Soonchunhyang University

Private set intersection (PSI) is a privacy‐preserving scheme that computes the intersection of two datasets without leaking any other information. Additionally, there is multiparty private set intersection (MPSI) to extend the number of parties for computing the intersection of multiple private datasets. In the traditional PSI and MPSI studies, protocol parties input their private datasets, and one or all of them can compute the intersection. However, there are some scenarios where an inputless external party requires the intersection between private datasets of other parties. Thus, the external party PSI protocols have been recently studied for applications such as pandemic contact tracing, computing human genome information and evaluating policy effects. However, they are limited in applications because the external party can compute the intersection of two datasets. In this paper, we propose a new external party compute‐MPSI (EPC‐MPSI) protocols that allow an external party to compute the intersection of multiple datasets. We provide the extension of the number of parties and solve the limitation of prior external party PSI protocols. In addition, we analyze the correctness, security and the efficiency in terms of communication and computation costs compared to the prior traditional MPSI protocols.

This study investigated the effects of rice protein (RP) addition on the retrogradation of waxy rice starch (WRS) gels during storage. The retrogradation behavior was evaluated using a rapid visco analyser, a differential scanning calorimeter, an X-ray diffractometer, a texture analyser, and a scanning electron microscope. The incorporation of RP inhibited the swelling of WRS, leading to reductions in trough viscosities, as well as setback values. DSC analysis indicated a reduction in melting enthalpy, while XRD showed decreased peak intensity in retrograded WRS with RP addition. After 3 days of storage at 4 °C, gel hardness increased, but RP significantly mitigated this effect. SEM observations showed that a 5% RP addition produced larger pore sizes and a looser network structure, supporting the inhibition of retrogradation. These findings suggest that RP effectively retards the retrogradation of WRS, offering potential applications in enhancing the texture and shelf-life of starch-based food products.

Background and Aims Colonic stenting using self‐expandable metallic stents (SEMS) as a bridge to surgery offers an effective alternative to emergency surgery for the management of malignant colorectal obstruction. However, the optimal timing of elective surgery after stenting remains controversial. Methods This retrospective multicenter cohort study analyzed 380 patients with obstructive colorectal cancer who were treated with SEMS as a bridge to surgery. Patients were categorized into four groups based on the time from stent insertion to surgery: within 7 days, 8–14 days, 15–21 days, and 22 days or more. Results The study cohort had a slight male predominance (55.8%), with an average age of 65.8 years. Most surgeries (74.2%) were laparoscopically performed. No significant differences were observed in stoma formation rates or postoperative complications between the different timing groups. Similarly, recurrence‐free survival, overall survival, locoregional recurrence, and distant metastasis rates showed no significant variations with the timing of post‐stenting surgery. A restricted cubic spline curve indicated that surgery within the 15–21‐day period post‐SEMS insertion resulted in the lowest incidence of stoma formation. Conclusions Delaying elective surgery for up to 3 weeks post‐SEMS placement for obstructive colorectal cancer is recommended, particularly within the 15–21‐day period, to minimize stoma formation rates without compromising on long‐term outcomes.

Background This study aimed to assess prognostic significance of FDG PET/CT parameters in predicting progression-free survival (PFS) and overall survival (OS) in patients with hepatocellular carcinoma (HCC) treated with atezolizumab plus bevacizumab therapy. Patients and Methods We retrospectively enrolled 78 patients with HCC who underwent FDG PET/CT before atezolizumab plus bevacizumab therapy and identified intrahepatic target tumor lesions on pretreatment imaging studies. From PET/CT images, we measured SUVmax, tumor-to-normal liver uptake ratio, metabolic tumor volume, and total lesion glycolysis (TLG) for intrahepatic tumor lesions, as well as SUVmax for extrahepatic metastatic lesions (extrahepatic SUVmax). Results In comparisons of PET/CT parameters, patients with progressive disease demonstrated significantly higher TLG values than those achieving complete or partial response ( P < 0.05). In the multivariate survival analysis, TLG independently predicted both PFS ( P = 0.019) and OS ( P = 0.003). Metabolic tumor volume was significantly associated with OS alone ( P = 0.010), and extrahepatic SUVmax was significantly associated with only PFS ( P = 0.045). Patients with high TLG values experienced poorer PFS and OS than those with low TLG values ( P < 0.05). Conclusions TLG in intrahepatic HCC lesions was significantly associated with treatment response and served as an independent prognostic factor for PFS and OS. TLG could be a potential imaging biomarker for predicting clinical outcomes in patients with HCC receiving atezolizumab plus bevacizumab therapy.

Background Ankle sprains are the most frequent musculoskeletal injury in sports. Patients reporting pain at the lateral malleolus tip following ankle sprains or sports activities frequently have separated ossicles, referred to as an os subfibulare (OSF). Commonly, small ossicles accompanied by chronic lateral ankle instability (CLAI) are treated with ossicle resection combined with the modified Broström operation (MBO). We compared the clinical and radiological results between groups in which a small OSF was or was not removed. Methods We retrospectively enrolled all patients with a small OSF who underwent arthroscopic MBO by one surgeon in one institution between 2015 and 2022. The study included skeletally mature patients who had an OSF among those who had MBO surgery and follow-up for at least 1 year. An ossicle was defined as small if the longitudinal diameter was < 5 mm on an anteroposterior plain radiograph. Results There were no significant differences between the groups preoperatively or 6 or 12 months postoperatively. The radiographic findings did not differ significantly between groups. Conclusions When performing arthroscopic MBO on patients with CLAI, OSF ≤ 5 mm removal did not alter clinical or radiological outcomes, suggesting that excision may not be needed in asymptomatic patients. Considering the risks of the removal process, leaving it alone may be a treatment option. Clinical trial number Not applicable.

Cognitive decline is a common issue in Parkinson’s disease (PD) and significantly affects patients’ quality of life. This study explored the relationship between cognitive functions and dysautonomia in de novo PD. We reviewed records of newly diagnosed PD patients from July 2017 to September 2023 who underwent cognitive and autonomic assessments. Cognitive functions were measured using the Korean version of the Montreal Cognitive Assessment (MoCA-K) and the Seoul Neuropsychological Screening Battery, while autonomic functions were evaluated with the SCOPA-AUT questionnaire. Among 155 patients, 82 with de novo PD were included. The mild cognitive impairment (MCI) group exhibited higher SCOPA-AUT scores, particularly in gastrointestinal dysfunction. Multivariable logistic regression identified total SCOPA-AUT scores as significant predictors of MCI, even after adjusting for demographic and clinical factors. Partial correlation analysis showed significant negative associations between SCOPA-AUT scores and cognitive functions, such as memory and executive function. This study highlights a strong link between autonomic dysfunction, including gastrointestinal issues, and cognitive impairment in de novo PD. Monitoring dysautonomia in early-stage PD may aid in identifying patients at risk of cognitive decline.

Recommendation systems (RSs) are crucial in personalizing user experiences in digital environments by suggesting relevant content or items. Collaborative filtering (CF) is a widely used personalization technique that leverages user-item interactions to generate recommendations. However, it struggles with challenges like the cold-start problem, scalability issues, and data sparsity. To address these limitations, we develop a Graph Convolutional Networks (GCNs) model that captures the complex network of interactions between users and items, identifying subtle patterns that traditional methods may overlook. We integrate this GCNs model into a federated learning (FL) framework, enabling the model to learn from decentralized datasets. This not only significantly enhances user privacy— a significant improvement over conventional models but also reassures users about the safety of their data. Additionally, by securely incorporating demographic information, our approach further personalizes recommendations and mitigates the cold-start issue without compromising user data. We validate our RSs model using the open MovieLens dataset and evaluate its performance across six key metrics: Precision, Recall, Area Under the Receiver Operating Characteristic Curve (ROC-AUC), F1 Score, Normalized Discounted Cumulative Gain (NDCG), and Mean Reciprocal Rank (MRR). The experimental results demonstrate significant enhancements in recommendation quality, underscoring that combining GCNs with CF in a federated setting provides a transformative solution for advanced recommendation systems.

Recently, Network Functions Virtualization (NFV) has become a critical resource for optimizing capability utilization in the 5G/B5G era. NFV decomposes the network resource paradigm, demonstrating the efficient utilization of Network Functions (NFs) to enable configurable service priorities and resource demands. Telecommunications Service Providers (TSPs) face challenges in network utilization, as the vast amounts of data generated by the Internet of Things (IoT) overwhelm existing infrastructures. IoT applications, which generate massive volumes of diverse data and require real-time communication, contribute to bottlenecks and congestion. In this context, Multi-access Edge Computing (MEC) is employed to support resource and priority-aware IoT applications by implementing Virtual Network Function (VNF) sequences within Service Function Chaining (SFC). This paper proposes the use of Deep Reinforcement Learning (DRL) combined with Graph Neural Networks (GNN) to enhance network processing, performance, and resource pooling capabilities. GNN facilitates feature extraction through Message-Passing Neural Network (MPNN) mechanisms. Together with DRL, Deep Q-Networks (DQN) are utilized to dynamically allocate resources based on IoT network priorities and demands. Our focus is on minimizing delay times for VNF instance execution, ensuring effective resource placement, and allocation in SFC deployments, offering flexibility to adapt to real-time changes in priority and workload. Simulation results demonstrate that our proposed scheme outperforms reference models in terms of reward, delay, delivery, service drop ratios, and average completion ratios, proving its potential for IoT applications.

Postoperative delirium (POD) is a frequent complication in older people undergoing general anesthesia surgery. We investigated the potential link between Alzheimer’s disease and POD by comparing plasma amyloid-beta oligomer levels (measured using the multimer detection system, MDS-OAβ) in patients who developed POD after general anesthesia surgery with those who did not. A total of 104 eligible participants were screened daily for delirium for three days postoperatively. After propensity score matching based on the ApoE4 allele, the final analysis included 31 patients with POD and 31 without POD. In the ICU, patients with delirium underwent immediate assessment with the Korean version of the Delirium Rating Scale-98 (K-DRS-98) and plasma MDS-OAβ levels. The control group (those without POD) received the same tests on the third postoperative day. Patients with POD had significantly higher MDS-OAβ values compared to those without POD. Within the POD group, MDS-OAβ values positively correlated with K-DRS-98 scores (both severity and total scores). These findings suggest an association between POD in older people undergoing general anesthesia surgery and elevated plasma amyloid oligomer levels. To definitively establish causality, further prospective studies are necessary.

Alzheimer’s disease (AD) and Parkinson’s disease (PD) are representative neurodegenerative diseases with abnormal energy metabolism and altered distribution and deformation of mitochondria within neurons, particularly in brain regions such as the hippocampus and substantia nigra. Neurons have high energy demands; thus, maintaining a healthy mitochondrial population is important for their biological function. Recently, exosomes have been reported to have mitochondrial regulatory potential and antineurodegenerative properties. This review presents the mitochondrial abnormalities in brain cells associated with AD and PD and the potential of exosomes to treat these diseases. Specifically, it recapitulates research on the molecular mechanisms whereby exosomes regulate mitochondrial biogenesis, fusion/fission dynamics, mitochondrial transport, and mitophagy. Furthermore, this review discusses exosome-triggered signaling pathways that regulate nuclear factor (erythroid-derived 2)-like 2-dependent mitochondrial antioxidation and hypoxia inducible factor 1α-dependent metabolic reprogramming. In summary, this review aims to provide a profound understanding of the regulatory effect of exosomes on mitochondrial function in neurons and to propose exosome-mediated mitochondrial regulation as a promising strategy for AD and PD.

Walking speed (WS), the 'sixth vital sign,' is critical for health assessment. However, its estimation using only insole pressure sensors (IPS) remains a challenge. We introduce a deep learning framework combining multi-modal ResNet-50 with multi-output regression (MOR) for estimating WS, stride length, and step length. In experiments with 25 healthy participants walking at self-selected speeds, the method achieved high accuracy (RMSE: 3.93 cm/s, adjusted R²: 0.89), outperforming previous studies. Ablation studies demonstrated that removing either the multi-modal architecture or MOR significantly reduced performance, underscoring the critical role of both components in enhancing the model's accuracy. Feature importance analysis revealed that mean midfoot center of pressure (CoP) velocity and second double support (SDS) time are pivotal to WS estimation, while participant-specific variables (shoe size, height) also contribute substantially. Key factors included mean midfoot CoP velocity and SDS time, highlighting potential for integrating contralateral data and additional biomechanical parameters (e.g., leg length) in future refinements. Overall, these findings demonstrate that WS can be robustly estimated using only IPS, avoiding the need for extra sensors.