Table 1.

Summary of artificial intelligence assisted joint surgery in China in 2021

Studies	System	Origination	Main findings
Li et al, 2019	Digital twin	publication [3]	The Tang's classification divided fractures into anterior, posterior, lateral, and medial types. The majority of fracture lines (85.9%, 433/504) on the anterior maps were along the intertrochanteric line where the iliofemoral ligament was attached.
Kong et al, 2020	5G + Mako	publication [8]	First report on the application of 5G communication technology in the field of joint surgery. The increase of Harris scores of the two patients after surgery were 71 and 62, respectively.
Lei et al, 2019	MR + 3D	publication [10]	The first study to use MR combined with 3D printing technology in THA. The distance between the preoperative design rotation center and the postoperative rotation center was 5.5 mm. The postoperative anteversion angle was 30.67° and inclination was 43.16°, compared to preoperative design (anteversion angle 25° and inclination 40°).
Wu et al, 2019	5G + MR	publication [12]	The first 5G + MR-assisted remote consultation and screw fixation of a thoracic fracture in July 2019. The average depth for C1 and C2 pedicle screws were (29.82 ± 1.36) and (32.24 ± 1.21) mm on the left side, and (30.38 ± 0.95) and (31.42 ± 1.05) mm on the right side.
Guo et al, 2007	V-Xp Stryker	publication [13]	The acetabular abduction angle of computer navigation-assisted THA was in the range of 30°-54°, with an average of (40.6 ± 5.1) °, and an acetabular abduction angle of > 50° was observed only in 1 case; in patients who underwent conventional THA, the acetabular abduction angle was in the range of 28°-70°, with an average of (44.2 ± 8.7) °, and an acetabular abduction angle of > 50° was observed in 10 cases.
Zhang et al, 2011	BrainLab	publication [14]	Nine knee implants (28%) in the conventional group, compared with no knee implants in the computer-navigation group, deviated > 3° from the mechanical axis on the coronal plane. The coefficient variation of data in the conventional group was three times greater than that in the computer-navigation group.
Liu et al, 2020	StealthStation S7 (Medtronic)	publication [17]	The hidden blood loss in the patients underwent computer navigation-assisted TKA was significantly lower than in those who underwent conventional TKA (448.7 ± 300.7 mL vs. 688.6 ± 405.8 mL by GROSS method and 374.3 ± 360.8 mL vs. 667.2 ± 425.5 mL by HB-balance method).
Zhou et al, 2021	Mako	publication [18]	The Mako robot-assisted THA accuracy rates for implanting the acetabular cup into the Lewinnek and Callanan safety zones were 94.9 and 74.6%, compared to conventional THA (79.7 and 50.8%).
Li et al, 2021	Mako	publication [19]	The postoperative LLD of Mako robot-assisted THA and conventional THA were 2.3 mm and 6.7 mm (P<0.001).
Chai et al, 2020	Mako	publication [20]	The proportion of cases with the acetabular rotation center in the safety zone in the Mako robot-assisted group was 94.29% (P = 0.042), which was higher than conventional THA group (67.56%).
Chai et al, 2020	Mako	publication [21]	Robot-assisted surgery is suitable for complex THA, the postoperative Harris scores were 83/86 (left/right) for hip dysplasia, 87 for post-traumatic hip osteoarthritis and 62 (poor) for ankylosing spondylitis after robot-assisted surgery.
Fu et al, 2017	Mako	publication [25]	The recommended learning curve (LC) of Mako robot-assisted UKA was 8 cases.
Zhu et al, 2019	Mako	publication [27]	The increase of Knee Society clinical and functional scores were 37.2 and 29, and the range of motion of the knee joint increased 37.2° in Robotic-assisted UKA patients.
Wang et al, 2021	HURWA	publication [28]	No significant difference in the correction rate of HKA between robot surgery group and traditional surgery group (63.6% vs. 69.2%, P = 0.651).
Xia et al, 2021	Skywalker	publication [29]	100% of the absolute error of the HKA angles were within 3°; 90.32% of the postoperative lower limb alignment angles in 31 patients were close to 180° after the operation.
Chai et al, 2020	YUANHUA	publication [31]	The deviation between the planned cutting thickness before operation and the measured value after the operation was < 1 mm and that the error of the osteotomy angle was < 2° in the goat models.
Chai et al, 2020	YUANHUA	publication [32]	The postoperatively measured hip-knee-ankle angle was 177.1°–179.7°, the coronal femoral component angle was 87.9°–91.4°, and coronal tibial component angle was 87.3°–91.4°, with errors being within ±3° of the angles planned before operation in cadaver trials.
Chai W, 2021	Mako	conference speech (2021 ICRA)	A abnormal hip-spine relationship is possible in THA patients with dislocation due to unclear causes. An improved safety zone calculation method was established to provide accurate guidance for acetabular cup implantation.
Zhang XG, 2021	Mako	conference speech (2021 ICRA)	Among 64 Mako robot-assisted THA cases, 2 experienced loosening of the iliac assembly that caused registration failure; 1 case suffered iliac bone fracture; 3 cases had errors in acetabulum registration that caused the misalignment of the acetabular cup due to anteversion; and 1 case showed acetabular cup loosening.
Zhang Y, 2021	YUANHUA	conference speech (2021 ICRA)	Four types of “Learning Curve (LC)” was redefined in robotic-assisted TKA: LC in terms of operative measures was 6–8 cases; No LC exist in terms of radiologic results; LC in terms of postoperative function scores (KSS, HSS) was 14 cases; the terms of traumatic variables, such as blood loss, erythrocyte sedimentation rate, C-reactive protein, were less than conventional group (P < 0.05).
Zhang Y, 2019	StealthStat-ion S7 (Medtronic)	conference speech (2019 Chinese Knee Society)	In a clinical study of 14 knee OA with constitutional femoral varus, 7 patients underwent computer navigation-assisted TKA, none had pie-crust releasing of MCL or peeling off its posteromedial tibial attachment; compared to 4 cases of releasing of the superficial MCL and 3 cases of peeling off the tibial attachment in 7 patients of conventional TKA.
Zhou YX, 2021	Mako	conference speech (2021 ICRA)	An improved registration method of robot-assisted surgery was developed for acetabular revision. Among 45 cases of acetabular revision, the acetabular cup was within the Lewinnek safety zone in 43 cases and within the Callanan safety zone in 38 cases.
Zhou ZK, 2021	YUANHUA	conference speech (2021 ICRA)	The postoperative lower limb axis deviation ranges were 0–2° for the robot-assisted group and 0–5° for the conventional group; The mean posterior slope of the tibial component was 2.8° in the test group and 5.3° in the control group, with fewer outliers in the test group.