• Communications for Statistical Applications and Methods
• /
• v.31 no.2
• /
• pp.203-212
• /
• 2024

The area under the ROC curve (AUC) is one of the most common criteria used to measure the overall performance of binary classifiers for a wide range of machine learning problems. In this article, we propose a L₁-penalized AUC-optimization classifier that directly maximizes the AUC for high-dimensional data. Toward this, we employ the AUC-consistent surrogate loss function and combine the L₁-norm penalty which enables us to estimate coefficients and select informative variables simultaneously. In addition, we develop an efficient optimization algorithm by adopting k-means clustering and proximal gradient descent which enjoys computational advantages to obtain solutions for the proposed method. Numerical simulation studies demonstrate that the proposed method shows promising performance in terms of prediction accuracy, variable selectivity, and computational costs.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.9 no.11
• /
• pp.4604-4622
• /
• 2015

It is not easy to reconstruct the geometrical characteristics of the distorted images captured by the devices. One of the most popular optimization methods is fast iterative shrinkage/ thresholding algorithm. In this paper, to deal with its approximation error and the turbulence of the decrease process, an adaptive proximal conjugate gradient (APCG) framework is proposed. It contains three stages. At first stage, a series of adaptive penalty matrices are generated iterate-to-iterate. Second, to trade off the reconstruction accuracy and the computational complexity of the resulting sub-problem, a practical solution is presented, which is characterized by solving the variable ellipsoidal-norm based sub-problem through exploiting the structure of the problem. Third, a correction step is introduced to improve the estimated accuracy. The numerical experiments of the proposed algorithm, in comparison to the favorable state-of-the-art methods, demonstrate the advantages of the proposed method and its potential.

• Bulletin of the Korean Mathematical Society
• /
• v.54 no.4
• /
• pp.1347-1359
• /
• 2017

The purpose of this paper is to give some new iterative methods for finding a common zero of a finite family of monotone operators in Hilbert spaces. We also give the applications of the obtained result for the convex feasibility problem and constrained convex optimization problem in Hilbert spaces.

• The Journal of the Korean bone and joint tumor society
• /
• v.2 no.1
• /
• pp.8-17
• /
• 1996

The purpose of this study is a comparative evaluation of range motion, especially extension deficit between the group of total patellectomy and that of intact patella, after reconstruction of the patellar tendon in the prosthetic replacement of a proximal tibia. Between 1990 and 1994, 15 patients who had a primary malignancy on proximal tibia were operated on. All patients were evaluated clinically and radiographically. Two patients were excluded because one had a deep infection treated with arthrodesis of the knee and the other was a composite allograft. The mean follow-up of the 13 patients was 27 months(15-47), including 10 osteosarcomas, 1 chondrosarcoma, 1 malignant fibrous histiocytoma and 1 malignant giant cell tumor. Eleven patients had a resection of the proximal tibia and 2 had an extracapsular total knee resection with distal femur. Reconstruction of the defect was done in 8 cases with a custom-made Link Endo-Model Total Rotation Knee Joint Prosthesis, and in 5 with How Medica Modular Resection System (HMRS). We used two methods to reconstruct the ligamentum patellae. Fixation of the patellar tendon to the prosthesis only with suturing and/or stapling(group SS) was done in 7. Transposition of gastrocnemius muscle to enhance fixation and to cover the prosthesis(group TG) was done in 6. Regardless of fixation methods, total patellectomy was done in 5 either to lengthen the patellar tendon or to make primary skin closure easier or for both. In 8 cases, patella was left intact or resurfaced with polyethylene prosthesis. Active extension was measured while the patient was in a sitting position. There is no statistically meaningful difference in terms of extension deficit (Wilcoxon rank test, p=0.8800) between patellectomy group and intact patella group, and between group of fixation only with suturing and that of gastrocnemius transposition. Two cases of extension deficit over 30 degree were seen in group SS and in the group of intact patella. Conclusively, total patellectomy could be an option without increasing the risk of extension deficit when primary skin closure is difficult or patellar tendon is a little bit short to be fixed. There is no rating in the Enneking system of functional evaluation that this finding into consideration.

• Journal of Gastric Cancer
• /
• v.20 no.3
• /
• pp.277-289
• /
• 2020

Purpose: To evaluate the feasibility and safety of intracorporeal overlapping gastrogastrostomy between the proximal anterior wall and antrum posterior wall (PAAP; PAAP anastomosis) of the stomach in minimally invasive pylorus-preserving gastrectomy (PPG) for early gastric cancer (EGC). Materials and Methods: From December 2016 to December 2019, 17 patients underwent minimally invasive PPG with PAAP anastomosis for EGC in the high body and posterior wall of the stomach. Intraoperative gastroscopy was performed with the rotation maneuver during proximal transection. A longer antral cuff (>4-5 cm) was created for PAAP than for conventional PPG (≤3 cm) at the point where a safe distal margin and good vascular perfusion were secured. Because the posterior wall of the proximal remnant stomach was insufficient for intracorporeal anastomosis, the anterior wall was used to create an overlapping anastomosis with the posterior wall of the remnant antrum. The surgical and oncological outcomes were analyzed, and the stomach volume was measured in patients who completed the 6-month follow-up. The results were compared to those after conventional PPG (n=11 each). Results: PAAP anastomosis was successfully performed in 17 patients. The proximal and distal resection margins were 2.4±1.9 cm and 4.0±2.6 cm, respectively. No postoperative complications were observed during the 1-year follow-up esophagogastroduodenoscopy (n=10). The postoperative remnant stomach (n=11) was significantly larger with PAAP than with conventional PPG (225.6±118.3 vs. 99.1±63.2 mL; P=0.001). The stomach length from the anastomosis to the pylorus was 4.9±2.4 cm after PAAP. Conclusions: PAAP anastomosis is a feasible alternative for intracorporeal anastomosis in minimally invasive PPG for highly posteriorly located EGC.

• Clinics in Shoulder and Elbow
• /
• v.11 no.2
• /
• pp.137-142
• /
• 2008

Purpose: To evaluate the clinical and radiographic results of the treatment of a proximal humerus fracture with a locking compression plate(LCP). Materials and Methods: This study reviewed the results of 24 cases of a proximal humerus fracture treated with a LCP from January 2005 to April 2007, after a follow up of more than 12 months. There were 8 males and 16 females with a mean age of 68.9(33-90) years. The clinical results were evaluated using the Neer's evaluation criteria, and the radiographic results were evaluated using the bone union time and Paavoleinen method. Results: The mean time for bone union was 11.9 weeks. Using the Neer's functional evaluation, 21 cases of the 24 patients (87%) showed excellent or satisfactory results. Twenty two cases (91%) showed good results according to the Paavoleinen method. The complications encountered were metal failure (1 patient), AVN of the humeral head (1 patient) and joint stiffness (1 patient). Conclusion: The LCP demonstrated good results in the treatment of a proximal humerus fracture and has relatively fewer complications than other internal fixators.

• Journal of Korean Foot and Ankle Society
• /
• v.8 no.2
• /
• pp.121-125
• /
• 2004

Purpose: We try to retrospectively evaluated the amount of dorsal angulation angle of the first metatarsal commonly occurring as the complication of proximal dome osteotomy for hallux valgus. Materials and Methods: Between January 2004 and March 2004, 34 patients who underwent proximal dome osteotomy for moderate to severe hallux valgus. Two of 34 patients were male, and thirty-two were female. The average age was 57.6 years. We measured and compared hallux valgus angle, 1st-2nd intermetatarsal angle, dorsal angulation angle of 1st metatarsal on preoperative, postoperative, postoperative 3 weeks', postoperative 3 months' X-ray. Results: Osteotomy sites were completely united on plane X-ray in all cases. The hallux valgus angle averaged $41.2^{\circ}$ ($30{\sim}60^{\circ}$) at preoperative, $4.3^{\circ}$ ($-10{\sim}20^{\circ}$) at postoperative, $5.5^{\circ}$ ($-1{\sim}20^{\circ}$) at 3 weeks after operation, $7.8^{\circ}$ ($-2{\sim}20^{\circ}$) at 3 months after operation. The 1st-2nd intermetatarsal angle averaged $17.1^{\circ}$ ($12{\sim}24^{\circ}$) at preoperative, $6.3^{\circ}$ ($0{\sim}13^{\circ}$) at postoperative, $7.2^{\circ}$ ($0{\sim}15^{\circ}$) at 3 weeks after operation, $8.7^{\circ}$ ($0{\sim}18^{\circ}$) at 3 months after operation. The dorsal angulation angle averaged $0.4^{\circ}$ ($0{\sim}3^{\circ}$) at postoperative, $1.6^{\circ}$ ($0{\sim}7^{\circ}$) at 3 weeks after operation, $2.1^{\circ}$ ($0{\sim}8^{\circ}$) at 3 months after operation. There were no statistically correlation between increase of dorsal angulation angle of the distal segment of the first metatarsal and increase of hallux valgus angle or 1st-2nd intermetatarsal angle. Conclusion: Our results shows that the dorsal angulation of distal fragment occurring after the proximal dome osteotomy in the treatment of hallux valgus may be minimized with meticulous surgery and patient's education.

• Journal of Korean Orthopaedic Sports Medicine
• /
• v.11 no.1
• /
• pp.24-29
• /
• 2012

Purpose: To evaluate follow-up after the clinical results of arthroscopic assisted tenodesis of the proximal biceps tendon using an interference screw. Materials and Methods: We reviewed 30 cases of patients receiving tenodesis of the proximal biceps tendon between January, 2008 and January, 2010 who treated with arthroscope. The results were judged by VAS, ASES, tenderness in the biceps groove, fixation failure and the degree of deformity. Results: VAS, ASES scores increased by a statistically significant during the final observation in all patients than before surgery (p<0.05). For those with rotator cuff tear had low ASES results (p<0.05), respectively. But, tenderness in the biceps groove was 17% (5 cases), pepeye deformity and incresead BAD (biceps apex distance) were 10% (3 cases). Conclusion: Arthroscopic tenodesis of the proximal biceps tendon using an interference screw to patients with proximal lesion in their biceps tendon, can achieve good results, but it is possible to groove tenderness and popeye deformity.

• Journal of Korean Neurosurgical Society
• /
• v.29 no.9
• /
• pp.1188-1194
• /
• 2000

Objective : Internal carotid-posterior communicating artery(IC-PC) aneurysms can be clipped easily without any special preparations. Occasionally, however, it is difficult to clip the low-lying IC-PC aneurysms without some kinds of additional procedures. Clinical Material and Methods : We experienced four cases of low-lying IC-PC aneurysms, which the intradural anterior clinoidectomy and/or anterior petroclinoid fold(APF) resection was essential to expose the proximal side of the aneurysmal neck and/or proximal control. One patient harbored two low-lying IC-PC aneurysms bilaterally. The patients were divided into two groups according to the necessity of anterior clinoidectomy : Group I(n=4) that needed an intradural clinoidectomy and/or APF resection and Group II(n=29) that had IC-PC aneurysms, easily clipped without any special preparation. Also, various radiometric parameters were measured through the preoperative angiograms. Results : The incidence of such aneurysms was 12% among a total of thirty-three surgically treated IC-PC aneurysms during lasr 3 years. Among four cases, three cases presented with subarachnoid hemorrhage and all aneurysmal sac projected to postero-inferior direction. In our study, We initially considered the necessity of intraoperative anterior clinoid process(ACP) removal and/or resection of APF in cases of shorter distance less than 5.6mm between the proximal aneurysmal sac and tip of the ACP(p<0.001), and the proximal portion of aneurysmal neck has located below the interclinoid line(p=0.001). Conclusion : Through a careful preoperative evaluation, some radiometric parameters can be used to determine whether the ACP should be removed in clipping of the low-lying IC-PC aneurysms. Unlike to total removal of the ACP, the intradural partial anterior clinolidectomy and/or APF resection, which are more familiar to surgeons, reduce the risks of the premature rupture, operative time, and also contribute a more precise clip placement with proximal control than the extradural clinoidectomy.

• Journal of the Korean Association of Oral and Maxillofacial Surgeons
• /
• v.33 no.4
• /
• pp.307-311
• /
• 2007

Purpose: The aim of this study was simply assessing linear measurements in the lateral and medial approach, respectively, for bone harvesting using anatomic and three-dimensional(3D) computed tomographic(CT) analyses on a dried cadaveric proximal tibia. In addition, the availability of the three-dimensional computed tomographic(3D-CT) analysis was also estimated. Materials and methods: Ten dried proximal tibia were obtained from five Korean cadavers. Four the reference points, the SM(superior-medial), IM(inferior-medial), SL(superior-lateral), and IL(inferior-lateral) were marked around the tibial tuberosity. The PM(posterior-medial) and PL(posterior-lateral) points were randomly marked at points farthest from the lateral and medial reference points, respectively, in the posterior border of the superior articular surface of both condyles. All measurements were obtained on the dried proximal tibia. After computed tomography had been performed, the three dimensional images were reconstructed using V works $4.0^{TM}$(Cybermed Inc., Seoul, Korea), and the length between the reference points were measured three dimensionally using the method described above. The error between the mean actual and mean 3D-CT measurements was calculated in order to determine the availability of the three dimensional computed tomographic analysis. Results: The length between the reference points was greatest at the IL-PM, which averaged $65.39mm{\pm}10.35$. This was followed by the SL-PM with $63.24mm{\pm}8.10$, the IM-PL with $58.09mm{\pm}10.02$, and the SM-PL with $51.99mm{\pm}9.06$. The differences between the IL-PM and SM-PL were 13.4 mm. The mean values were 55.04 mm in the medial approach and 64.32 mm in the lateral approach, and the differences between medial and lateral were 9.28 mm. The error between the mean actual and mean 3D-CT measurements was 0.31% and the standard deviation was 0.28%. Conclusion: The anatomical and three dimensional computed tomographic analysis indicates that there was only a 9.28 mm linear difference between the lateral and medial approach. This is consistent with previous studies, which showed that there was little difference between the two approaches in terms of the bone volume. In addition, the error(0.31%) and the standard deviation(0.28%) were considered low, demonstrating high accuracy of 3D-CT. Therefore it can be used in preoperative treatment planning.