Youngmi Lee

Cheonggukjang (CGJ) exerts anti-obesity and lipid-lowering effects; however, its effects on gut microbiota-derived short-chain fatty acid (SCFA) production, adipose G protein-coupled receptor (GPCR) expression, and the mechanistic associations between them have not been investigated. Thus, the current study aimed to investigate these relationships in relation to lipid metabolism in white adipose tissue (WAT). Two distinct CGJ samples were administered to high-fat/high-cholesterol diet-induced (HCFD) obese mice. CGJ slightly lowered body weight gain and significantly improved dyslipidemia and hepatic lipid accumulation, also improving WAT lipid metabolism-related gene expression. Additionally, CGJ reversed gut microbiota dysbiosis, increased the abundance of genera associated with SCFA production, and increased colonic SCFA levels. Adipose Gpr41 and Gpr109a expressions were upregulated, and their mRNA levels were strongly correlated with systemic lipid indicators and WAT lipid-metabolism genes. To the best of our knowledge, this is the first study to provide evidence that CGJ modulates gut microbiota-derived SCFA levels and elevates adipose GPCR gene expression, suggesting that CGJ may exert anti-obesity and lipid-metabolism-alleviating effects through the gut microbiota–SCFA–adipose GPCR–WAT lipid metabolism axis.

Transfer learning has emerged as a powerful technique in machine learning, wherein knowledge acquired from one task or dataset is leveraged to improve performance on related tasks and datasets. Although univariate time series forecasting plays a pivotal role in numerous real-world applications, the adoption of transfer learning in this domain has been relatively underexplored. This paper proposes a new transfer learning approach for univariate time series forecasting, combining the Neural Basis Expansion Analysis for interpretable Time Series forecasting (N-BEATS) and Multi-Layer Perceptron (MLP). Furthermore, the Boltzmann-weighted ensemble is proposed for prediction, serving as an advanced inference method that improves forecasting accuracy without incurring significant computational overhead. We constructed a pre-trained model for knowledge transfer. Then, we fine-tuned it to predict time series data from seven domains (temperature, solar radiation, electricity demand, air pollution, wind speed, traffic volume, and coin price). Our experimental results show that transfer learning enhances predictive performance, and investigations into additional layers reveal that MLP architectures are superior to long short-term memory and original transformer architectures in terms of Root Mean Square Error (RMSE), whereas the advanced time-series transformer PatchTST shows potential in transfer learning. Additionally, we conducted various experiments for qualitative analyses. The results of experiments consistently demonstrate the superior performance of transfer learning for univariate time series forecasting, offering valuable insights into designing more robust deep learning solutions in data-driven environments.

With the widespread adoption of computer vision technology in agriculture, securing high-quality training data has become essential. However, existing data construction methods are limited by the substantial time and cost required. In this work, we develop a semi-automatic annotation system that integrates SAM3 (Segment Anything Model 3), a state-of-the-art multimodal foundation model. SAM3(Segment Anything Model 3). The proposed system is implemented in a GUI environment and consists of three stages: (1) object recognition based on text prompts, (2) generation of precise masks and their conversion into trainable polygon coordinates, and (3) quality assurance through user verification. In performance evaluations using 600 images, SAM3 achieved a matching rate of 92.9% and a mean Average Precision (mAP) of 0.790, reducing dataset construction time by 96-98% compared to manual annotation. These results demonstrate superior performance in both accuracy and efficiency compared to baseline foundation models such as SAM+CLIP and Grounding DINO+SAM. This study highlights how the zero-shot capabilities of foundation models can drastically improve agricultural data labeling efficiency and accelerate related AI research.

Background/Objectives: To develop a DL-based model predicting recurrence risk in HER2-low breast cancer patients and to compare performance of the MRI-alone, clinicopathologic-alone, and combined models. Methods: We analyzed 453 patients with HER2-low breast cancer who underwent surgery and preoperative breast MRI between May 2018 and April 2022. Patients were randomly assigned to either a training cohort (n = 331) or a test cohort (n = 122). Imaging features were extracted from DCE-MRI and ADC maps, with regions of interest manually annotated by radiologists. Clinicopathological features included tumor size, nodal status, histological grade, and hormone receptor status. Three DL prediction models were developed: a CNN-based MRI-alone model, a clinicopathologic-alone model based on a multi-layer perceptron (MLP) and a combined model integrating CNN-extracted MRI features with clinicopathological data via MLP. Model performance was evaluated using AUC, sensitivity, specificity, and F1-score. Results: The MRI-alone model achieved an AUC of 0.69 (95% CI, 0.68–0.69), with a sensitivity of 37.6% (95% CI, 35.7–39.4), specificity of 87.5% (95% CI, 86.9–88.2), and F1-score of 0.34 (95% CI, 0.33–0.35). The clinicopathologic-alone model yielded the highest AUC of 0.92 (95% CI, 0.92–0.92) and sensitivity of 93.6% (95% CI, 93.4–93.8), but showed the lowest specificity (72.3%, 95% CI, 71.8–72.8) and F1-score of 0.50 (95% CI, 0.49–0.50). The combined model demonstrated the most balanced performance, achieving an AUC of 0.90 (95% CI, 0.89–0.91), sensitivity of 80.0% (95% CI, 78.7–81.3), specificity of 83.2% (95% CI: 82.7–83.6), and the highest F1-score of 0.55 (95% CI, 0.54–0.57). Conclusions: The DL-based model combining MRI and clinicopathological features showed superior performance in predicting recurrence in HER2-low breast cancer. This multimodal approach offers a framework for individualized risk assessment and may aid in refining follow-up strategies.

Minwoo Lee, Youngmi Lee. Korean J Appl Stat 2025;38:265-79. https://doi.org/10.5351/KJAS.2025.38.2.265

South Korea faces serious challenges regarding population imbalance and sustainability due to low birth rates and the aging population. This study utilizes future population projection scenarios provided by Statistics Korea to estimate population distributions at both regional and grid levels. The analysis applies an urban growth model for administrative divisions and a modified gravity model for grid-level estimations. The modified gravity model enhances the prediction accuracy by effectively accounting for multi-centered urban structures and excluding non-residential areas such as mountains, rivers, and parks. Additionally, a novel spatial weight matrix considering train station connectivity is introduced. The results show that incorporating public transportation infrastructure around cities slows the rate of population decline, highlighting its mitigating effects on regional extinction. The study predicts that over 30 cities will face depopulation risks by 2072, while population concentration in the Seoul metropolitan area will persist. The grid-level analysis reveals detailed patterns of population imbalance within regions, particularly identifying uninhabited areas and their spatial distribution. These findings carry significant implications for infrastructure planning and regional development. By employing innovative modeling approaches and high-resolution projections, this study provides policymakers with valuable insights into South Korea’s demographic challenges and potential advances in sustainable urban and regional development.