Comparison of early clinical outcomes between handheld milling-based Smith & nephew CORI robotic and cutting-guide ARTHROBOT knee systems in total knee arthroplasty

Abstract

A comparative analysis was conducted to assess the early clinical results following total knee arthroplasty. (TKA) assisted by the handheld robotic milling CORI system (CRATKA group) versus the cutting-guide ARTHROBOT system (JRATKA group). This study included 103 patients suffering from end-stage knee osteoarthritis in a retrospective analysis. The JRATKA group demonstrated significantly shorter registration time, osteotomy time, and total operative time, alongside a steeper learning curve (13 cases vs. 17 cases). Radiographically, the JRATKA group achieved a coronal femoral component angle significantly closer to 90° and a superior sagittal femoral component angle. No significant intergroup differences were noted in hospital stay, haemoglobin loss, inflammatory markers, 1-month post-operative functional scores (VAS, KSS, WOMAC, ROM), complication rates, or patient satisfaction. Both systems facilitated precise implantation and excellent early functional recovery. The JRATKA system showed advantages in procedural efficiency, whereas the CORI system offers distinct flexibility. The choice should be individualised based on institutional resources and surgical team expertise.

Introduction

Total knee arthroplasty (TKA) is a well-established surgical procedure aimed at restoring lower limb alignment and achieving optimal soft-tissue balance, which are critical for long-term implant survivorship and favorable functional outcomes in end-stage knee osteoarthritis (KOA) [1, 2]. Conventional TKA relies on mechanical alignment instruments and surgeon experience, often resulting in a relatively high rate of alignment deviations, which remains a significant factor affecting postoperative outcomes and long-term implant survivorship. Resently, the integration of robotic-assisted technology in surgery has brought revolutionary advancements in enhancing the precision and predictability of TKA. This technology significantly improves the accuracy of bone resection and prosthesis placement through preoperative three-dimensional planning, intraoperative real-time navigation, and precise execution [3–5]. Currently, various surgical robot systems with distinct design philosophies have emerged in the market. Based on the characteristics of the robotic arm end-effector, they can be classified into two categories: milling-based systems and guide-based/oscilating saw systems. The Smith & Nephew CORI and ARTHROBOT Knee Systems are representative of milling-based and guide-based types, respectively. Both systems allow the use of a “four-in-one” cutting guide after distal femoral resection.While each system has unique features in preoperative planning, registration, soft tissue balancing, and bone resection, there is currently a lack of clinical reports comparing their efficacy in practical application. Therefore, this study aims to conduct a retrospective comparative analysis to systematically evaluate the differences between milling-based Smith & Nephew CORI robotic-assisted TKA (CRATKA group) and guide-based ARTHROBOT Knee Systems-assisted TKA (JRATKA group) in terms of perioperative outcomes, learning curves, radiographic assessments, and follow-up data, thereby providing objective and detailed evidence-based medical guidance for clinical practice.

Methods

Patient selection

Inclusion criteria: ① Met the diagnostic criteria for severe knee osteoarthritis established by the American College of Rheumatology; ② Inadequate response to standardized conservative treatment; ③ Varus knee deformity ranging from 0° to 20°; ④ Primary unilateral knee surgery performed by a single surgical team, all using the same knee prosthesis model. Exclusion criteria: ① Severe cardiopulmonary dysfunction or hepatic/renal impairment rendering the patient unfit for surgery; ② Presence of either local knee infection or active systemic infection; ③ Patients with neuromuscular dysfunction such as paralysis or muscle weakness.

General data

Ethical approval for this study was obtained from the Scientific Research Ethics Committee of Panzhihua Central Hospital (Approval No.: PankeLunShenZi 2024-008). Following the acquisition of written informed consent from all participants, clinical data were retrospectively analyzed for patients who received robot-assisted TKA for end-stage knee osteoarthritis between February 2025 and September 2025. A total of 103 patients who satisfied all eligibility criteria were enrolled. Based on the robotic system used, patients were divided into the Smith & Nephew CORI robot group (CRATKA group) and the ARTHROBOT Knee Systems group (JRATKA group). The two groups demonstrated comparable baseline demographic and clinical characteristics, including age, gender, operative side, BMI, and disease duration. (P > 0.05), as shown in Table 1.

Table 1.

Comparison of baseline demographic and clinical characteristics between the two groups before treatment

Metrics	CRATKA group (n = 51)	JRATKA group (n = 52)	t/χ2	P
Age (years, x̅ ± s)	64.1 ± 8.5	62.5 ± 8.2	0.972	0.333
Gender (n, male/female)	20/31	18/34	0.234	0.686
Operative Side (R/L)	28/23	25/27	0.480	0.556
BMI (kg/m², x̅ ± s)	25.3 ± 2.9	25.7 ± 2.0	-0.816	0.416
Disease Duration (months, x̅±s)	95.6 ± 54.6	101.3 ± 50.2	-0.552	0.582

Surgical technique

All surgical procedures were performed by a consistent team of senior attending surgeons. All operations were conducted under combined intravenous-inhalational anaesthesia with patients in the supine position. Tourniquet application was standardized across all total knee arthroplasty procedures in this series. A standardized approach, involving a midline skin incision and medial parapatellar arthrotomy, was employed for all total knee arthroplasty procedures. Preoperative planning for all patients specified the use of a 6 mm tibial insert and a 3° posterior tibial slope angle. The planned postoperative lower limb alignment target was neutral mechanical alignment. Patellar resurfacing was not performed in any case, and no surgical drains were employed.Perioperative management for both groups followed the guidelines outlined in the Chinese Expert Consensus on Accelerated Rehabilitation Strategies for Hip and Knee Arthroplasty—Perioperative Management [6]. This included standardized protocols for anti-inflammatory therapy, analgesia, anticoagulation, oedema reduction, and functional exercise.

CRATKA group

The Smith & Nephew CORI Orthopaedic Intelligent Robot system was employed. Following joint exposure via the aforementioned approach, 3D infrared camera tracking arrays were affixed to the medial aspect of the distal femur and the medial proximal tibia. The milling handpiece was mounted and registered to the handheld flexible robotic arm. Registration was sequentially performed as guided by the robotic system. The center of the femoral head was determined by dynamic hip circumduction, assessing lower limb alignment and range of flexion-extension. Femoral and tibial bone registration was performed following system guidance, and soft tissue tension was tested. Patient-specific kinematic data were acquired to formulate a surgical plan.Under robotic navigation monitoring, distal femoral resection was executed and verified. A “four-in-one” cutting guide was then positioned to complete subsequent femoral cuts. The guide’s position was adjusted for tibial resection, with continuous monitoring of resection depth and alignment throughout. Tibial plateau resection was subsequently performed and validated.A trial reduction with knee prosthesis components was conducted, and medial-lateral balance as well as flexion-extension gaps were re-evaluated. Upon achieving satisfactory alignment and stability. The joint cavity was irrigated using pulsed lavage, and the Smith & Nephew Legion knee prosthesis was implanted with the appropriate polyethylene insert. The wound was closed in layers and dressed.

JRATKA group

Based on preoperative CT scans of the hip, knee, and ankle, a patient-specific plan was generated using the ARTHROBOT Knee Systems. This plan then guided the subsequent surgical procedure, which was initiated via a standard midline skin incision and medial parapatellar approach to expose the knee joint. Osteophytes were excised. Robotic positioning reflective markers were securely fixed to the distal femur and the tibial plateau. Registration and calibration were then performed according to system prompts, targeting anatomical regions including the distal femur, the anterior cortex of the distal femur, and the tibial plateau.Following successful registration, femoral distal resection was conducted under the robotic, with subsequent verification of the cut. A “four-in-one” guide was then positioned to complete the remaining femoral resections. The guide’s position was adjusted for tibial resection, with continuous intraoperative monitoring of resection depth and alignment.After completion of all bone cuts, trial reduction with provisional components was performed to assess the balance of the flexion and extension gaps, as well as the medial and lateral stability. Upon confirmation of satisfactory alignment and stability, the joint cavity was irrigated using pulsed lavage. The Smith & Nephew Legion knee prosthesis was implanted with the appropriate polyethylene insert. The wound was closed in layers and dressed.

Efficacy evaluation metrics

Perioperative parameters were recorded, including registration time, osteotomy time, total operative time, accuracy of single-cut alignment for distal femur and tibia, insert thickness, incision length, wound healing status, time to ambulation, and length of hospital stay.Postoperative day 3 laboratory values were collected to assess blood loss and inflammatory response: haemoglobin (HB), haematocrit (HCT), white blood cell count (WBC), neutrophil-to-lymphocyte ratio (NLR), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP).Functional outcomes were evaluated using time to full weight-bearing, and pre- and one-month postoperative scores including the Visual Analogue Scale (VAS) for pain, American Knee Society Score (KSS), Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC), and knee range of motion (ROM). Complication rates and patient satisfaction were documented. Radiographic assessments included: Coronal tibial component angle (FTC, optimal value 90°); Sagittal femoral component angle (LFC, optimal value 0°); Sagittal tibial component angle (LTC, optimal value 87°);Coronal femoral component angle (FFC, optimal value 90°); Hip-knee-ankle angle (HKA, optimal value 180°) deviation to assess lower limb alignment. All evaluations were conducted under blinded conditions to eliminate assessment bias.

Statistical methods

All analyses were performed in SPSS 26.0. Continuous data are summarized as mean ± SD. Based on distribution, intergroup comparisons used independent t-tests (normal) or Mann-Whitney U tests (non-normal). Longitudinal intragroup comparisons employed one-way ANOVA with LSD post-hoc tests. Categorical data were analyzed by Chi-square/Fisher’s exact test, with P < 0.05 (two-tailed) considered significant. A post hoc power analysis was conducted for the primary efficiency metric of total operative time. With an alpha of 0.05, and sample sizes of 51 and 52, the achieved statistical power was calculated to be > 99%, indicating a sufficiently powered analysis for this outcome.

Results

Perioperative data

All patients successfully completed the surgery. The CRATKA group exhibited significantly longer registration time, osteotomy time, and total operative time compared to the JRATKA group (P < 0.05). The two groups demonstrated comparable accuracy in distal femoral and tibial single-cut alignment (P > 0.05). In the CRATKA and JRATKA groups, 7 and 3 cases, respectively, required secondary osteotomy due to unsatisfactory alignment after the initial cut, with all achieving acceptable alignment after revision.The two groups were comparable in terms of insert thickness, incision length, time to ambulation, length of hospital stay, or postoperative day 3 laboratory values (Hb, HCT, WBC, NLR, ESR, CRP) (P > 0.05). All surgical incisions achieved primary healing (Table 2).

Table 2.

Comparison of perioperative parameters and clinical findings at postoperative day 3 between the two groups

Metrics	CRATKA group(n = 51)	JRATKA group(n = 52)	t/χ2/Fisher	P
Registration time (min, x̅ ± s)	6.6 ± 2.0	4.3 ± 1.8	6.137	<0.001
Osteotomy time (min, x̅ ± s)	10.4 ± 2.2	5.7 ± 1.8	11.877	<0.001
Operative time (min, x̅ ± s)	92.6 ± 7.9	75.6 ± 10.1	9.502	<0.001
Accuracy rate of single-cut alignment for distal femur and tibia	94.12% (48/51)	86.54% (45/52)		0.319
Insert thickness (n, 6/7/8/9/10 mm)	44/7/0/0/0	47/5/0/0	0.423	0.555
Incision length (cm, x̅ ± s)	14.1 ± 0.6	13.9 ± 0.7	1.556	0.123
Wound healing (n, Grade A/B/C)	51/00/00	52/00/00		ns
Time to ambulation (days, x̅ ± s)	0.7 ± 0.2	0.7 ± 0.1	0.000	1.000
Length of hospital stay (days, x̅ ± s)	9.4 ± 1.8	9.5 ± 1.3	-0.324	0.747
Hb (g/L, x̅±s)	115.6 ± 12.5	118.5 ± 10.5	-1.276	0.205
HCT (%, x̅±s)	37.7 ± 3.1	36.8 ± 3.9	1.295	0.198
WBC (×109, x̅±s)	7.5 ± 1.5	7.9 ± 1.9	-1.184	0.239
NLR (x̅±s)	3.5 ± 1.2	3.6 ± 1.8	-0.331	0.741
ESR (mm/h, x̅±s)	58.7 ± 20.2	60.2 ± 18.6	-0.392	0.696
CRP (mg/L, x̅±s)	52.5 ± 20.5	56.0 ± 21.0	-0.856	0.394

Learning curve

The learning curve was assessed by cumulative summation (CUSUM) analysis, with the transition point between the learning and proficiency phases defined by its peak. In the CRATKA group, the cubic curve demonstrated the highest goodness of fit (P < 0.05, coefficient of determination R² = 0.988). The fitted equation was CUSUM(n) = CUSUM(n) = 0.003n³-0.285n² + 7.495n + 24.812 (where n represents the number of surgeries), indicating a learning phase of 17 cases (33.33%).In the JRATKA group, the cubic curve also showed the highest goodness of fit (P < 0.05, R² = 0.986). The fitted equation was CUSUM(n) = 0.004n³-0.331n² + 7.034n + 38.620, the learning phase comprised 13 cases.(25%). The duration of the learning phase was comparable between groups (P > 0.05) ( Fig. 1).

Fig. 1.

Comparison of learning curves between the two groups

Radiographic evaluation

At the 1-month follow-up, all knee prostheses in both groups were well-positioned and stable, with no signs of loosening. Furthermore, the two cohorts exhibited comparable alignment outcomes in the HKA, FTC, and LTC angles (P > 0.05). However, The FFC and LFC angles differed significantly between the two groups., the JRATKA group demonstrated values significantly closer to the ideal targets of 90° and 0°, respectively, compared to the CRATKA group (P < 0.05) (Table 3).

Table 3.

Comparison of imaging documents between the two groups

Metrics	Parameter	CRATKA group (n = 51)	JRATKA group (n = 52)	t	P
FFC	Angle (°, x̅±s)	88.0 ± 0.9	89.4 ± 1.1	-1.836	<0.001
FTC	Angle (°, x̅±s)	88.7 ± 1.0	89.1 ± 1.2	-1.836	0.069
LFC	Angle (°, x̅±s)	7.2 ± 1.3	6.5 ± 1.6	2.434	0.017
LTC	Angle (°, x̅±s)	86.3 ± 1.2	86.4 ± 1.5	-0.373	0.710
HKA	Angle (°, x̅±s)	177.1 ± 1.0	177.4 ± 1.7	-1.089	0.279

Follow-up results

Both groups were followed up postoperatively. Both preoperatively and in the early functional recovery period, the two groups demonstrated no significant differences, as evidenced by comparable scores in VAS, KSS, WOMAC, knee ROM, and the time to full weight-bearing (P > 0.05), both demonstrated significant improvement in all metrics at the 1-month follow-up compared to their baseline values (P < 0.05). Each group reported three cases of intermuscular venous thrombosis in the lower limbs. Postoperative wound exudation occurred in one patient from the CORI group but resolved conservatively with dressing changes (P > 0.05). No major intraoperative or postoperative complications occurred, such as mechanical failure, pulmonary embolism, neurovascular injury, periprosthetic joint infection, nausea, vomiting, or reflux aspiration. Patient satisfaction during hospitalization was comparable between the two groups, as detailed in Table 4.

Table 4.

Comparison of follow-up documents between the two groups

Metrics	CRATKA group(n = 51)	JRATKA group(n = 52)	t/χ2/Fisher	P
Time to Full Weight-Bearing (days)	25.5 ± 4.5	26.6 ± 3.6	-1.371	0.173
VAS Score (points, x̅ ± s)
Preoperative	4.6 ± 1.1	4.7 ± 1.6	-0.369	0.713
1 Month Postoperative	1.7 ± 0.7	1.9 ± 0.6	-1.558	0.122
P	<0.001	<0.001
KSS Clinical Score (points, x̅ ± s)
Preoperative	43.5 ± 5.6	45.6 ± 7.2	-1.650	0.102
1 Month Postoperative	82.3 ± 5.9	83.6 ± 6.6	-1.053	0.295
P	<0.001	<0.001
KSS Clinical Score (points, x̅ ± s)
Preoperative	42.4 ± 4.7	43.5 ± 6.2	-1.013	0.313
1 Month Postoperative	80.6 ± 5.9	79.5 ± 5.6	0.971	0.334
P	<0.001	<0.001
WOMAC Score (points, x̅ ± s)
Preoperative	74.7 ± 5.9	73.6 ± 6.5	0.899	0.371
1 Month Postoperative	59.6 ± 4.9	60.9 ± 5.6	-1.253	0.213
P	<0.001	<0.001
Knee Flexion-Extension ROM (°, x̅ ± s)
Preoperative	102.7 ± 10.5	101.7 ± 9.5	0.507	0.613
1 Month Postoperative	115.5 ± 9.5	112.5 ± 7.6	1.771	0.080
P	<0.001	<0.001
Complication Rate [n (%)]	4(7.84)	3(5.77)		0.715
Satisfaction Rate During Hospitalization (%)	49/51(96.08)	49/52(94.23)		1.000

Representative cases

See Figs. 2 and 3.

Fig. 2.

Case 1. Imaging findings of a 61-year-old female with right knee osteoarthritis (Kellgren-Lawrence IV) treated with robot-assisted TKA (ARTHROBOT Knee system). (A) Preoperative full-length lower limb radiographs views. (B) Intraoperative photograph showing distal femoral osteotomy performed with the robotic system. (C1-C3) Postoperative knee anteroposterior, lateral, and full-length lower limb radiographs

Fig. 3.

Case 2. Imaging findings of a 61-year-old female with right knee osteoarthritis (Kellgren-Lawrence IV) treated with robot-assisted TKA (CORI system). (A) Preoperative full-length lower limb radiographs views. (B)The distal femoral osteotomy was conducted using the CORI robotic. (C1-C3) Postoperative knee anteroposterior, lateral, and full-length lower limb radiographs

Discussion

Driven by an aging population and rising life expectancy, osteoarthritis (OA) affects over 300 million individuals globally. Notably, the prevalence of OA among Chinese adults aged ≥ 40 years reaches 46.3%, underscoring its substantial national burden [7–11]. The volume of TKA procedures is projected to increase steadily. Consequently, with the progress in precision medicine and surgical technology, robot-assisted TKA has transitioned from an innovative concept to a routinely applied clinical modality [12, 13]. This study retrospectively analyzed the short-term outcomes of guide-based ARTHROBOT Knee Systems-assisted TKA versus semi-active collaborative milling-based Smith & Nephew CORI robotic-assisted TKA.The study demonstrated that in terms of surgical efficiency, the ARTHROBOT Knee Systems-assisted procedures were associated with shorter operative times, primarily attributable to differences in registration and osteotomy duration.The ARTHROBOT Knee Systems exhibited shorter registration time, which relies on preoperative CT-based modeling and planning—intraoperatively, only key planes require validation. In contrast, the CORI system utilizes intraoperative 3D registration technology, which demands more comprehensive data acquisition across multiple planes, resulting in relatively longer registration times.Regarding osteotomy, both systems required the surgeon to perform only the distal femoral resection initially; the remaining femoral cuts were completed using a “four-in-one” cutting guide. During distal femoral and tibial resections, the ARTHROBOT Knee Systems employed an oscillating saw guided by a restricted cutting guide, leading to higher osteotomy efficiency.This workflow seamlessly aligns with the conventional TKA technique of using an oscillating saw, resulting in a shorter learning curve (13 cases in this study). This finding is consistent with Wang et al. [14], who reported a similar learning phase for the guide-based ROSA (Zimmer Biomet, USA) knee system in TKA.In contrast, the CORI system utilizes a high-speed milling burr operated by the surgeon for point-by-point milling and layered grinding. Due to the small volume of the burr and its limited grinding range, the process is relatively slow. In practice, we found that employing a “dragging” technique during milling could enhance efficiency. However, this technique requires the surgeon to adapt to the milling operation and its tactile feedback, resulting in a longer learning curve, which stabilized at 17 cases in our study.The compact milling burr minimizes soft tissue trauma during osteotomy and, in cases of complex anatomy, enables precise adjustments in resection parameters to preserve joint balance and create a smoother cut surface—factors that may mitigate prosthetic wear. Furthermore, its use reduces intraoperative vibration and the risk of dislodging optical trackers. Nevertheless, bone debris generated during milling—particularly when dealing with sclerotic bone—can scatter and potentially contaminate the surgical field if ejected. This may increase the risk of local infection. Strategies to mitigate bone debris scatter include the use of protective shields, controlled irrigation during milling, and active suction assistance by the surgical team.

In terms of controllability, the ARTHROBOT Knee Systems provides only a guide to restrict the osteotomy direction without setting a safety boundary, which places higher demands on the surgeon’s intraoperative experience in bone resection and awareness of protecting critical soft tissues such as ligaments. In contrast, the CORI system requires the surgeon to operate while monitoring the screen during milling; if the predefined safety margin is exceeded, the milling burr automatically retracts and stops rotating, reducing the surgeon’s active involvement in soft tissue protection. Regarding operational precision, since both groups utilized a “four-in-one” cutting guide after distal femoral resection, the surgeon’s experience and stability in using the oscillating saw with the guide significantly influenced the accuracy of the osteotomy. The ARTHROBOT Knee Systems, which employs an oscillating saw for both distal femoral and tibial plateau resections, is more susceptible to variations in cutting stability and operator experience. When dealing with patients exhibiting bone sclerosis, the elastic nature of the saw blade and vibration during oscillating saw use may lead to reduced resection volume in sclerotic areas, potentially resulting in unsatisfactory osteotomy angles and necessitating secondary resection. However, due to the high speed of oscillating saw osteotomy, overall efficiency remains largely unaffected.In contrast, the CORI system ensures greater precision in osteotomy through real-time dynamic tracking compensation, facilitating optimal prosthesis-bone interface fit. Both robotic systems demonstrated comparable capabilities in predicting osteotomy gaps and soft tissue balance, achieving resection targets while preserving bone stock. The authors believe that thorough removal of osteophytes affecting soft tissue tension prior to registration is critical for ensuring post-osteotomy soft tissue balance. For osteophytes that potentially impact soft tissue tension but cannot be fully removed pre-registration, slightly reducing resection on the corresponding side followed by post-osteophyte removal may be considered.Both systems required similar incision lengths and surgical exposure. With strict application of intraoperative tourniquet use and adherence to ERAS protocols, Postoperative hemoglobin loss and inflammatory response were comparable between the two groups. Previous studies indicate that robot-assisted TKA does not increase postoperative blood loss or inflammatory response compared to conventional TKA, a view supported by Yang Yongze [15] and Ma Qiaoqiao [16]. Luo Xuefeng [17] further suggests that robot-assisted TKA may offer advantages in controlling perioperative blood loss. Regarding registration, the ARTHROBOT Knee Systems eliminates the need for secondary verification pins, preventing potential retention of forgotten pins in the surgical site. The primary advantage of robot-assisted TKA lies in its ability to execute preoperative planning accurately, ensuring precise prosthesis placement and restoration of lower limb alignment [18–20], which is a crucial criterion for surgical success [21, 22]. Short-term follow-up confirmed well-positioned prostheses without signs of loosening in both groups.In this study, the ARTHROBOT Knee Systems achieved FFC and LFC values closer to ideal, attributable to preoperative CT-based planning that predefined the femoral notch point on the anterior cortex, combined with intraoperative real-time validation against this reference. This enhanced precision was consistently observed in our previous comparisons between Hehua and CORI systems [23].Similar findings were reported by Bhor [24] and Zhou [25], confirming that CT-based robotic systems improve prosthesis positioning accuracy and optimize coronal and sagittal alignment. However, preoperative CT imaging increases waiting time, hospitalization costs, and radiation exposure. While a formal cost-effectiveness analysis was beyond the scope of this study, institutions must weigh against the potential benefits in alignment precision observed in our results.

Both groups achieved satisfactory postoperative HKA angles, demonstrating the consistent ability of robotic-assisted systems to restore lower limb alignment.Both robotic systems demonstrated safe operational performance during TKA procedures, with no occurrences of mechanical failures, neurovascular injuries, or other complications. Postoperatively, no adverse events such as wound healing issues, infections, deep vein thrombosis, or pulmonary embolism were observed in either group. All patients achieved satisfactory knee function, significant pain reduction, and improved comfort levels. The average time to assisted ambulation with a walker was 9 h postoperatively, and full weight-bearing without crutches was achieved within 25–26 days. These outcomes suggest that preoperative CT dependency does not significantly impact early functional recovery or range of motion in robot-assisted TKA patients.In clinical practice, concerns regarding whether a robotic system is image-dependent may be unwarranted. Surgeons can select a robotic platform based on the specific advantages of each system and their institutional resources. In terms of soft tissue balancing, the CORI system utilizes dynamic monitoring of ligament balance and real-time assessment of soft tissue tension, placing greater emphasis on intraoperative soft tissue equilibrium. From a design perspective, the CORI robot employs a handheld milling device instead of a large robotic arm, resulting in a compact and portable system that minimizes operating room space requirements and offers flexible configuration options, making it suitable for hospitals of varying scales. Additionally, the system eliminates the need for sterile draping of robotic components and complex instrument assembly, saving time and reducing operational burden.

In summary, both robotic systems demonstrate excellent efficacy in achieving the core objectives of precise prosthesis placement and optimal restoration of knee function. The milling-based CORI system excels in operational flexibility and control refinement, while the guide-based ARTHROBOT Knee Systems demonstrates advantages in surgical efficiency and procedural ergonomics. Future clinical decision-making should be based on a comprehensive evaluation of hospital resources, surgeon expertise, and individualized patient needs. The mutual integration and cross-learning between these two technological approaches are expected to further advance the field of joint surgery toward higher levels of precision, intelligence, and personalization. It is also important to note that our findings are applicable to a specific patient population with varus deformities of mild to moderate severity. The performance and potential differential advantages of these two robotic systems in knees with severe deformity, significant bone defects, or poor bone quality remain an important area for future investigation.

It is important to acknowledge the limitations of this study. Firstly, its design as a single-centre, retrospective analysis may introduce selection bias, potentially limiting the generalisability of our findings. Secondly, the relatively small sample size and short follow-up period preclude a robust evaluation of long-term outcomes, which only allows for an assessment of early clinical and radiological outcomes. This short timeframe precludes any evaluation of mid- to long-term endpoints, such as implant survivorship, the maintenance of mechanical alignment, functional progression, or the potential impact of observed alignment differences on polyethylene wear. Additionally, our radiographic evaluation was limited to two-dimensional coronal and sagittal plane measurements on plain radiographs. We were unable to assess three-dimensional parameters, such as component rotation or joint line restoration. Future studies utilizing post-operative CT scans would provide a more comprehensive evaluation of implant positioning. Finally, all procedures were performed by a single surgical team, which may affect the reproducibility of the results in other clinical settings. Future research should involve multicentre, prospective randomised controlled trials with long-term follow-up to validate these findings.

Abbreviations

TXA

Tranexamic acid

VAS

Visual analogue scale

TKA

Total knee arthroplasty

HB

Hemoglobin

HCT

Hematocrit

WBC

White Blood Cell

ESR

Erythrocyte Sedimentation Rate

CRP

C-reactive Protein

FTC

Frontal tibia component

LFC

Lateral femoral component

LTC

Lateral tibia component

FFC

Frontal femoral component

HKA

Hip-knee-ankle angle

JRATKA

ARTHROBOT Knee Systems-assisted total knee arthroplasty

CRATKA

Stryker Cori robot-assisted total knee arthroplasty

KSS

American Knee Society Score

ROM

Range of motion

VAS

Visual analogue scale

LOS

Length of stay

Author contributions

MYW and ZDT conducted the research, collected data, performed literature review, and drafted the manuscript. LYP handled surgical planning and execution, and contributed to revision. HPW and YZH assisted in research, data collection, and literature review. XZS undertook data analysis, interpretation, and literature review. All authors endorsed the final manuscript.

Funding

This research was funded by Sichuan Provincial Rehabilitation Medical Association’s 2025 Annual Scientific Research Projects (No.SCKFKY20250210), Sichuan Provincial Primary Health Development Research Center in 2024, North Sichuan Medical College (No. SWFZ24-Q-86) and 2025 Annual Science and Technology Bureau of Panzhihua City project (2025ZD-S-2).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

This study received ethical approval from The Institutional Review Board of Panzhihua Central Hospital (Approval No.: PankeLunShenZi 2024-008). All participants provided written informed consent before enrollment.

Consent for publication

Written informed consent was obtained from the patient for publication of this paper.

Competing interests

The authors declare no competing interests.