0.076
Global angular deviations(°)	0.533 (0.357,0.787)	0.578 (0.316,0.822)	-0.403	0.687
Platform horizontal deviations (buccolingual, mm)	0.185 (0.106,0.288)	0.284 (0.203,0.345)	-1.776	0.076
Platform horizontal deviations (mesiodistal, mm)	0.224 ± 0.191	0.225 ± 182	-0.007	0.995
Apex horizontal deviations (buccolingual, mm)	0.188 ± 0.136	0.296 ± 0.153	-2.410	0.021
Apex horizontal deviations (mesiodistal, mm)	0.172 (0.088,0.308)	0.198 (0.099,0.315)	-0.705	0.481
Platform vertical deviations(apicocoronal, mm)	0.227 ± 0.180	0.361 ± 0.181	-1.779	0.080
Apex vertical deviations(apicocoronal, mm)	0.227 ± 0.180	0.362 ± 0.286	-1.802	0.079

Analysis of deviation directions in buccolingual, mesiodistal, and vertical dimensions

Figure 5 illustrates the distribution and directions of implant deviations at the platform and apex in the vertical, buccolingual, and mesiodistal directions. Among all 42 implants, buccal deviation accounted for approximately 76.2%, whereas lingual deviation accounted for 23.8%. In the mesiodistal dimension, mesial and distal deviations accounted for approximately 47.6% and 52.4%, respectively. In the vertical dimension, 61.9% of implants were placed slightly shallower than designed, while 38.1% were placed deeper. Overall, implants tended to deviate toward the buccal and distal directions and were positioned slightly more superficially than designed.

Fig. 5.

The 3D scatter plot shows the deviations at the implant platform (A) and apex (B) in buccolingual, mesiodistal, and apicocoronal directions. Red dots represent the actual coordinates, and gray dots represent the projected points

Discussion

Following tooth loss, the mandible gradually atrophies due to the long-term absence of functional loading, rendering it increasingly susceptible to spontaneous fractures. In an edentulous mandible, where the residual alveolar and basal bone is weak and the residual bone height is less than 10 mm, inappropriate implant positioning or angulation may further compromise the structural integrity of the mandible, thereby increasing the risk of iatrogenic mandibular fractures. When fractures occur in edentulous jaws, the lack of dental support, reduced bone density, and insufficient bone height often make open reduction and internal fixation necessary for fracture treatment. This not only increases surgical trauma but also imposes substantial physical and psychological burdens on patients [22, 29]. Accurate three-dimensional implant positioning ensures optimal alignment of the implant with the surrounding anatomical structures, thereby facilitating accurate occlusal reconstruction, minimizing the risk of complications such as iatrogenic fractures, adjacent tooth injury, nerve damage, and maxillary sinus floor perforation, and ultimately improving patient satisfaction [30]. Therefore, precise implant placement is particularly important in complex clinical situations, such as implant rehabilitation in edentulous patients.

Dental implant placement is highly dependent on the clinician’s skill and experience in freehand implant surgery. In edentulous patients, the absence of natural teeth and the limited availability of residual alveolar bone may increase the difficulty of placing implants in the ideal position. As a result, implant positioning accuracy and treatment outcomes may vary even among experienced clinicians when faced with different clinical scenarios. A growing body of clinical evidence has confirmed that, in edentulous patients, CAIS can provide significantly greater implant placement accuracy [31–34]. A prospective study demonstrated that the mean angular deviation was 10.09° in the freehand group of edentulous patients, whereas it was reduced to 4.98° and 5.75° in the S-CAIS and D-CAIS groups, respectively. Furthermore, the three-dimensional deviations at both the implant platform and the apex were approximately 3.40–3.60 mm in the freehand group, compared with 1.40–1.86 mm in the other groups, demonstrating significantly greater positional accuracy in the CAIS groups [35]. A study using a double-factor technique that combined S-CAIS and D-CAIS achieved mean linear deviations of 1.42 mm at the implant platform and 1.25 mm at the implant apex, indicating a high level of placement accuracy with this combined guidance strategy in edentulous patients [36].

However, due to the complex anatomical conditions in edentulous patients, CAIS may be insufficient to fully satisfy the clinical demands for accurate implant placement. In recent years, R-CAIS has gradually been applied in the treatment of completely edentulous patients. The YaKebot dental implant robot employs a robotic arm end-effector equipped with an implant handpiece, which allows the robotic arm end-effector to and exits the oral cavity, precisely navigates to the designed surgical site, and automatically performs osteotomy preparation and implant placement. After completing the designed procedure, the system stops drilling and returns the robotic arm to its initial position. Throughout the procedure, the surgeon controls the system via a foot pedal, thereby reducing the influence of operator-dependent factors such as operator experience and variability [37, 38]. In addition, the system provides preoperative instrument path planning and postoperative accuracy evaluation, offering important technical support for the precision and standardization of implant surgery.

In this study, R-CAIS performed using the YaKebot system achieved deviations within 0.6 mm at the implant platform and apex. The mean angular deviation was approximately 0.6°, and the mean vertical deviation was within 0.3 mm. These results are comparable to those of Xie et al. [39], who evaluated the accuracy of R-CAIS for dental implant placement in edentulous patients and reported deviations of less than 0.6 mm at both the implant platform and apex, with a mean angular deviation of 1.83°. Similarly, another study using haptic robotic guidance in a completely edentulous participant reported a mean global angular deviation of 2.56°, mean deviations of 1.04 mm at the implant platform and 0.95 mm at the apex, and a signed depth deviation of 0.42 mm [20]. Collectively, these findings suggest that R-CAIS can achieve high accuracy even in complex cases of edentulous implant rehabilitation and may reduce the influence of patient- and operator-related factors on implant placement accuracy.

Analysis of the relationship between implant position and deviation revealed that the buccolingual deviation at the apex was greater in mandibular implants. This may be attributed to the heterogeneous bone quality of the mandible and the greater buccolingual density differences near the apical region [40]. Although the robotic arm can constrain deviation through its force-feedback control system, it still tends to follow the path of least resistance. In addition, mandibular movement is typically greater and more frequent during surgery. This phenomenon is also observed in freehand implant surgery and CAIS. However, the robotic system can still improve implant accuracy through faster and more precise responses. These findings suggest that clinicians should reserve sufficient buccolingual space when performing robotic computer-assisted implant surgery in the edentulous mandible. This helps to compensate for potential intraoperative deviations and ensure adequate surrounding bone for implant support. From a technological perspective, such deviation patterns could be incorporated into the robotic system through preoperative programming and algorithm optimization.

In terms of deviation directions, implants tended to shift toward the buccal and distal directions and were positioned slightly more superficially than designed. Buccal deviation is considered to be associated with heterogeneous bone quality. Distal deviation may be influenced by the instrument entry and exit path as well as the patients’ mouth opening. In addition, factors such as instrument wear and resistance from apical bone during implant insertion may prevent implants from reaching the designed depth.

In D-CAIS, the preoperative virtual trajectory must be translated into the actual drilling path by clinician-operators using a handheld implant handpiece. This process requires a high level of hand-eye coordination, interpretation of real-time navigation feedback, and rapid operative response. Therefore, timely recognition of and adjustment to navigation prompts are essential for achieving high implant placement accuracy [41, 42]. By contrast, R-CAIS enables a more standardized and programmed translation of the preoperative design into the intraoperative execution path, thereby reducing the dependence of surgical outcomes on clinician-operators’ experience and real-time manual control. This may also help shorten the learning curve.

Although R-CAIS improves placement accuracy, deviations from the designed implant position still occur, meaning that the system does not completely eliminate surgical errors. Factors influencing the accuracy of R-CAIS in edentulous patients may also include image fusion, scanning accuracy, measurement accuracy, joint drive compliance, and real-time tracking performance [36, 43, 44]. Due to metal artifacts in CBCT imaging, postoperative DICOM data may not accurately reflect the true implant position, potentially leading to measurement errors in deviation analysis [45]. In edentulous patients, the fabrication of a cortical bone screw guide and a patient-specific calibration guide is required, which increases the number of patient visits and prolongs the overall treatment time. Moreover, inaccuracies may arise in the fabrication, positioning, and fixation of surgical guides. Thus, R-CAIS may be more valuable in complex cases, whereas freehand implant surgery remains a reasonable option for simple single-implant cases with adequate bone volume.

In addition to numerical deviation, the relationship between the implant position and the surrounding bone envelope should be considered. A prosthetically driven implant position may occasionally be close to the buccolingual cortical contour, particularly in edentulous patients with limited residual alveolar bone volume. Moreover, the apparent buccolingual position of an implant on reconstructed images may be influenced by cross-sectional orientation, three-dimensional rendering, CBCT artifacts, and image-registration characteristics. Therefore, software-based visualization should be interpreted together with multiplanar CBCT assessment and clinical findings. Clinically, limited buccolingual bone coverage may affect long-term peri-implant hard-tissue and soft-tissue stability, even when the measured implant deviation is small. Such cases require careful long-term clinical and radiographic follow-up to assess marginal bone stability, soft-tissue recession, and peri-implant inflammatory changes.

The study still has several limitations. Although a standardized protocol was employed to minimize errors associated with CBCT image quality, image registration, and measurement procedures, the influence of these factors could not be completely eliminated and may have affected the reported accuracy values. This was a retrospective study with a relatively small sample size. Because multiple implants were placed within the same patient, implant-level observations cannot be considered fully statistically independent. Given the limited number of patients included in this study, the potential clustering effect at the patient level could not be robustly accounted for in the statistical analysis. Accordingly, implant-level subgroup analyses should be interpreted as exploratory.

Although all implants remained in place at the 3-month follow-up, this finding reflects only short-term survival and should not be interpreted as evidence of long-term implant success or peri-implant tissue stability. Further follow-up is required to evaluate long-term clinical outcomes. In the future, we plan to conduct prospective multicenter studies with larger sample sizes and longer follow-up periods to further evaluate not only the accuracy of fully autonomous robotic implant surgery in edentulous patients but also its clinical efficacy, implant survival, and long-term peri-implant tissue stability.

Conclusions

Within the limitations of this study, R-CAIS demonstrated high implant placement accuracy in edentulous patients. The system was able to autonomously perform implant osteotomy preparation and implant placement according to the preoperative design, indicating its potential to enhance the precision and predictability of implant rehabilitation in edentulous patients.

Authors’ contributions

Jing Lv and Honghong Liu jointly conceived and designed the study. Data collection was performed by Jing Lv. Data analysis was conducted by Yingying Zhang, Jia Chen, and Qian Han. Jing Lv drafted the manuscript. Honghong Liu and Zhihong Zhang critically revised the manuscript for important intellectual content. All authors reviewed and approved the final version of the manuscript.

Funding

This study was supported by the Anhui Provincial Key Research and Development Program (Social Development Project, No. 2023s07020006).

Data availability

The original contributions presented in this study are included in the article. Further data and materials are available from the corresponding author upon request.

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of the First Affiliated Hospital of the University of Science and Technology of China. This study was conducted in accordance with the principles of the Declaration of Helsinki. Written informed consent was obtained from all participating patients.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.