Hyundai Engineering (South Korea)

In retail lending, offering preferential interest rates is a core marketing instrument for balancing customer acquisition with portfolio profitability. Accurately predicting the effect of interest-rate discounts for each customer is pivotal for optimizing the discount strategy: offering overly generous discounts erodes margins, while insufficient discounts drive price-sensitive customers to defect. Off‑the‑shelf machine learning uplift models rarely respect the complex operational constraints of financial business, such as tiered rate grids, regulatory guard‑rails, and marketing budget ceilings. We propose an integrated system that fuses causal inference and domain adaptation to produce constraint‑aware, customer‑specific discount recommendations. To further enhance practitioner adoption, a large language model layer translates model outputs into actionable narratives. Developed in Hyundai Capital Services, the system boosted transaction volume by 13%, demonstrating both technical soundness and material business impact.

High-skilled esports players have unique cortical activity and gaze control, both of which are closely linked to their performance. To enhance key aspects of esports performance—shot time and accuracy—the current study implemented two training methods: a cortical activity-based method (Experiment 1) and a gaze control-based method (Experiment 2). In Experiment 1, a three-channel electroencephalogram headset was used to provide participants with real-time feedback on central lobe activity associated with sustained attention. Twenty-one participants were randomly assigned to the training group (10 participants) and sham group (11 participants). The training group received auditory feedback based on the power of theta and alpha power, while the sham group received auditory feedback not related to central lobe activity. In Experiment 2, an eye-tracker was used to measure the gaze movement and provide feedback. The feedback was provided to help participants maintain their gaze on-center. Twenty-one participants were randomly assigned to the training group (11 participants) and the sham group (10 participants). The training group received warning sounds when their gaze deviated from the off-center, promoting centralized gaze control, whereas the sham group received no such feedback. Participants‘ age and gaming experience were controlled to minimize confounding factors. The results demonstrated a significant decrease in shot time for the training group in both Experiment 1 (Mdn = -30.3ms) and Experiment 2 (Mdn = -47.2ms), with no significant changes in the sham group (all p > .05). Therefore, biofeedback training methods can reduce shot time, one of the critical performance indicators in esports. • We implement attention training and peripheral vision methods for enhancing the reaction speed and accuracy of esports players. • Attention training significantly reduced shot times (Mdn: -30.3ms) in esports players by improving their attention, and the result was supported by a reduction in central alpha power. • Peripheral vision training significantly reduced shot times (Mdn: -47.2ms) in esports players by enhancing peripheral vision skills, and this finding was supported by a narrower gaze distribution. • This reduction in shot time was achieved without a significant reduction in accuracy.

PurposeThis retrospective case study explored the effect of contralateral acupuncture (Geoja technique) on patients with one-sided musculoskeletal pain disorders.Patients and MethodsThis retrospective observational study reviewed 109 patients treated for unilateral musculoskeletal pain. Their demographics and clinical characteristics were recorded. Patients were categorized into two groups based on pulse strength differences, diagnosed using a classic Korean medicine method rooted in Donguibogam. Disability and pain intensity were measured using the Numeric Rating Scale (NRS) and Range of Motion (ROM). The treatment effect was assessed by comparing the improvement rates between these groups.ResultsAll 109 patients received contralateral acupuncture treatment. Significant improvements in disability and pain intensity were observed in the 65 patients with pulse strength differences. The mean differences and improvement rate of NRS showed clinically relevant changes in both groups; however, the improvement rate in the group with pulse strength differences was statistically significantly higher than in the 44 patients without pulse strength differences.ConclusionThis study suggests the 2, potential of contralateral acupuncture for pain relief, ROM improvement, and maintaining normal function in patients with musculoskeletal diseases. It also highlights the importance of pulse strength difference for effective treatment. Further studies are required to ensure appropriate patient selection, integrate treatment guidelines, and ascertain optimal outcome measurement criteria to draw robust conclusions regarding the effectiveness of treating musculoskeletal disorders.

This paper proposes a sensorless phase synchronization method for single-phase grid-connected inverters that avoids the requirement for a voltage sensor. The method models the actual grid impedance as a virtual impedance and represents the grid voltage in the corresponding synchronous reference frame. To construct the synchronous voltage, a 90-degree phaseshifted virtual voltage is then generated, which is used to estimate the grid phase and frequency. The estimated voltage subsequently undergoes processing through a low-pass filter (LPF) and a disturbance observer to extract the final phase and frequency components. The proposed method features low computational complexity and is amenable to real-time implementation on low-cost microcontrollers. It also exhibits robustness against grid-impedance mismatch and harmonic distortion, maintaining stable phase and frequency estimation. The effectiveness of the proposed approach was demonstrated through both simulation and experimental results.

This paper discusses the design and performance analysis of a permanent magnet synchronous machine (PMSM) for low-speed, high-torque shaft generators. A 216-slot and 60pole $(216 \mathrm{~S} 60 \mathrm{P})$ combination was selected, based on a 36S10P design with a coil span of three, which allows for a modular structure of both windings and cores. A comprehensive 2-D finite element analysis was conducted under rated conditions to evaluate the effects of segmented stator and rotor structures on average torque, torque ripple, core losses, and eddy current loss of the permanent magnet. The study revealed that an optimized segmentation of the rotor yoke with a non-magnetic rotor rim led to a substantial reduction in subharmonic fluxes, resulting in a significant decrease in the eddy current loss by as much as 80%. The findings open new possibilities for the design and application of modular PMSMs, promising improved efficiency and maintenance convenience in large-scale marine applications.

This paper presents the design optimization of a permanent magnet shaft generator (PMSG) to improve the overall shaft generator system efficiency for large vessels. The study focuses on selecting appropriate system constraints and determining the optimal design parameters of the PMSG, with consideration for both cost-effectiveness and operational system-level efficiency. To validate the proposed design, a prototype was manufactured and tested with a variable frequency drive under realistic operating conditions. The results indicate that the optimized shaft generator system achieves up to a 0.79% increase in total system efficiency compared to a conventional configuration, demonstrating the potential advantages of applying advanced electrical machine optimization to marine propulsion systems.

Carotid atherosclerosis is a key predictor of cardiovascular disease (CVD), necessitating early detection. While foundation models (FMs) show promise in medical imaging, their optimal selection and fine-tuning strategies for classifying carotid atherosclerosis from retinal images remain unclear. Using data from 39,620 individuals, we evaluated four vision FMs with three fine-tuning methods. Performance was evaluated by predictive performance, clinical utility by survival analysis for future CVD mortality, and explainability by Grad-CAM with vessel segmentation. DINOv2 with low-rank adaptation showed the best overall performance (area under the receiver operating characteristic curve = 0.71; sensitivity = 0.87; specificity = 0.44), prognostic relevance (hazard ratio = 2.20, P-trend < 0.05), and vascular alignment. While further external validation on a broader clinical context is necessary to improve the model's generalizability, these findings support the feasibility of opportunistic atherosclerosis and CVD screening using retinal imaging and highlight the importance of a multi-dimensional evaluation framework for optimal FM selection in medical artificial intelligence.

Digital twin (DT) technology, integrated with building information modeling (BIM), enables real-time feedback and predictive analytics in construction. This study presents a BIM-enabled DT framework to optimize in situ production and yard-stock management of precast concrete (PC) components with a focus on minimizing CO2 emissions. Using Oracle Crystal Ball, scenario-based simulations revealed up to an 8.9% reduction in environmental impact. Distinct from prior research that largely emphasized cost or off-site strategies, this study uniquely addresses on-site sustainability by embedding carbon metrics into the decision-making process. The framework was validated through a large-scale logistics warehouse project that showcased its practical utility. This research contributes a replicable method for enhancing sustainability in precast construction through digital technologies.