OpenAI’s GPT-4 could detect, reason, and revise errors in head CT reports, demonstrating its feasibility as a tool for proofreading radiology reports.

Background: The early prediction of antibiotic resistance in patients with a urinary tract infection (UTI) is important to guide appropriate antibiotic therapy selection. Objective: In this study, we aimed to predict antibiotic resistance in patients with a UTI. Additionally, we aimed to interpret the machine learning models we developed. Methods: The electronic medical records of patients who were admitted to Yongin Severance Hospital, South Korea were used. A total of 71 features extracted from patients’ admission, diagnosis, prescription, and microbiology records were used for classification. UTI pathogens were classified as either sensitive or resistant to cephalosporin, piperacillin-tazobactam (TZP), carbapenem, trimethoprim-sulfamethoxazole (TMP-SMX), and fluoroquinolone. To analyze how each variable contributed to the machine learning model’s predictions of antibiotic resistance, we used the Shapley Additive Explanations method. Finally, a prototype machine learning–based clinical decision support system was proposed to provide clinicians the resistance probabilities for each antibiotic. Results: The data set included 3535, 737, 708, 1582, and 1365 samples for cephalosporin, TZP, TMP-SMX, fluoroquinolone, and carbapenem resistance prediction models, respectively. The area under the receiver operating characteristic curve values of the random forest models were 0.777 (95% CI 0.775-0.779), 0.864 (95% CI 0.862-0.867), 0.877 (95% CI 0.874-0.880), 0.881 (95% CI 0.879-0.882), and 0.884 (95% CI 0.884-0.885) in the training set and 0.638 (95% CI 0.635-0.642), 0.630 (95% CI 0.626-0.634), 0.665 (95% CI 0.659-0.671), 0.670 (95% CI 0.666-0.673), and 0.721 (95% CI 0.718-0.724) in the test set for predicting resistance to cephalosporin, TZP, carbapenem, TMP-SMX, and fluoroquinolone, respectively. The number of previous visits, first culture after admission, chronic lower respiratory diseases, administration of drugs before infection, and exposure time to these drugs were found to be important variables for predicting antibiotic resistance. Conclusions: The study results demonstrated the potential of machine learning to predict antibiotic resistance in patients with a UTI. Machine learning can assist clinicians in making decisions regarding the selection of appropriate antibiotic therapy in patients with a UTI.

Background: Although the roles and responsibilities of hospitalists have grown considerably in recent years, research on the current job status and satisfaction levels of Korean hospitalists is lacking. Objective: We investigate the present state of Korean hospitalists and the factors influencing their job satisfaction 6 years after the pilot program’s launch. Design: This cross-sectional analysis was based on an online survey conducted from January 30 to February 18, 2023. Participants: Korean hospitalists (N = 303) Main Measures: The survey encompassed participant demographics, hospital information, education, clinical practice, research involvement, and job satisfaction. We employed multiple logistic regression analyses to identify determinants of satisfaction as a hospitalist. Key Results: The analysis was based on 79 hospitalists’ responses (response rate 26%). Respondents had a median age of 39 years; approximately half were male internal medicine specialists, possessing over 3 years of hospitalist experience. Most respondents were interested in clinical work (94.4%), with only 21.5% interested in research and evidence-based medicine. Over two-thirds indicated that non-clinical duties occupied less than 20% of their time. Overall, job satisfaction among hospitalists averaged 51.9%. Notably, the availability of a research mentor was significantly associated with job satisfaction (P = .011). While hospitalists with more than 3 years of experience, more hospitalists per facility, and autonomy were associated with increased job satisfaction, these associations were not statistically significant. Furthermore, there was no association between night shift work, work type, or work hours and job satisfaction. Conclusions: Although Korean hospitalists primarily focus on clinical practice, our study underscores the positive impact of mentorship from research mentors on job satisfaction, supported by comprehensive univariate and multivariate analyses. These findings signal a progressive transformation in the role of Korean hospitalists, as they increasingly engage in research alongside patient care. Graphical

The prediction of corticosteroid responses in coronavirus disease 2019 (COVID-19) patients is crucial in clinical practice, and exploring the role of artificial intelligence (AI)-assisted analysis of chest radiographs (CXR) is warranted. This retrospective case–control study involving mild-to-moderate COVID-19 patients treated with corticosteroids was conducted from 4 September 2021, to 30 August 2022. The primary endpoint of the study was corticosteroid responsiveness, defined as the advancement of two or more of the eight-categories-ordinal scale. Serial abnormality scores for consolidation and pleural effusion on CXR were obtained using a commercial AI-based software based on days from the onset of symptoms. Amongst the 258 participants included in the analysis, 147 (57%) were male. Multivariable logistic regression analysis revealed that high pleural effusion score at 6–9 days from onset of symptoms (adjusted odds ratio of (aOR): 1.022, 95% confidence interval (CI): 1.003–1.042, p = 0.020) and consolidation scores up to 9 days from onset of symptoms (0–2 days: aOR: 1.025, 95% CI: 1.006–1.045, p = 0.010; 3–5 days: aOR: 1.03 95% CI: 1.011–1.051, p = 0.002; 6–9 days: aOR; 1.052, 95% CI: 1.015–1.089, p = 0.005) were associated with an unfavorable corticosteroid response. AI-generated scores could help intervene in the use of corticosteroids in COVID-19 patients who would not benefit from them.

Objectives: This study assessed the effectiveness of in-context learning using Generative Pre-trained Transformer-4 (GPT-4) for labeling radiology reports. Methods: In this retrospective study, radiology reports were obtained from the Medical Information Mart for Intensive Care III database. Two structured prompts—the “basic prompt” and the “in-context prompt”— were compared. An optimization experiment was conducted to assess consistency and the occurrence of output format errors. The primary labeling experiments were performed on 200 unseen head computed tomography (CT) reports for multilabel classification of predefined labels (Experiment 1) and on 400 unseen abdominal CT reports for multi-label classification of actionable findings (Experiment 2). Results: The inter-reader accuracies in Experiments 1 and 2 were 0.93 and 0.84, respectively. For multi-label classification of head CT reports (Experiment 1), the in-context prompt led to notable increases in F1-scores for the “foreign body” and “mass” labels (gains of 0.66 and 0.22, respectively). However, improvements for other labels were modest. In multi-label classification of abdominal CT reports (Experiment 2), in-context prompts produced substantial improvements in F1-scores across all labels compared to basic prompts. Providing context equipped the model with domain-specific knowledge and helped align its existing knowledge, thereby improving performance. Conclusions: Incontext learning with GPT-4 consistently improved performance in labeling radiology reports. This approach is particularly effective for subjective labeling tasks and allows the model to align its criteria with those of human annotators for objective labeling. This practical strategy offers a simple, adaptable, and researcher-oriented method that can be applied to diverse labeling tasks.

Background/Objectives: Hyperglycemia in hospitalized patients is associated with an increased risk of complications, morbidity, mortality, and healthcare costs, regardless of a prior diagnosis of diabetes. The hospitalist system can improve various outcomes, including length of stay, medical costs, patient satisfaction, and mortality rates. However, the effects of hospitalist care on blood glucose control in hospitalized patients remain unclear. This study aimed to assess the specific effects of hospitalist services on blood glucose control in hospitalized patients, with a focus on hyperglycemia management and patient outcomes. Methods: This retrospective study reviewed the electronic medical records of patients diagnosed with diabetes at Yonsei Severance Hospital in Yongin, between March 2020 and February 2022. It included adults aged ≥20 years who were hospitalized and had undergone blood glucose measurements during hospitalization. Glycemic control was assessed using hemoglobin A1c, and the blood glucose levels were measured four times daily during hospitalization. Variability was quantified using the coefficient of variation and compared between hospitalist-led and traditional specialty care groups, over a 14-day hospitalization period. Results: Despite a higher baseline risk profile, patients receiving hospitalist-led care experienced significantly more stable glycemic variability over time (p = 0.002), suggesting better inpatient glucose management than those receiving traditional specialty care. Conclusions: Hospitalist-led care was associated with more stable glycemic variability over time in hospitalized patients with diabetes, despite a higher baseline burden of comorbidities and poorer glycemic control at admission.