Ankle fractures are common and, compared to other injuries, tend to be overlooked in the emergency department. We aim to develop a deep learning algorithm that can detect not only definite fractures but also obscure fractures. We collected the data of 1226 patients with suspected ankle fractures and performed both X-rays and CT scans. With anteroposterior (AP) and lateral ankle X-rays of 1040 patients with fractures and 186 normal patients, we developed a deep learning model. The training, validation, and test datasets were split in a 3/1/1 ratio. Data augmentation and under-sampling techniques were administered as part of the preprocessing. The Inception V3 model was utilized for the image classification. Performance of the model was validated using a confusion matrix and the area under the receiver operating characteristic curve (AUC-ROC). For the AP and lateral trials, the best accuracy and AUC values were 83%/0.91 in AP and 90%/0.95 in lateral. Additionally, the mean accuracy and AUC values were 83%/0.89 for the AP trials and 83%/0.9 for the lateral trials. The reliable dataset resulted in the CNN model providing higher accuracy than in past studies.

To develop and validate a novel artificial intelligence model for predicting the initial empiric dose of vancomycin, with the aim of achieving an area under the concentration-time curve (AUC) of 400-600 mg∙h/L, using individual clinical data. Machine learning models were developed and validated internally and externally using data from adult patients who received intravenous vancomycin treatment between November 2020 and June 2023, using records from July to September 2023, sourced from two hospitals. This study included 205, 44, and 35 patients in the training, internal validation, and external validation sets, respectively. The Random Forest, XGBoost, and Ensemble models were evaluated based on the mean squared error, root mean squared error, R-squared value, and accuracy (±20 % of the actual dose). In internal validation, the Random Forest model achieved the highest 20% Accuracy at 56.8%, while the XGBoost model achieved 56.8% and Voting Ensemble models attained 54.5% accuracy. In external validation, the XGBoost model achieved the highest 20% Accuracy at 74.3%, followed by Random Forest and Voting Ensemble, both also achieving 74.3% accuracy. The estimated glomerular filtration rate was the most significant predictor in all models, with body weight, serum creatinine, height, and age also prominently influencing the XGBoost model's predictions. The proposed model is capable of accurately predicting the optimal dose of vancomycin, which could aid physicians in making more informed treatment decisions regarding early therapy, particularly when AUC estimation systems are not accessible or readily available in routine clinical practice.

Background: In military settings, determining whether a patient with abdominal pain requires emergency care can be challenging due to the absence or inexperience of medical staff. Misjudging the severity of abdominal pain can lead to delayed treatment or unnecessary transfers, both of which consume valuable resources. Therefore, our aim was to develop an artificial intelligence model capable of classifying the urgency of abdominal pain cases, taking into account patient characteristics. Methods: We collected structured and unstructured data from patients with abdominal pain visiting South Korean military hospital emergency rooms between January 2015 and 2020. After excluding patients with missing values, 20,432 patients were enrolled. Structured data consisted of age, sex, vital signs, past medical history, and symptoms, while unstructured data included preprocessed free text descriptions of chief complaints and present illness. Patients were divided into training, validation, and test datasets in an 8:1:1 ratio. Using structured data, we developed four conventional machine learning models and a novel mixed model, which combined one of the best performing machine learning models with emergency medical knowledge. And we also created a deep learning model using both structured and unstructured data. Results: Xgboost demonstrated the highest performance among the six models, with an area under the precision-recall curve (AUPRC) score of 0.61. The other five models achieved AUPRC scores as follows: logistic regression (0.24), decision tree (0.22), multi-layer perceptron (0.21), deep neural network (0.58), and mixed model (0.58). Conclusion: This study is the first to develop an AI model for identifying emergency cases of abdominal pain in a military setting. With more balanced and better-structured datasets, clinically significant AI model could be developed based on the findings of this study.

Methicillin-resistant Staphylococcus aureus (MRSA) remains a major cause of bloodstream infection worldwide. In Korea, sequence type (ST) 72 has predominated, whereas ST8, including the USA300 lineage, has recently emerged. Comparative data on these genotypes in MRSA bacteremia (MRSAB) are limited. We conducted a retrospective cohort study of adult patients with MRSAB admitted to a 2700-bed tertiary care hospital in Republic of Korea between July 2008 and December 2020. Clinical features and outcomes of patients with ST8 MRSA were compared with those of patients with ST72 MRSA. Among 1975 cases of S. aureus bacteremia, 998 (50.5%) were due to MRSA, including 327 (32.7%) ST72 and 23 (2.3%) ST8 isolates. Demographics and comorbidities were similar, though pneumonia appeared more frequent in ST8 cases without statistical significance. ST8 isolates exhibited greater resistance to ciprofloxacin and erythromycin and more frequent vancomycin E-test MICs ≥1 mg/L, while broth microdilution MICs were comparable. spa type distribution differed, with t324 predominating in ST72 and t008 in ST8. Management practices, persistent bacteremia, recurrence, and 30- and 90-day mortality did not differ significantly. In multivariable analysis, liver cirrhosis and Charlson comorbidity index >4, but not MRSA genotype, independently predicted 30-day mortality. These findings highlight the importance of continued surveillance of emerging ST8 clones.

Background: Although public access defibrillation programs have expanded, real‐world bystander automated external defibrillator (AED) use remains limited. The concept of functional accessibility, the realistic ability of bystanders to retrieve and use an AED within a time window that affects survival, has not yet been systematically evaluated. This study aimed to evaluate the functional accessibility of publicly installed AEDs by integrating factors influencing their usability in out‐of‐hospital cardiac arrest events. Methods: We conducted a population‐based geospatial analysis of 39 563 out‐of‐hospital cardiac arrests in Gyeonggi‐do, Korea (2021–2023). Each incident was matched to the nearest public AED using network‐based walking routes. Functional accessibility was measured by walking time and distance, stratified by urbanization, time of day, and location of arrest. Multivariable logistic regression was used to assess the association of AED accessibility with bystander AED use and prehospital return of spontaneous circulation. Results: The median walking time to the nearest AED was 1.9 minutes; only 25.4% of rural arrests were within 3 minutes. Among 17 285 resuscitation‐attempted cases, bystander AED use occurred in 2.9%. Shorter walking time was a statistically significant predictor of AED use and prehospital return of spontaneous circulation. Nighttime and residential arrests showed markedly reduced AED use irrespective of proximity. Moreover, although home arrests accounted for 70% of cases, bystander defibrillation occurred in <2%. Conclusions: Functional accessibility determines both device use and survival outcomes. Policies should prioritize 24/7 external AED availability, integration of dispatcher‐assisted public access defibrillation programs, and novel delivery systems, such as drones, to bridge the spatial and temporal gaps in access.

Background and Objectives: Emergency physicians practice in high-pressure environments and face occupational stressors that may affect their mental health. This study was designed to evaluate the prevalence of depression among emergency physicians in South Korea and examined environmental, sociolegal, and individual factors associated with depressive symptoms in the post-pandemic period. Materials and Methods: This nationwide cross-sectional study analyzed data from the 2025 Korean Emergency Physician Survey. Screening positive for depressive symptoms was defined as a Patient Health Questionnaire-9 (PHQ-9) score ≥ 10, indicating moderate-to-severe depressive symptom severity. Measures included the PHQ-9, the Korean Epworth Sleepiness Scale (KESS), and the Adult APGAR, a brief self-administered instrument assessing overall wellness. Multivariable logistic regression was performed to identify factors associated with depression after adjusting for demographic, clinical, and work-related variables. Results: Of the 1050 physicians who responded (response rate: 37.5%), 743 emergency physicians completed the PHQ-9 section (completion rate: 70.8%; mean age, 43.2 ± 7.78 years; 86.5% male), and 111 (14.9%) screened positive for depressive symptoms. Objective workload indicators, including total work hours and number of night shifts, did not differ between physicians with and without depression. However, emergency physicians screening positive for depression reported higher perceived burdens related to staffing shortages and patient-related stressors. Protective factors included being married (adjusted odds ratio [AOR], 0.22; 95% confidence interval [CI], 0.08–0.58), longer sleep duration (AOR, 0.65; 95% CI, 0.50–0.86), better sleep quality (AOR, 0.45; 95% CI, 0.27–0.74), fixed mealtimes (AOR, 0.60; 95% CI, 0.39–0.93), and higher Adult APGAR scores (AOR, 0.72; 95% CI, 0.60–0.86). Factors associated with increased odds of depression included a history of cancer (AOR, 14.63, 95% CI, 2.53–84.61), current alcohol consumption (AOR, 2.54, 95% CI, 1.14–5.68), daytime sleepiness (AOR, 1.17; 95% CI, 1.04–1.31), and more frequent verbal abuse during the previous 12 months (AOR, 1.25; 95% CI, 1.08–1.44). Conclusions: Depression was prevalent and was associated with perceived work burden, sleep health, lifestyle regularity, and psychosocial factors. Interventions should address sleep quality, workplace safety, and social support.