http://repository.sungshin.ac.kr/handle/2025.oak/158 2026-04-02T11:18:40Z 2026-04-02T11:18:40Z 강대성 윤태영 http://repository.sungshin.ac.kr/handle/2025.oak/8854 2025-12-29T00:41:21Z 2025-07-08T15:00:00Z

Title: CurriMAE: curriculum learning based masked autoencoders for multi-labeled pediatric thoracic disease classification Author(s): 강대성; 윤태영 Abstract: Masked autoencoders (MAE) have emerged as a powerful framework for self-supervised learning by reconstructing masked input data. However, determining the optimal masking ratio requires extensive experimentation, resulting in significant computational overhead. To address this challenge, we propose CurriMAE, a curriculum-based training approach that progressively increases the masking ratio during pretraining to balance task complexity and computational efficiency. In CurriMAE, the training process spans 800 epochs, with the masking ratio gradually increasing in four stages: 60% for the first 200 epochs, followed by 70%, 80%, and finally 90% in the last 200 epochs. This progressive masking approach, inspired by curriculum learning, allows the model to learn from simpler tasks before tackling more challenging ones. To ensure stable convergence, a cyclic cosine learning rate scheduler is employed, resetting every 200 epochs, effectively dividing the training process into four distinct stages. At the end of each stage, corresponding to one complete cycle of the learning rate schedule, a snapshot model is saved, resulting in four pretrained models. These snapshots are then fine-tuned to obtain the final classification results. We evaluate CurriMAE on multi-labeled pediatric thoracic disease classification, pretrainin

2025-07-08T15:00:00Z 강대성 윤태영 http://repository.sungshin.ac.kr/handle/2025.oak/8847 2025-12-29T00:41:13Z 2025-07-07T15:00:00Z

Title: Efficient pretraining of ECG scalogram images using masked autoencoders for cardiovascular disease diagnosis Author(s): 강대성; 윤태영 Abstract: Cardiovascular diseases (CVDs) are the leading cause of mortality worldwide, emphasizing the need for accurate and early diagnosis. Electrocardiograms (ECG) provide a non-invasive means of diagnosing various cardiac conditions. However, traditional methods of interpreting ECG signals require substantial expertise and time, motivating the development of automated deep learning models to enhance diagnostic precision. This study proposes a novel approach that leverages masked autoencoders (MAE) to pretrain a model on ECG scalogram images, thereby enhancing the diagnostic accuracy for seven CVDs. Through extensive experimentation, we demonstrated that pretraining with an 85% masking ratio over 500 epochs yields optimal results. The pretrained ViT-S(MAE-scalo) network demonstrated remarkable performance in detecting CVDs, achieving an AUC of 0.986 and 92.43% accuracy in Lead II. Furthermore, the ensemble learning approach applied across 12 ECG leads enhanced the model’s diagnostic capabilities, resulting in an AUC of 0.994 and 92.72% accuracy. The MAE-based models outperformed traditional models such as ResNet-34 and ViT-S pretrained on ImageNet or random weights, as well as other SSL models such as MoCo-v2 and BYOL. Notably, the MAE-based models demonstrated superior performance even with a significantly smaller dat

2025-07-07T15:00:00Z 강태욱 Boyoung Jeon Sung Wook Choi http://repository.sungshin.ac.kr/handle/2025.oak/8769 2025-12-29T00:39:56Z 2025-05-20T15:00:00Z

Title: Lifestyle factors and health outcomes associated with infertility in women: A case-control study using National Health Insurance Database Author(s): 강태욱; Boyoung Jeon; Sung Wook Choi Abstract: BackgroundApproximately one in six people is experiencing infertility at some point in their lives. In response, health insurance coverage for infertility treatments has been strengthened. However, studies examining lifestyle factors that affect infertility remain lacking, highlighting the need to generate objective evidence to address infertility issues using national-level datasets.MethodsThe General Healthcare Screening Program dataset from National Health Insurance Service database was employed in this study to examine infertility and childbirth among women aged 22-49 years. In 2020, 25,333 women with infertility and 73,759 women who had given birth were initially identified. After applying propensity score matching for age, Charlson Comorbidity Index score, and income level, the final study population included 24,325 women with infertility and 24,325 women who with childbirth. Employing a case-control study design, lifestyle factors (drinking, smoking, and physical activity) and health checkup outcomes (underweight, overweight, hypertension, diabetes, kidney function, anemia, and menstrual disorders) were assessed in this study. Statistical analyses included chi-squared tests, t-tests, and logistic regression.ResultsThis study revealed significant risk factors for infertility: two high-r

2025-05-20T15:00:00Z 강태욱 Jung-Won SHIN Heather SWAN Kyungho PAIK Chang-Hun HUH Hyun Jung KIM http://repository.sungshin.ac.kr/handle/2025.oak/8711 2025-12-29T00:39:14Z 2025-01-07T15:00:00Z

Title: Birth Outcomes of Infants Born to Mothers with Alopecia Areata: A Nationwide Population-based Study in Korea Author(s): 강태욱; Jung-Won SHIN; Heather SWAN; Kyungho PAIK; Chang-Hun HUH; Hyun Jung KIM Abstract: Data on pregnancy outcomes in patients with alopecia areata (AA) are limited. The aim of this study is to determine the association between maternal AA and risk of adverse birth outcomes in children. A retrospective cohort study was conducted on 45,328 children born to mothers with AA and 4,703,253 controls born to mothers without AA using the Korean National Health Insurance Claims database from 2002 to 2016. Multivariate logistic regression analyses were performed to evaluate the association between maternal AA and the birth outcomes of their children. Infants born to mothers with AA exhibited a significantly higher risk of preterm birth (odds ratio [OR] 1.39, 95% CI 1.33-1.45; adjusted OR [aOR] 1.07, 95% CI 1.01-1.13), low birthweight (OR 1.36, 95% CI 1.30-1.42; aOR 1.11, 95% CI 1.05-1.17), and Caesarean section birth (OR 1.24, 95% CI 1.22-1.26; aOR 1.12, 95% CI 1.08-1.15) than controls. In addition, the risk of congenital malformations was also significantly higher in infants born to mothers with AA (OR 1.19, 95% CI 1.15-1.22; aOR 1.10, 95% CI 1.07-1.14), especially for malformations of the urinary (OR 1.33, 95% CI 1.19-1.48; aOR 1.16, 95% CI 1.04-1.29) and musculoskeletal (OR 1.19, 95% CI 1.12-1.27; aOR 1.12, 95% CI 1.05-1.19) systems, than controls. Maternal AA is associated with an increased risk of adver

2025-01-07T15:00:00Z