Background: Real-time ergonomic risk assessment in manufacturing environments is challenged by severe class imbalance in high-risk postures and the need for deployment-efficient models. Conventional oversampling techniques may violate biomechanical constraints, limiting their suitability for human motion data. Objectives: This study aimed to compare multiple machine learning models for real-time ergonomic risk assessment while addressing data imbalance using biomechanically appropriate learning strategies and evaluating both predictive performance and deployment efficiency. Design: Comparative study. Methods: A large-scale workplace safety dataset comprising image-based skeletal keypoints was analyzed. To mitigate class imbalance without generating biomechanically implausible samples, cost-sensitive learning and focal loss were employed instead of synthetic oversampling. Subject-wise data splitting was applied to prevent data leakage. Five model families, including Random Forest, convolutional neural networks, and a lightweight graphbased network, were evaluated using accuracy, F1-score, area under the receiver operating characteristic curve (AUC), and high-risk recall. Statistical significance was assessed using bootstrap confidence intervals and McNemar and DeLong tests. Results: The lightweight graph-based model demonstrated competitive classification performance while maintaining reduced computational complexity. Although none of the models achieved the predefined high-risk recall threshold, statistically significant performance differences were observed across model families. Conclusion: The findings suggest that biomechanically informed imbalance handling improves methodological validity in ergonomic risk assessment. While deployment feasibility appears promising, further empirical validation on edge hardware is required.

This case report describes the successful management of a recurrent eyelid tumor in a 10-year-old, neutered male Labrador Retriever using surgical debulking combined with cryotherapy. The mass was located on the margin of the left upper eyelid, near the medial canthus, and had recurred two years after a previous excision. Due to the patient's history of nephrectomy and the owner's concerns regarding general anesthesia, the procedure was performed under light sedation. After debulking of the tumor, two freeze-thaw cycles (25-second freeze, 40-second thaw) were applied to the remaining tumor bed. Histopathological examination confirmed a meibomian gland epithelioma. The surgical site healed well with only mild depigmentation, and no recurrence was observed during the 8-month follow-up period. This combined approach represents an effective, minimally invasive option for treating canine eyelid tumors, particularly in high-risk patients.

Background: Due to the variety of etiological factors in chronic low back pain (CLBP), there is significant variability in functional measurements. Objects: This study aimed to determine if using metrics in addition to inferential statistics could change how the impact of poor prognosis risk for pain among volunteers with CLBP is interpreted. Methods: In this cross-sectional observational study, 74 adult volunteers were allocated to four groups: a pain-free control group (CG) and three CLBP groups stratified by the STarT Back Screening Tool into low (LR), medium (MR) and high risk (HR). Spatiotemporal gait parameters outcomes were self-selected walking speed (SWS), optimum walking speed (OWS) and the locomotor rehabilitation index (LRI). Data were analyzed using a generalized estimating equation model. Reproducibility, responsiveness (minimum detectable change [MDC]) and effect sizes were also computed. Results: No differences were found for OWS. SWS and LRI were significantly higher in CG than in all CLBP groups, but observed differences did not exceed MDC, indicating they are likely to reflect measurement error. Nevertheless, large effect sizes suggest these reductions in SWS and LRI are clinically meaningful. Comparisons among the LR, MR, and HR groups revealed no significant differences or meaningful effect sizes. Conclusion: Combining complementary metrics with inferential statistics confirms that individuals with CLBP walk more slowly and exhibit lower LRI than pain-free controls, while prognostic risk strata do not influence these spatiotemporal gait parameters.

Veterinary drugs can remain as residues in animal-derived food products, and therefore, many countries conduct residue monitoring programs for imported livestock products. However, because the types and authorizations of veterinary drugs vary among countries, it is necessary for importing nations to establish residue monitoring systems tailored to their specific circumstances. This study aimed to develop an algorithm to quantitatively evaluate and score the risk of veterinary drug residues that may be present in imported livestock products, thereby enabling risk-based prioritization. The overall risk score was calculated as the product of exposure and toxicity factors. To minimize uncertainty, the algorithm utilized objective and accessible data obtained from both domestic and international sources. The exposure factor was determined using the number of residue violations and the estimated exposure value, which was calculated based on withdrawal periods and maximum residue limits (MRLs). The toxicity factor was evaluated using the acceptable daily intake (ADI) and the regulatory importance of the substances. The regulatory importance was classified according to the antimicrobial resistance (AMR) ranking criteria of the World Health Organization (WHO) and the World Organisation for Animal Health (WOAH), while substances not covered by these classifications were ranked based on their impact on the human intestinal microbiota. According to the results of residue violation grading by country and substance, when focusing on meat (excluding dairy products), the United States had the highest number of Grade 5 substances (seven), followed by Canada, Brazil, Mexico, Spain, Uruguay, and Chile, which each contained Grade 5 substances. In domestic livestock products, 14 substances—including cefazolin and amoxicillin—were classified as Grade 5 in beef, eight substances—including amoxicillin and cefquinome—as Grade 5 in pork, and bifenthrin as a Grade 5 substance in poultry. Based on MRL grading, phenylbutazone, norgestomet, and flumethasone were classified as Grade 5 in beef; phenylbutazone, altrenogest, and flumethasone in pork; and phenylbutazone and dexamethasone in poultry. For ADI-based grading, oleandomycin, cefadroxil, avilamycin, norgestomet, and dexamethasone were identified as Grade 5 substances. Withdrawal period grading indicated that gentamicin was categorized as Grade 5 across all livestock types, including cattle, swine, poultry, and milk. In terms of regulatory importance, danofloxacin, ceftiofur, spiramycin, erythromycin, and enrofloxacin were classified as Grade 5 substances. The risk-prioritization algorithm developed in this study identified five substances—ampicillin, closantel, phenylbutazone, ractopamine, and zeranol—as having the highest possible risk score (25 points) in imported beef. This algorithm enables risk-based prioritization using the results of national residue monitoring programs conducted by exporting countries, thereby allowing importers to establish inspection priorities tailored to their own contexts. Consequently, the developed algorithm can be effectively utilized to identify high-risk veterinary drugs by exporting country and livestock type, supporting the establishment of more efficient monitoring plans for imported livestock products.

본 연구는 12주간 아쿠아로빅 운동프로그램이 비만 여성노인의 심혈관질환 위험요인에 미치는 영향을 구명하기 위해 아쿠아로빅 운동군(AEG, n=15), 대조군(CG, n=15)으로 구분하여 실시하였다. 아쿠 아로빅 운동프로그램은 주 2회 1회 운동 60분을 실시하였으며, 운동 강도는 1-4주, 40-50%HRR, RPE 12-13, 5-8주, 50-60%HRR, RPE 13-14, 9-12주, 60-65%HRR, RPE 14-15로 설정하여 실시하였다. 자료처리는 측정항목에 대한 평균값(M)과 표준편차(SD)를 산출하였고, 그룹 및 시기 간 상호작용 효과검 증은 two-way repeated measures ANOVA를 실시하였다. 사후분석으로 그룹 내 시기 간 차이 검증은 paired t-test, 그룹 간 차이 검증은 independent t-test, 각 항목별 통계분석의 기본 유의수준(α)은 .05로 설정하였으며, 사후분석에서 보정된 유의수준(α)은 .01로 설정하였다. 그 결과, 비만요인은 변화가 나타나 지 않았고, 체력 검사의 3가지 항목은 상호작용 효과(p<.001), SBP와 HOMA-IR에서 상호작용 효과 (p<.05), 혈청지질 중 TC(p<.001), LDL-C(p<.01)은 상호작용 효과, HDL-C는 시기 간 주효과(p<.01)가 나타났다. 이상의 결과를 통해, 12주간 실시한 아쿠아로빅 운동프로그램은 비만 여성노인의 체력과 심혈관 질환 위험인자의 대부분에 긍정적인 영향을 미치며, 이를 통해 향후 심혈관질환의 노출 비율을 감소시킬 수 있고, 근골격계의 무리 없이 수행할 수 있는 운동 프로그램이라 사료된다. 하지만, 비만요인과 TG의 개 선이 나타나지 않은 이유는 본 프로그램의 운동의 횟수, 지속시간, 강도가 비만 여성노인의 식이통제 없이 에너지 소비에 부족할 수 있다는 것을 확인하였다. 이를 통해 추후 연구에서는 식이통제나 운동 프로그램 을 개선하여 보다 나은 연구의 기초자료로 사용될 수 있길 기대한다.

Using highway accident data, this study predicts the probability of rollover, overturning, and fire accidents and identifies the related risk factors. Whereas existing studies rely primarily on limited explanatory variables and classical statistical models, this study simultaneously enhances predictive performance and interpretability by applying and comparing machine learning-based nonlinear prediction-analysis systems (XGBoost and Shapley additive explanations) with logistic regression, which offers advantages in statistical reasoning. The analysis identifies speeding, segment characteristics (tunnel, ramp, shoulder), and vehicle type (SUV, truck, trailer, and tank lorry) as common key risk factors. These results suggest the necessity of establishing a multilayered management system for speeding, improving facilities centered on high-risk sections (tunnel in/out, ramp, and downhill), performing custom inspections for each vehicle type (load, tire, and brake system), and improving driving behavior (enhancing forward attention, introducing a drowsiness warning system, etc.). This study provides a datadriven empirical basis for identifying the causes of major highway accidents and for designing effective prevention policies.

Crash risk in metropolitan areas arises from the interaction among driver behavior, enforcement infrastructure, and urban environmental conditions; however, microspatial evidence remains scarce. This study examines the effects of automated speed-enforcement cameras on the crash risk in Seoul at the legal-dong level using the accident risk index, which accounts for both crash frequency and injury severity. The dataset combines crash records, enforcement infrastructure, school-zone information, demographic indicators, and land-use characteristics. Methodologically, a Bayesian negative binomial regression model was employed to address overdispersed crash data, whereas gradient-boosting machine and eXtreme Gradient Boosting models with SHAP interpretations were applied to capture nonlinear effects, heterogeneity, and variable interactions. The results reveal that the crash risk is spatially concentrated, with a small proportion of districts contributing disproportionately to the overall exposure. Regression analysis highlights enforcement and behavioral factors as significant predictors, whereas machine-learning models emphasize the added importance of structural and demographic characteristics, such as road area and floating population. This divergence reflects the sensitivity of the regression to collinearity and linearity assumptions in contrast to the flexibility of tree-based methods in uncovering nonlinear and context-dependent influences. In general, the findings reflect the value of integrating statistical and machine-learning approaches for a more comprehensive understanding of crash risk at the microspatial scale. This study advances the methodological diversity in traffic-safety research and provides practical evidence that cameradeployment strategies should be context sensitive while targeting areas with concentrated risks and distinct structural and demographic profiles.

This study compared and analyzed the toxic concentration and impact range that could kill 50% of the people in the Samae 2 Tunnel by using the ALOHA PROGRAM and Probit analysis methods for each substance, including ammonia, in the event of an accidental gas leak at the exit of a Level 3 road tunnel while transporting toxic substances in tanks fixed to vehicles, and applied the results to Google Earth. This study showed that the impact range differs by substance when toxic gases stay and move in the tunnel. Therefore, it is necessary to change the direction of installing additional or reducing evacuation connection routes by referring to the impact range using simulations for each substance. The results of this study estimated that there is a high probability of 50% of casualties due to toxic concentrations, so it is necessary to recognize that toxic gases in tunnels also pose a potential risk of casualties just like smoke, and in the future, it is necessary to establish new standards for smoke extraction or exhaust to expel toxic gases out of tunnels.

본 연구에서는 개별 상수도관이 상수도관망 유지관리 데이터베이스에서 관의 격리 밸브 사이의 개체로 정의되지 않은 경우 상수도관의 사고 위험도를 산정할 수 있는 기법을 개발하였다. 이 기법은 사고 위험도 산정 과정에서 상수도관에서 사고 발생시 수리 및 복구 작업에 따른 상수도 관망의 일부 구간의 격리로 인한 사고 영향의 확장을 반영할 수 있도록 설계되었다. 이 기법을 활용하여 우리나라의 한 지자체 상수도 관망내 관들에 대한 사고 위험도를 평가하였다. 또한, 연구대상 상수도 관망의 관들에 대해 노후도를 산정하였으며, 이를 추정된 사고 위험도와 같이 활용하여 연구대상 상수도 관망의 유지관리 우선순위를 설정하였다. 마지막으로, 사고 위험도와 노후도에 적용한 가중치가 상수도 관망의 유지관리 우선순위 산정 결과에 미치는 영향을 분석하였다.

This study aims to establish a data-driven framework for identifying fishing vessel risk factors based on the Korean Maritime Accident Verdicts. As fishing vessels accounted for 64.66% of maritime accidents and 77.45% of fatalities in Korea (2020 – 2024), they represent a key target for maritime safety management. The narrative structure of verdicts — covering background, cause, and consequence — was transformed into 4M (Man, Machine, Method, Media)-based causal data, and the contribution ratios of each factor were calculated by an accident type. To complement documentary analysis, a HAZID (Hazard Identification) workshop was conducted to verify findings through field assessment. The proposed analytical framework converts narrative verdict records into numerical contribution values and reproducible causal sequences, enabling quantitative comparison of accident mechanisms across accident categories. This allows the identification of which causal factors and combinations should be prioritized for prevention efforts in fishing vessels, providing an objective basis for determining safety-check items and risk-control priorities. By integrating quantitative data analysis with field-based validation, this study establishes a practical and data-driven foundation for risk assessment in fishing-vessel design and safety management.