Volume 132 · Issue 21 · 2019

人工智能（AI）由约翰·麦卡锡教授于1956年首次提出，旨在利用计算机再现人类智能。机器学习（ML）是人工智能的一种形式，它使用计算算法，根据经验进行学习和改进。^[1]ML的两种主要形式是监督和无监督。在监督ML中，算法被给予标记数据，用于预测新患者的疾病结果。相比之下，无监督ML用于在没有训练的情况下识别模式；该算法通过搜索共同特征来学习数据的内在结构。^[1]

Available research about the anatomic patterns of intertrochanteric fractures is lacking, and fracture mapping has not previously been performed on intertrochanteric fractures. This study aimed to determine the major trajectories of intertrochanteric fracture lines using computed tomography data from a series of surgically treated patients.

In this study, 504 patients with intertrochanteric fractures were retrospectively analyzed. Fracture patterns were graded according to Arbeitsgemeinschaft für Osteosynthesefragen (AO) classification. Fracture lines were transcribed onto proximal femoral templates and graphically superimposed to create a compilation of fracture maps that were subsequently divided into anterior, posterior, lateral, and medial fracture maps to create a three-dimensional (3D) pattern by reducing fragments in the 3D models. The fracture maps were then converted into frequency spectra. The major fracture patterns were assessed by focusing on the lateral femoral wall, lesser trochanter, intertrochanteric crest, and inner cortical buttress.

Anterior, posterior, lateral, and medial fracture maps were created. The majority of fracture lines (85.9%, 433/504) on the anterior maps were along the intertrochanteric line where the iliofemoral ligament was attached. In the medial plane, the majority of fracture lines (49.0%, 247/504) shown on the frequency spectrum included the turning point involving the third quadrant. In the posterior plane, the majority of fracture lines (52.0%, 262/504) involved the intertrochanteric crest from the greater to the lesser trochanter. In the lateral plane, the majority of fracture lines (62.7%, 316/504) involved the greater trochanter at the gluteus medius attachment.

The fracture patterns observed in the present study might be used to describe morphologic characteristics and aid with management strategies. Further classifications or modifications that incorporate the fracture patterns identified in this study may be used in future research.

Reverse intertrochanteric fractures are usually initially treated with closed reduction. However, sometimes these fractures are not amenable to closed reduction and require open reduction. To date, few studies have been conducted on predictors of and reduction techniques for irreducible reverse intertrochanteric fractures. Therefore, this study aimed to summarize the displacement patterns of irreducible reverse intertrochanteric fractures and corresponding reduction techniques, and explore predictors of irreducibility.

We reviewed 1174 cases of trochanteric fractures treated in our hospital from January 2006 to October 2018, 113 of which were reverse intertrochanteric fractures. An irreducible fracture was determined according to intra-operative fluoroscopy imaging after closed manipulation. Fractures were assessed for displacement patterns, radiographic features of irreducibility, and reduction techniques. Logistic regression analysis was performed on potential predictors for irreducibility, including gender, age, body mass index, AO Foundation/Orthopaedic Trauma Association (AO/OTA) classification, and radiographic features.

Seventy-six irreducible fractures were identified, accounting for 67% of reverse intertrochanteric fractures. Six patterns of fracture displacement after closed manipulation were identified; the most common pattern was medial displacement and posterior sagging of the femoral shaft relative to the head-neck fragment. Multivariate logistic regression analysis identified three predictors of irreducibility: a medially displaced femoral shaft relative to the head-neck fragment on the anteroposterior (AP) view (odds ratio [OR], 8.00; 95% confidence interval [CI], 3.04-21.04; P < 0.001), a displaced lesser trochanter (OR, 3.61; 95% CI, 1.35-9.61; P = 0.010), and a displaced lateral femoral wall (OR, 2.92; 95% CI, 1.02-8.34; P = 0.046).

A high proportion of reverse intertrochanteric fractures are not amenable to closed reduction. Six patterns of fracture displacement after closed manipulation were identified. Different reduction techniques are required for different displacement patterns. Predictors of irreducibility include a medially displaced femoral shaft relative to the head-neck fragment on the AP view, a displaced lesser trochanter, and a displaced lateral femoral wall. These patients warrant special consideration in terms of recognition and management.

The effect of short-segment decompression/fusion versus long-segment decompression/fusion and osteotomy for Lenke-Silva type VI adult degenerative scoliosis (ADS) has not been clarified. This study aimed to compare the clinical and radiographic results of short-segment fusion vs. long-segment fusion and osteotomy for patients with Lenke-Silva type VI ADS.

Data of 28 patients who underwent spinal surgery for ADS from January 2012 to January 2014 in the General Hospital of Northern Theater Command were reviewed. Of the 28 patients, 12 received long-segment fusion and osteotomy and 16 received short-segment fusion. Radiographic imaging parameters and clinical outcomes, including the sagittal vertical axis (SVA), lumbar lordosis (LL) angle, pelvic tilt (PT), sacral slope (SS), the visual analog scale (VAS), Japanese Orthopedic Association (JOA), Oswestry disability index (ODI), and lumbar stiffness disability index (LSDI) scores, were recorded. The difference between groups was compared using the dependent t test or Chi-squared test.

The Cobb and LL angles and SVA improved in both groups; however, PT and SS angles did not improve following short fusion. There were significant differences in the post-operative SVA (26.8 ± 5.4 mm vs. 47.5 ± 7.6 mm, t = -8.066, P < 0.001), PT (14.7 ± 1.8° vs. 29.1 ± 3.4°, t = -13.277, P < 0.001), and SS (39.8 ± 7.2° vs. 26.1 ± 3.3°, t = 6.175, P < 0.001) between the long and short fusion groups. All patients had improved ODI, JOA, and VAS scores post-operatively (all P < 0.001), with no significant difference between the groups (all P > 0.05). The post-operative LSDI score was 3.5 ± 0.5 in the long fusion group, which was significantly higher than that of the short fusion group (1.4 ± 0.7; P < 0.001).

The clinical outcomes of patients with Lenke-Silva type VI ADS who underwent short-segment decompression/fusion were comparable to those of patients who underwent long-segment decompression/fusion and osteotomy despite poor correction of sagittal imbalance. Moreover, short-segment decompression/fusion showed a short operation time and reduced surgical trauma.

The potential benefit of arthroscopic surgery for osteoarthritic knee associated with medial meniscus tear is controversial. This study was conducted to determine the effect of pre-operative medial meniscus extrusion (MME) on arthroscopic surgery outcomes in the osteoarthritic knee associated with medial meniscus tear during a minimum 4-year follow-up.

This was a retrospective review of a total of 131 patients diagnosed with osteoarthritic knee associated with medial symptomatic degenerative meniscus tear who underwent arthroscopic surgery from January 2012 to December 2014 and were observed for more than 4 years. Patients were classified into two groups: MME ≥3 mm (major MME group, n = 54) and MME <3 mm (non-major MME group, n = 77). Clinical assessments, including the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) score, and radiographic assessments, including the Kellgren-Lawrence (K-L) grade and medial joint space width (JSW), were evaluated pre-operatively and at final follow-up. The longitudinal changes of clinical and radiographic parameters (WOMAC and the medial JSW change, K-L grade progression) were compared between groups unadjusted and adjusted for age, sex, and body mass index. Four-year survival rates (without progression to knee replacement [KR]) were also evaluated using a log-rank test and Cox proportional hazard regression model.

Major MME was present in 41% of patients. After a minimum 4-year follow-up, the mean WOMAC total and pain scores improved significantly in both groups. However, the medial JSW and K-L grade worsened significantly. Patients with pre-operative major MME worsened more in WOMAC total (adjusted mean difference [MD] 3.800, 95% confidence interval [CI]: 0.900, 11.400; P = 0.037) and function (adjusted MD 3.100, 95% CI: 0.700, 6.300; P = 0.038) scores than patients with pre-operative non-major MME, and no significant difference was observed in WOMAC pain and stiffness score between groups. The group with major MME had significantly higher joint space narrowing (adjusted MD -0.630, 95% CI: -1.250, -0.100; P = 0.021) and K-L rate progression (adjusted mean relative risk [RR] 1.310, 95% CI: 1.100, 1.600; P = 0.038) than the group with non-major MME. There was a significantly more KR progression in patients with major MME compared with those with non-major MME (adjusted RR 3.100, 95% CI: 1.100, 9.200; P = 0.042 and adjusted hazard ratio 3.500, 95% CI 1.100, 9.500; P = 0.022).

Osteoarthritic knee patients associated with medial meniscus tear with non-major MME are more responsive to arthroscopic surgery in terms of the clinical and radiologic outcomes and survival for at least 4-year follow-up; however, in terms of pain relief, arthroscopic surgery in patients with major MME is also beneficial as well as in patients with non-major MME.

The size of the glenoid bone defect is an important index in selecting the appropriate treatment for anterior shoulder instability. However, the reliability of glenoid bone defect measurement is controversial. The purpose of the present study was to investigate the reliabilities of measurements of the glenoid bone defect on computed tomography and to explore the predisposing factors leading to inconsistency of these measurements.

The study population comprised 69 consecutive patients who underwent surgery for recurrent anterior shoulder dislocation in Peking University Fourth School of Clinical Medicine from March 2016 to January 2017. The glenoid bone defect was measured by three surgeons on 'self-confirmed’ and 'designated’ 3-D en-face views, and repeated after an interval of 3 months. Measurements included the ratio of the defect area to the best-fit circle area, and the ratio of the defect width to the diameter of the best-fit circle. The inter- and intra-observer reliabilities of the measurements were evaluated using intraclass correlation coefficients (ICCs). The maximum absolute inter- and intra-observer differences and the cumulative percentages of cases with inter- and intraobserver differences greater than these respective levels were calculated.

Almost all linear defect values were bigger than the areal defect values. The inter-observer ICCs for the areal defect were 0.557 and 0.513 in the 'self-confirmed’ group and 0.549 and 0.431 in the 'designated’ group. The inter-observer reliabilities for the linear defect were moderate or fair in the 'self-confirmed’ group (ICC = 0.446, 0.374) and 'designated’ group (ICC = 0.402, 0.327). The ICCs for intra-observer measurements were higher than those for inter-observer measurements. The respective maximum interand intra-observer absolute differences were 13.9% and 13.2% in the 'self-confirmed’ group, and 15.8% and 9.8% in the 'designated’ group.

The areal measurement of the glenoid bone defect is more reliable than the linear measurement. The reliability of the glenoid defect areal measurement is moderate or worse, suggesting that a more accurate and objective measurement method is needed in both en-face view and best-fit circle determination. Subjective factors affecting the glenoid bone loss measurement should be minimized.

The predominant method for Manske type IIIB and IV thumb hypoplasia is pollicization. However, for those who are not willing to sacrifice the index finger, a method that could reconstruct a functionally capable and aesthetically acceptable thumb remains desirable. This study aimed to investigate and assess the functional and radiographic outcomes of utilizing a reversed vascularized second metatarsal composite flap for thumb reconstruction as a new alternative.

From May 2014 to January 2017, 15 patients with Manske type IIIB or IV thumb hypoplasia who were admitted to the Department of Hand Surgery, Beijing Jishuitan Hospital were included in this study. An osteocutaneous flap containing a section of second metatarsal and its distal head was transferred in reversed position to reconstruct carpometacarpal joint. The donor site was reconstructed by a split half of the third metatarsal. Various functional reconstructions were commenced at second stage. The reconstructed thumbs were evaluated using the Kapandji score, pinch force, and the capacities of performing daily activities through a detailed questionnaire.

Among these 15 patients (seven type IIIB and eight type IV), there were ten boys and five girls with median age of 4.2 years (range: 2.0-7.0 years). There were seven right, three left, and five bilateral thumbs for whom only the right thumb received surgery. There were 14 metatarsal flaps survived (14/15). With an average follow-up of 19.2 months, the reconstructed thumbs had acceptable functional and aesthetic outcomes and the donor foot presented in decent appearance without signs of impaired function. All 15 children have improved the Kapandji score (from 0 to an average of 6.7), pinch force (from 0 to an average of 1.5 kg), with ability of grip and pen holding. X-ray indicated continuous bone growth. Patients and parents had good acceptance of the new thumb.

Reconstruction of an unstable hypoplastic thumb (Manske type IIIB and IV) with use of a vascularized metatarsal is an effective strategy. It offers an alternative solution for parents insisting on saving the thumb.

The angiography with micro-computed tomography (micro-CT) has been proved its great advantages on investigating the intra-osseous vascularity of carpal bones. But few researches have focused on the intra-hamate vascularity. This study aimed to illustrate the intra-osseous arteries of the hamate and the relationship between the intra-hamate vascularity and the avascular risk of different types of hamate fractures.

Six normal cadaveric hamates were investigated with red lead (Pb₃O₄) micro-CT angiography. The intra-osseous arteries of specimens were clearly enhanced and the three-dimensional model was reconstructed. In order to study the features of the arterial entrances and intra-hamate vascularity, the diameters, quantities, locations of enhanced arteries, and the locations of transversal/proximal pole fracture lines on the body of the hamate were statistically compared. Besides, in order to analyze the relationship between intra-hamate vascularities and different hamate fractures, 127 cases of hamate fractures who presented in our hospital from March 2003 to June 2017 were retrospectively studied.

A total of 94 cases were followed up (range: 4-37 months; mean: 12.4 months) effectively. The overall union rate of hamate fractures was as high as 92.6% (87 of 94 cases), while non-union of fracture on hamate hook was more common (P = 0.031). The arterial entrances were located around the dorsal, volar, radial, ulnar non-articular surfaces of the hamate body and the hook of the hamate. Generally, there were one to two trunk arteries on the volar non-articular surface and one to three trunk arteries on the dorsal non-articular surface. They formed one or two arterial arches, from which some branches were emitted and supplied the proximal parts. The intra-osseous vascularities of the hamate body were generally located in the radial part. The blood supply of the hook was mainly from the volar non-articular surface in most specimens. Hamate fractures could be classified into four types: fractures of the transversal/proximal pole, medial tuberosity, dorsal coronal of the hamate body, and fractures of the hamate hook.

This study showed new features of intra-hamate vascularity and the results will guide surgeons to reduce the vascular damage during the hamate fracture operations. The fracture lines of different types of hamate fractures may disrupt the intra-hamate arteries. The intra-hamate vascularities will have different influences on the avascular risks of different hamate fractures.

Plant homeodomain finger protein 23 (PHF23) is a novel autophagy inhibitor gene that has been few studied with respect to orthopedics. This study was to investigate the expression of PHF23 in articular cartilage and synovial tissue, and analyze the relationship between PHF23 and chondrocyte autophagy in osteoarthritis (OA).

Immunohistochemical staining and western blot were applied to show the expression of PHF23 in cartilage of different outbridge grades and synovial tissue of patient with OA and healthy control. The normal human chondrocyte pre-treated with rapamycin or 3-methyladenine, treated with interleukin-1β (IL-1β). IL-1β induced expression level of PHF23 and autophagyrelated proteins light chain 3B-I (LC3B-I), LC3B-II, and P62, were examined by Western blot. A PHF23 gene knock-down model was constructed with small interfering RNA. Western blot was performed to detect the efficiency of PHF23 and the impact of PHF23 knockout on IL-1β-induced expression of autophagy-related and apoptotic-related proteins in chondrocyte.

The expression of PHF23 was significantly increased in the high-grade cartilage and synovial tissue of patients with OA. The IL-1β-induced expression of PHF23 was gradually enhanced with time. The level of LC3B-II, P62 changed with time. After knockdown of PHF23, the level of autophagy-related proteins increased and apoptotic-related proteins decreased in IL-1β-induced OA-like chondrocytes.

The expression of PHF23 increased in human OA cartilage and synovium, and was induced by IL-1β through inflammatory stress. PHF23 can suppress autophagy of chondrocytes, and accelerate apoptosis.

Refractory femoral neck fractures cannot be anatomically reduced by closed traction reduction which may affect fracture healing. We evaluated the biomechanical effects of positive, negative, and anatomic reduction of various degrees of displacement in Pauwels I femoral neck fractures by a finite element analysis.

Five reduction models of Pauwels type I femoral neck fracture were established using the Mimics 17.0 (Materialize, Leuven, Belgia) and Hypermesh 12.0 (Altair Engineering, Troy, MI, USA). According to the degree of fracture displacement, there were three models of positive support, an anatomic reduction model, and a negative 2 mm reduction model. Finite element analysis was conducted using the ABAQUS 6.9 software (Simulia, Suresnes, France). The von Mises stress distribution and the stress peak of internal fixation in different models, the displacement between fracture blocks, and the principal strain of the femoral neck cancellous bone model were recorded under the axial stress of 2100 N.

The peak von Mises stress on screw of each model was located at the thread of the screw tip. The peak von Mises stress was the lowest at the tip of the anatomic reduction model screw (261.2 MPa). In the positive 4 mm model, the von Mises stress peak was the highest (916.1 MPa). The anatomic reduction model showed the minimum displacement (0.388 mm) between fracture blocks. The maximum displacement was noted in the positive 4 mm model (0.838 mm). The displacement in the positive 3 mm model (0.721 mm) was smaller than that in the negative 2 mm model (0.786 mm). Among the five models, the strain area of the femoral neck cancellous bone was mainly concentrated around the screw hole, and the area around the screw hole could be easily cut.

Compared with negative buttress for femoral neck fracture, positive buttress can provide better biomechanical stability. In Pauwel type I fracture of femoral neck, the range of positive buttress should be controlled below 3 mm as far as possible.

Reports on the efficacy of modifications to the thread design of pedicle screws are scarce. The aim of the study was to investigate initial and early fixation of pedicle screws with a plasma-sprayed titanium coating and dual pitch in the pedicle region (dual pitch titanium-coated pedicle screw [DPTCPS]) in a polyetheretherketone (PEEK) rod semi-rigid fixation system.

Fifty-four sheep spine specimens and 64 sheep were used to investigate initial ( "0-week" controls) and early (postoperative 6 months) fixation, respectively. Sheep were divided into dual pitch pedicle screw (DPPS), standard pitch pedicle screw (SPPS), DPTCPS, and standard pitch titanium-coated pedicle screw (SPTCPS) groups. Specimens/sheep were instrumented with four screws and two rods. Biomechanical evaluations were performed, and histology at the implant-bone interface was investigated.

At 0-week, mean axial pull-out strength was significantly higher for the DPTCPS and SPTCPS than the SPPS (557.0 ± 25.2 vs. 459.1 ± 19.1 N, t = 3.61, P < 0.05; 622.6 ± 25.2 vs. 459.1 ± 19.1 N, t = 3.43, P < 0.05). On toggle-testing, the DPTCPS was significantly more resistant than the SPPS and SPTCPS (343.4 ± 16.5 vs. 237.5 ± 12.9 N, t = 3.52, P < 0.05; 343.4 ± 16.5 vs. 289.9 ± 12.8 N, t = 3.12, P < 0.05; 124.7 ± 13.5 vs. 41.9 ± 4.3 cycles, t = 2.18, P < 0.05; 124.7 ± 13.5 vs.79.5 ± 11.8 cycles, t = 2.76, P < 0.05). On cyclic loading, maximum displacement was significantly lower for the DPTCPS than the SPPS and SPTCPS (1.8 ± 0.13 vs. 3.76 ± 0.19 mm, t = 2.29, P < 0.05; 1.8 ± 0.13 vs. 2.46 ± 10.20 mm, t = 2.69, P < 0.05). At post-operative 6 months, mean axial pull-out strength was significantly higher for the DPTCPS and SPTCPS than the SPPS (908.4 ± 33.6 vs. 646.5 ± 59.4 N, t = 3.34, P < 0.05; 925.9 ± 53.9 vs. 646.5 ± 59.4 N, t = 3.37, P < 0.05). On toggle-testing, the DPTCPS was significantly more resistant than the SPPS and SPTCPS (496.9 ± 17.9 vs. 370.3 ± 16.4 N, t = 2.86, P < 0.05; 496.9 ± 17.9 vs. 414.1 ± 12.8 N, t = 2.74, P < 0.05; 249.1 ± 11.0 vs.149.9 ± 11.1 cycles, t = 2.54, P < 0.05; 249.1 ± 11.0 vs.199.8 ± 7.2 cycles, t = 2.61, P < 0.05). On cyclic loading, maximum displacement was significantly lower for the DPTCPS than the SPPS and SPTCPS (0.96 ± 0.11 vs. 2.39 ± 0.14 mm, t = 2.57, P < 0.05; 0.96 ± 0.11 vs. 1.82 ± 0.12 mm, t = 2.73, P < 0.05). Resistance to toggle testing (370.3 ± 16.4 vs. 414.1 ± 12.8 N, t = 3.29, P < 0.05; 149.9 ± 11.1 vs.199.8 ± 7.2 cycles, t = 2.97, P < 0.05) was significantly lower and maximum displacement in cyclic loading (2.39 ± 0.14 vs.1.82 ± 0.12 mm; t = 3.06, P < 0.05) was significantly higher for the SPTCPS than the DPTCPS. Bone-to-implant contact was significantly increased for the DPTCPS compared to the SPPS (58.3% ± 7.0% vs. 36.5% ± 4.4%, t = 2.74, P < 0.05); there was no inflammatory reaction or degradation of coated particles.

DPTCPSs might have stronger initial and early fixation in a PEEK rod semi-rigid fixation system.

人体骨骼是动态变化的。对于人类来说，骨骼通常在胚胎期的第7周开始形成，并继续发育，直到人类达到骨骼成熟。在这个发育阶段，人体骨骼的质量、密度、形态、硬度和强度都在不断变化。有研究描述了人体从胚胎期到青春期各部位骨骼形态的动态变化。例如，股骨颈扭转角(FNTA)在出生时约为30至40度，到青春期早期该角度逐渐后退至10至15度。婴儿具有较大的股骨颈轴角（NSA），并且该角度逐渐减小至成人常见的角度。最近的研究集中于骨骼感知、转导和响应机械负荷的生理和分子机制，并研究了衰老过程对皮质骨特性（如横截面几何形状和骨矿物质密度）和机械功能之间关系的影响。^[1,2]

空地无线电话通信为民用飞行员和空中交通管制员在飞行的所有阶段提供了唯一的通信方式。飞行过程中的通信质量对飞机安全至关重要。因此，由于飞行过程中缺乏视觉提示，飞行员具有良好的听觉功能至关重要。在经常暴露于嘈杂的职业噪音环境（如驾驶舱）的民用飞行员中，听力损失可能会受到威胁，并可能使机组人员和乘客处于危险之中。无线电话是一种半人工语言，基于标准化、程序化、清晰、精确的祈使句。在中国国内航线上广泛使用的普通话无线电话术语有其自身的特点，由中文普通话术语、英文大写字母和术语英文缩写三部分组成。在中国，只有能够满足值班听觉健康（AFFD）的飞行员^[1]中国民用航空局（CAAC；中国北京）的标准被认证为拥有足够的听力以确保安全飞行。目前CAAC的AFFD测试主要依赖于纯音测听（PTA）。然而，先前的研究表明，PTA可能不适合决定个人令人满意地执行工作的能力，因为在稳态噪声中识别语音的能力不能用听力图来预测。因此，在这项研究中，我们旨在开发一套专门为民用飞行员设计的语音测听材料，作为CAAC的功能性附加AFFD测试。

致编辑：扩张开放椎板成形术（EOLP）通常用于各种原因的多节段脊髓型颈椎病患者。然而，仍然存在一些并发症。其中，移位的铰链侧完全骨折和椎板再闭合可导致神经功能恶化。陈等人^[1]据报道，与传统缝合技术相比，使用微型接骨板可显著降低椎板再闭合的发生率。然而，微型钢板组和缝线悬吊组的完全骨折率约为56.8%。据报道，准确定位槽位置可降低铰链侧完全骨折的发生率。^[2]在这项研究中，制造了三维（3D）打印的患者特异性钻孔模板来引导EOLP中的谷位置，^[3]验证3D打印模板的安全性和有效性。

致编辑：脊柱后凸畸形是强直性脊柱炎（AS）的主要表现之一。通常，过度后凸沿腰椎、胸腰椎和胸椎分布，因为脊柱的骨化主要始于骶髂关节，然后向头部发展。^[1]偶见颈胸椎发生后凸畸形，导致下巴靠在胸部，称为下巴靠胸畸形。下巴对胸部畸形的患者可能在水平凝视、咀嚼和吞咽、保持和执行适当的个人卫生方面有问题，他们可能患有颈部疼痛和脊髓病。^[1]然而，AS中严重的整体脊柱后凸，导致上半身弯曲类似字母C，是一种极其罕见的同时伴有腰椎、胸椎和颈胸脊柱后凸的疾病。当这些患者发生颈椎骨折时，由于畸形的严重性和骨折的不稳定性，处理起来非常具有挑战性。在这项研究中，我们报告了我们在治疗一名报告这些问题的患者的经验。