Point-of-care ultrasound (POCUS) is an imaging modality that has become a fundamental part of clinical care provided in the emergency department (ED). The applications of this tool in the ED have ranged from resuscitation, diagnosis, and therapeutic to procedure guidance. This review aims to summarize the evidence on the use of POCUS for diagnosis and procedure guidance. To achieve this, CrossRef, PubMed, Cochrane Library, Web of Science, and Google Scholar databases were extensively searched for studies published between January 2000 and November 2023. Additionally, the risk of bias assessment was performed using the Quality Assessment of Diagnostic Accuracy Studies 2 (for studies on the diagnostic role of POCUS) and Cochrane Risk of Bias tool (for studies on the use of POCUS for procedure guidance). Furthermore, diagnostic accuracy outcomes were pooled using STATA 16 software (StatCorp., College Station, TX, USA), while outcomes related to procedure guidance were pooled using the Review Manager software. The study included 81 articles (74 evaluating the diagnostic application of POCUS and seven evaluating the use of POCUS in guiding clinical procedures). In our findings sensitivities and specificities for various conditions were as follows: appendicitis, 65% and 89%; hydronephrosis, 82% and 74%; small bowel obstruction, 93% and 82%; cholecystitis, 75% and 96%; retinal detachment, 94% and 91%; abscess, 95% and 85%; foreign bodies, 67% and 97%; clavicle fractures, 93% and 94%; distal forearm fractures, 97% and 94%; metacarpal fractures, 94% and 92%; skull fractures, 91% and 97%; and pleural effusion, 91% and 97%. A subgroup analysis of data from 11 studies also showed that the two-point POCUS has a sensitivity and specificity of 89% and 96%, while the three-point POCUS is 87% sensitive and 92% specific in the diagnosis of deep vein thrombosis. In addition, the analyses showed that ultrasound guidance significantly increases the overall success rate of peripheral venous access (p = 0.02) and significantly reduces the number of skin punctures (p = 0.01) compared to conventional methods. In conclusion, POCUS can be used in the ED to diagnose a wide range of clinical conditions accurately. Furthermore, it can be used to guide peripheral venous access and central venous catheter insertion.

Cardiogenic shock (CS) may have a negative impact on mortality in patients with heart failure (HF) or acute myocardial infarction (AMI). Early prediction of CS can result in improved survival. Artificial intelligence (AI) through machine learning (ML) models have shown promise in predictive medicine. Here, we conduct a systematic review and meta-analysis to assess the effectiveness of these models in the early prediction of CS. A thorough search of the PubMed, Web of Science, Cochrane, and Scopus databases was conducted from the time of inception until November 2, 2023, to find relevant studies. Our outcomes were area under the curve (AUC), the sensitivity and specificity of the ML model, the accuracy of the ML model, and the predictor variables that had the most impact in predicting CS. Comprehensive Meta-Analysis (CMA) Version 3.0 was used to conduct the meta-analysis. Six studies were considered in our study. The pooled mean AUC was 0.808 (95% confidence interval: 0.727, 0.890). The AUC in the included studies ranged from 0.77 to 0.91. ML models performed well, with accuracy ranging from 0.88 to 0.93 and sensitivity and specificity of 58%-78% and 88%-93%, respectively. Age, blood pressure, heart rate, oxygen saturation, and blood glucose were the most significant variables required by ML models to acquire their outputs. In conclusion, AI has the potential for early prediction of CS, which may lead to a decrease in the high mortality rate associated with it. Future studies are needed to confirm the results.

Objectives This study aimed to analyze the impact of the United States Medical Licensing Examination (USMLE) Step 1 transition to a pass/fail scoring system in 2022 on the performance of first-time test takers in three distinct groups: Doctor of Osteopathy (DO) and Doctor of Medicine (MD) examinees from US/Canadian schools and examinees from non-US/Canadian schools. The analysis spans a decade-long period from 2012 to 2022, offering insights into the implications of this pivotal change in medical education. Methods We analyzed the performance of first-time USMLE Step 1 examinees from US/Canadian MD and DO programs and non-US/Canadian schools from 2012 to 2022, including the transition year to a pass/fail scoring system. Data were obtained from USMLE performance data reports and organized into annual contingency tables. Descriptive statistics and comparative analysis were used to identify trends and differences in performance across the groups. Data visualization techniques were employed to illustrate these findings, and the results were contextualized within the broader changes in medical education. Results In 2021, first-time takers from US/Canadian MD and DO Degree programs had pass rates of 96% and 94%, respectively, while non-US/Canadian schools had a pass rate of 82%. However, in 2022, these rates dropped to 93%, 89%, and 74%, respectively. The most significant relative decline was observed among non-US/Canadian Schools' first-time takers, with an 8% decrease. Repeaters consistently had lower pass rates across all groups. Conclusion The study reveals a notable decline in pass rates following the transition to pass/fail scoring, although this is based on just one year of data. This underscores the importance of students not rushing into the exam and dedicating sufficient time for preparation. The potential impact of this research could be transformative for medical education, but more years of data post-transition will be needed to confirm these initial findings. These findings serve as a reminder that the change in scoring does not diminish the rigor of the exam, prompting students to approach their studies with diligence and patience and potentially paving the way for systemic improvements in medical education and healthcare delivery worldwide.

Background: Aortic stenosis (AS) is frequently identified at an advanced stage after clinical symptoms appear. The aim of this systematic review and meta-analysis is to evaluate the diagnostic accuracy of artificial intelligence (AI) algorithms for AS screening.

Cardiogenic shock is a life-threatening condition that significantly impacts the mortality of patients suffering from acute myocardial infarction and heart failure. Despite advancements in medical care, early identification of cardiogenic shock remains a critical challenge, with delayed diagnosis contributing to poor outcomes. Recent developments in artificial intelligence and machine learninghave provided innovative solutions for predictive medicine, offering new opportunities to enhance early cardiogenic shock detection and intervention. This mini-review highlights the application of machine learning models, particularly in the context of artificial intelligence in age-related cardiogenic shock prediction, for predicting cardiogenic shock in hospitalized patients with acute myocardial infarction and heart failure. A thorough search of the PubMed, Web of Science, Cochrane, and Scopus databases was conducted from inception to November 2, 2023, to find relevant studies. The studies analyzed were drawn from various cohorts, focusing on the predictive accuracy of machine learning algorithms in clinical settings. Six studies were included in our review, including four retrospective studies and two prospective cohort studies. The results revealed that machine learning models achieved a pooled mean area under the curve of 0.808 (95% confidence interval: 0.727, 0.890), indicating reliable performance. The area under the curve in the included studies ranged from 0.77 to 0.91. Machine learning models performed well, with accuracy ranging from 0.88 to 0.93 and sensitivity and specificity of 58%–78% and 88%–93%, respectively. Key predictive variables across studies included age, blood pressure, heart rate, oxygen saturation, and blood glucose—all of which were routinely measured in hospital settings. These findings suggest that machine learning can support clinicians by automating risk assessment and providing early alerts, leading to better patient stratification and timely interventions. The integration of machine learning into clinical workflows can bridge the gap between data analysis and decision-making, potentially reducing the workload on medical professionals and enhancing patient outcomes. The use of machine learning models for early cardiogenic shock prediction represents a promising development in cardiology, harnessing artificial intelligence to improve diagnostic precision and facilitate proactive care. Continued research and validation are necessary to refine these models, optimize data management, and expand their applicability. By overcoming current limitations, machine learning can be seamlessly integrated into healthcare to support early detection and improve survival rates in patients at risk for cardiogenic shock.

The integration of artificial intelligence (AI) into medicine offers transformative potential, particularly in the detection and management of atrial fibrillation (AF). However, the intersection of AI and AF has not been comprehensively evaluated. This systematic review focuses specifically on the applications of AI in AF risk prediction, monitoring, and management. Following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines, a comprehensive search was conducted in PubMed and Google Scholar using terms such as artificial intelligence, deep learning, machine learning, artificial neural networks, and AF diagnosis. Methodological quality and risk of bias were assessed using the JBI critical appraisal checklist for qualitative research. Of the 109 studies screened, 39 met the inclusion criteria. Of these, 19 studies focused on AI’s role in AF risk prediction, while 20 studies addressed its application in monitoring and management. Machine learning models, including AI-ECG approaches such as the optimal time-varying machine learning model and the observational medical outcomes partnership common data model, demonstrated superior sensitivity and specificity compared to traditional models (Framingham, atherosclerosis risk in communities (ARIC), congestive heart failure, hypertension, age ≥75, diabetes, stroke, vascular disease (CHADS-VASc), and cohorts for heart and aging research in genomic epidemiology model for atrial fibrillation (CHARGE-AF). Wearable devices, such as patch monitors and smartwatches, emerged as reliable, cost-effective, and noninvasive alternatives to implantable cardiac monitors for continuous AF detection and patient-centered management. Despite these advances, the reliability and consistency of AI-based tools remain variable across studies due to data heterogeneity and methodological inconsistencies. Identified gaps include the need for standardized, labeled datasets, robust validation through prospective clinical trials, and improved data governance frameworks to ensure reliability and reproducibility. In conclusion, AI holds immense potential for AF prediction and management, but addressing these challenges is essential for its integration into clinical practice.