Background: Recent advancements in large language models (LLMs) have accelerated their integration into clinical domains, including laboratory medicine. The performance of LLMs in answering board-level laboratory medicine questions has not been comprehensively evaluated. Given the importance of diagnostic accuracy in this field, rigorous and objective evaluations of LLM capabilities are essential. Methods: We assessed 12 LLMs from OpenAI, Anthropic, and Google using 320 Korean Residency Examination questions (2021–2024) spanning six laboratory medicine subspecialties. Standardized prompts were provided via their application programming interfaces under deterministic settings (temperature = 0). Questions were administered thrice to assess response reproducibility. Outputs were compared with validated answers and analyzed for accuracy, reasoning quality, and error typology. Results: Google’s Gemini 2.0 Pro achieved the highest accuracy (80.0%), followed by OpenAI’s GPT-4.5 (77.2%) and Anthropic’s Claude 3.7 Sonnet (74.1%). Accuracy decreased as the difficulty of questions increased (78.0% for easy vs. 45.1% for challenging). Subspecialty performance varied. Al models underperformed on questions on transfusion medicine (mean accuracy: 38.8%), primarily because of limitations in domain-specific and regional knowledge representations. Incorrect answers primarily resulted from reasoning errors. Reproducibility exceeded 95% for most models; however, some residual non-determinism appeared even with greedy decoding (temperature = 0). Conclusions LLMs demonstrated substantial potential for integration into laboratory medicine, particularly in clinical chemistry and immunology. Performance inconsistencies (particularly for high-difficulty questions) and knowledge gaps (notably for transfusion medicine) highlight the necessity for further development—potentially including domain-specific fine-tuning and retrieval-augmented generation integration—and robust expert oversight before clinical application.

Background: Fresh frozen plasma (FFP) is widely used to correct coagulopathy. However, many transfusions are administered without adherence to guidelines. The clinical utility of pre-transfusion coagulation tests and optimal dosing strategies remains poorly defined in practice. Methods: We retrospectively reviewed 181 adult patients who received FFP transfusions at a tertiary hospital from January to December 2023. Indications were selected by clinicians from a predefined checklist. Changes in prothrombin time-international normalized ratio (PT-INR), activated partial thromboplastin time (aPTT), and fibrinogen were analyzed. Multivariate linear regression was used to predict PT-INR correction. Time to post-transfusion testing was also evaluated. Results: Among 518 episodes, 39.6% of transfusions were administered to patients with a PT-INR＜1.7, indicating that the guideline criteria were not met. The median volume was 7.0 mL/kg, with only 16.2% exceeding 10 mL/kg.Pre-transfusion PT-INR was the strongest predictor of correction (β=–1.003, P＜0.001). Greater volumes of transfused FFP led to better correction, especially in individuals with PT-INR ≥3.0. The median time to follow-up testing was 8.5 hours, but showed no significant effect on correction. Conclusion FFP transfusion was most effective in patients with severe PT-INR prolongation. Frequent inappropriate transfusions and subtherapeutic dosing limited its effectiveness. These findings highlight the importance of targeted education and implementation of INR-based indications to improve FFP transfusion practices.

ABO-blood typing typically involves both front typing (cell typing) and back typing (serum typing) to ensure accurate determination of blood type. However, in neonates, back typing is frequently omitted due to their immature immune system. If an institution chooses red blood cells (RBCs) that match the neonate's blood type, omitting back typing could lead to unexplained cross-matching incompatibilities. Therefore, blood banks should be cautious with neonatal blood typing and transfusion practices to ensure transfusion safety.

Purpose: Triage tools play a vital role in classifying the severity of children in emergency departments (EDs). We investigated the association between the Korean Triage and Acuity Scale (KTAS) and severity of dyspnea in the ED. Methods: We conducted a retrospective study of children aged 3-14 years with dyspnea who visited the ED from January 2015 through December 2021. They were divided into severe (KTAS level 1-3) and non-severe (KTAS level 4-5) groups. Between the groups, we compared the clinical characteristics, including age, sex, associated symptoms, vital signs, route of visit, treatment at ED, and outcomes. Results: Among a total of 468 children with dyspnea, 267 and 201 were assigned to the severe and non-severe groups, respectively. The severe group had higher frequencies of fever (21.7% vs. 13.9%; P = 0.031), cough (53.2% vs. 43.3%; P = 0.034), systemic steroids (42.3% vs. 25.9%; P < 0.001), intravenous fluids (47.6% vs. 25.4%; P < 0.001), oxygen therapy (16.5% vs. 6.5%; P = 0.001), inotropics (4.1% vs. 1.0%; P = 0.042), and hospitalization (24.7% vs. 11.9%; P = 0.002). The severe group also showed a higher mean heart rate, respiratory rate, and temperature, and lower mean oxygen saturation (all Ps < 0.001). Among these findings, fever, heart rate, respiratory rate, temperature, intravenous fluids, oxygen therapy, inotropics, and hospitalization remained significantly different between the groups after defining the severe group as a KTAS level 1-2. Conclusion This study shows the association between KTAS and severity of dyspnea in the ED. Therefore, KTAS may reflect not only the initial clinical conditions but also emergency measures and outcomes in children with dyspnea who visit EDs.

Objective: A prolonged length of stay in the emergency department (EDLOS) is known to be associated with poorer outcomes in critically ill patients. However, this has not been proven in patients who visit the emergency department (ED) due to acute heart failure (AHF). We aimed to find out whether prolonged EDLOS is associated with major in-hospital complications in patients with AHF. Methods: This is a retrospective cohort study of AHF patients who were admitted to intensive care units (ICU) via the ED of a single academic hospital from January 2015 to December 2019. We divided the patients into two groups: EDLOS <24 hours and EDLOS≥24 hours. The primary outcome was major in-hospital complications, which included in-hospital death, application of continuous renal replacement therapy, or extracorporeal membrane oxygenation treatment. Secondary outcomes included in-hospital death, prolonged ICU stay (≥10 days), and prolonged hospital stay (≥14 days) excluding ED stay. Results: A total of 265 patients were enrolled. Of these 163 patients stayed in the ED for over 24 hours. The multivariable logistic analysis demonstrated that EDLOS ≥24 hours was independently associated with major in-hospital complications (odds ratio [OR], 3.296; 95% confidence interval [CI], 1.291-8.413; P=0.013). Analysis of the secondary outcomes showed that EDLOS ≥24 hours was associated with in-hospital death (OR, 2.607; 95% CI, 1.005-6.759; P=0.049) and prolonged hospital stay ≥14 days (OR, 2.458; 95% CI, 1.303-4.636; P=0.006). Conclusion Our study showed that in patients with AHF who visited ED and were admitted to the ICU, prolonged EDLOS was associated with major in-hospital complications.

Background: Transfusions in pediatrics need to be performed carefully because of various variables, such as the blood volume and immature immune system. As a result, adverse transfusion reactions may appear differently from adults. This study examined the frequency and types of adverse transfusion reactions in pediatric patients. Methods: From January 2018 to December 2021, this study was conducted on 58 children who requested red blood cells, platelets, and plasma blood components from Chungbuk National University Hospital. The frequency and types of adverse transfusion reactions were analyzed retrospectively by reviewing blood transfusion-related medical records and compared with previous studies. Results: Approximately 0.9% of total blood components were transfused into pediatric patients; 1,179 units of blood components were transfused. The number of transfusions for red blood cells, platelets, and plasma was 383, 712, and 84 units, respectively. Among 58 patients, 23 adverse transfusion reactions were observed in 15 (25.9%) patients. Of these, 18 were febrile nonhemolytic transfusion reactions, and five were allergic transfusion reactions. Febrile nonhemolytic transfusion reactions occurred in 66.7% of cases with red blood cells, and allergic transfusion reactions occurred with platelets in 60% of cases. Conclusion This paper reported the incidence and types of adverse transfusion reactions in pediatric patients. This is expected to be more frequent in pediatric patients than adults, but most of them were relieved by supportive treatment because the symptoms were mild. As the awareness of hemovigilance is still low, it is essential to recognize and deal with adverse transfusion reactions through continuous education.

Background: Pretransfusion testing is vital for safe transfusion. However, in situations without time to perform sufficient testing, all or part of the pretransfusion testing may be skipped to issue blood quickly. This study evaluated the safety of red blood cell (RBC) transfusion released by an emergency blood transfusion protocol through retrospective analysis at a tertiary hospital for eight years. Methods: All RBC transfusions following the emergency blood transfusion protocol from 2011 to 2018 at Seoul National University Hospital were included in the study. Crossmatching and unexpected antibody screening test results conducted after RBC release and the occurrence of hemolytic transfusion reactions were analyzed. Results: A total of 1,541 cases (5,299 RBCs issued) of emergency blood transfusion were identified. RBCs were issued after performing the immediate spin crossmatch without an unexpected antibody screening test in most cases (1,443; 93.64%), while RBCs were issued with no pretransfusion testing in 98 cases (6.36%). Antibody screening tests performed after the issue of RBCs showed that 17 (1.1%) cases were positive. Two units of RBCs from two different cases showed positive antiglobulin crossmatch test results. However, none of them were suspected to be associated with a hemolytic transfusion reaction. Conclusion The incidence of incompatible RBC release was very low in patients receiving RBC transfusion through the emergency blood transfusion protocol suggesting it can be used safely with minimal risk of hemolytic transfusion reactions caused by incompatible blood transfusions.

PURPOSE: The purpose of this study was to evaluate the diagnostic performance of S-Detect when applied to breast ultrasonography (US), and the agreement with an experienced radiologist specializing in breast imaging. METHODS: From June to August 2015, 192 breast masses in 175 women were included. US features of the breast masses were retrospectively analyzed by a radiologist who specializes in breast imaging and S-Detect, according to the fourth edition of the American College of Radiology Breast Imaging Reporting and Data System lexicon and final assessment categories. Final assessments from S-Detect were in dichotomized form: possibly benign and possibly malignant. Kappa statistics were used to analyze the agreement between the radiologist and S-Detect. Diagnostic performance of the radiologist and S-Detect was calculated, including sensitivity, specificity, positive predictive value (PPV), negative predictive value, accuracy, and area under the receiving operator characteristics curve. RESULTS: Of the 192 breast masses, 72 (37.5%) were malignant, and 120 (62.5%) were benign. Benign masses among category 4a had higher rates of possibly benign assessment on S-Detect for the radiologist, 63.5% to 36.5%, respectively (P=0.797). When the cutoff was set at category 4a, the specificity, PPV, and accuracy was significantly higher in S-Detect compared to the radiologist (all P<0.05), with a higher area under the receiver operator characteristics curve of 0.725 compared to 0.653 (P=0.038). Moderate agreement (k=0.58) was seen in the final assessment between the radiologist and S-Detect. CONCLUSION: S-Detect may be used as an additional diagnostic tool to improve the specificity of breast US in clinical practice, and guide in decision making for breast masses detected on US.
Breast* ; Decision Making ; Female ; Humans ; Information Systems ; Retrospective Studies ; Sensitivity and Specificity ; Ultrasonography* ; Ultrasonography, Mammary

Rhinovirus infection is typically associated with the common cold and has rarely been reported as a cause of severe pneumonia in immunocompetent adults. A 55-year-old previous healthy woman, who consumed half a bottle of alcohol daily, presented with respiratory failure after one week of upper respiratory infection symptoms. Radiography revealed bilateral, diffuse ground glass opacity with patchy consolidation in the whole lung field; bronchoalveolar lavage fluid analysis indicated that rhinovirus was the causative organism. After five days of conservative support, the symptoms and radiographic findings began to improve. We report this rare case of rhinovirus pneumonia in an otherwise healthy host along with a review of references.
Adult* ; Alcohol Drinking ; Bronchoalveolar Lavage Fluid ; Common Cold ; Female ; Glass ; Humans ; Lung ; Middle Aged ; Pneumonia ; Radiography ; Respiratory Insufficiency* ; Rhinovirus*