Background: Point-of-care ultrasound (POCUS) may support focused bedside decisions, but prehospital implementation remains limited by operational constraints, heterogeneous protocols, variable training, and limited outcome-focused evidence. This narrative review examines prehospital POCUS as a clinical decision-support tool across cardiac, thoracic, trauma, pediatric, and obstetric applications. Methods: A structured narrative review was conducted in MEDLINE/PubMed, Embase, and Cochrane Central from April 2017 to February 2026. Search terms combined the prehospital setting, ultrasound, and clinical applications. Of 464 records identified, 38 full-text articles were included, including nine systematic reviews on cardiac evaluation (n = 2), thoracic evaluation (n = 3), and trauma assessment (n = 4). Screening and full-text assessment were performed by a single reviewer. No formal risk-of-bias assessment, certainty grading, quantitative synthesis, or meta-analysis was performed. Results: In cardiac arrest, prehospital POCUS may identify cardiac activity and selected reversible causes, guide interventions, confirm return of spontaneous circulation, and support prognostic assessment, provided that image acquisition does not interrupt resuscitation or delay defibrillation. In nontraumatic shock and cardiac dysfunction, focused cardiac, lung, and vascular ultrasound may help characterize shock mechanisms, guide fluid therapy, and identify actionable findings such as pulmonary congestion, pericardial effusion, right ventricular strain, or major aortic pathology. In acute dyspnea, lung ultrasound may improve diagnostic confidence, particularly for cardiogenic pulmonary edema. In trauma, E-FAST has high specificity and may support triage, destination choice, trauma pathway activation, and preparation for definitive care; however, limited sensitivity prevents its use as a rule-out test. Pediatric and obstetric applications are feasible but remain emerging. Conclusions: Prehospital POCUS is an adjunctive decision-support tool rather than a standalone diagnostic test. Its value lies in answering actionable questions that may influence treatment, triage, transport destination, hospital prealert, or pathway activation. Implementation requires protocols, training, quality assurance, and governance, and must not delay time-critical care or transport.
BACKGROUND:This study evaluated short-term knowledge change and learner perceptions following a virtual reality (VR)-based point-of-care ultrasound (POCUS) training program for undergraduate medical students. METHODS:A single-center observational pre-post educational evaluation was conducted within an undergraduate POCUS training course at the National University of Singapore. Students completed a self-directed online module followed by a VR simulation session covering cardiac, lung, and eFAST ultrasound. Knowledge was assessed using a 19-item multiple-choice questionnaire administered immediately before and after the VR session. Post-session survey responses assessed learner experience, usability, perceived learning benefit, and adverse effects. RESULTS:Of 290 students who completed the pre-training assessment and participated in the VR session, 275 matched pre-post datasets were available for analysis. Mean MCQ score increased from 0.689 out of 1 (SD 0.162) before training to 0.809 out of 1 (SD 0.153) after training, corresponding to a mean paired improvement of 0.120, or approximately 2.3 additional correct answers out of 19 items (95% CI 0.104 to 0.135; p < 0.001; Cohen's dz = 0.93). Survey responses demonstrated favorable learner perceptions. Participants reporting adverse effects rated the session as less engaging than those without adverse effects. CONCLUSIONS:VR-based POCUS simulation, delivered within a blended learning pathway, was associated with improved immediate MCQ-assessed knowledge and favorable learner perceptions. Further studies should assess effectiveness by practical skills assessment and long-term knowledge retention.
The rapid integration of Artificial Intelligence (AI) into Point-of-Care Ultrasound (POCUS) represents a transformative shift, offering the potential to democratize diagnostic expertise while simultaneously presenting significant risks regarding clinical validation, workforce preparedness, and health equity. Informed by a recent global survey indicating that while 81% of practitioners are optimistic about AI, major concerns remain regarding training and evidence gaps, the World Interactive Network Focused On Critical UltraSound (WINFOCUS) proposes a unified strategic framework. This manifesto outlines seven foundational pillars to guide the ethical and effective adoption of AI-augmented POCUS: (1) earning trust through rigorous, prospective evidence; (2) building an AI-literate workforce through evolved curricula; (3) championing global equity to prevent widening health disparities; (4) ensuring algorithmic transparency and accountability; (5) designing for seamless human-AI collaboration; (6) establishing a sustainable, privacy-centric data infrastructure; and (7) committing to continuous, patient-centered evaluation. We present this roadmap as a global call to action for clinicians, researchers, and industry partners to collectively shape a future where technology amplifies clinical wisdom and improves patient outcomes.
BACKGROUND:Ultrasound-guided (USG) peripheral intravenous catheter (PIVC) insertion improves cannulation success, particularly in patients with difficult intravenous access. Whether USG also improves post-insertion outcomes, including catheter failure, dwell time, and complications, remains uncertain. METHODS:We conducted a systematic review and meta-analysis of randomized controlled trials and comparative cohort studies comparing USG with landmark-guided PIVC insertion and reporting at least one post-insertion outcome. PubMed, Embase, Cochrane CENTRAL, and CINAHL were searched from January 2000 to March 2026. Risk of bias was assessed using RoB 2 and the Newcastle-Ottawa Scale. Random-effects meta-analysis used the DerSimonian-Laird estimator; REML with Hartung-Knapp-Sidik-Jonkman adjustment was performed as a post-hoc sensitivity analysis because few studies contributed to pooled outcomes. RESULTS:Fourteen studies (5 randomized trials, 9 cohort studies; 78,209 participants) were included. Catheter failure did not differ significantly between USG and landmark groups (k = 4; RR 1.23, 95% CI 0.99-1.51; p = 0.056), although the estimate was sensitive to model choice and exclusion of the largest study. Dwell time and extravasation were not pooled because of substantial heterogeneity (I² = 91.9% and 95.7%, respectively). Infiltration also showed no significant difference (k = 2; RR 0.68, 95% CI 0.12-3.83). Several studies reporting favorable USG outcomes were substantially confounded by systematic differences in catheter length or material between groups. CONCLUSIONS:Current evidence does not demonstrate a consistent post-insertion advantage of USG over landmark PIVC placement. Reported benefits in some studies may reflect catheter-specification differences rather than ultrasound guidance itself. Larger, well-controlled trials are needed.
BACKGROUND:The objective of this study was to evaluate whether deep learning radiomic (DLR) models utilizing B-mode ultrasound (BUS) and contrast-enhanced ultrasound (CEUS) could improve the preoperative prediction of lymph node metastasis (LNM) in patients with hilar cholangiocarcinoma (HCCA). METHODS:The study included 110 HCCA patients from two clinical centers, divided into primary and external validation cohorts. Pathological verification of lymph node status was performed, and the ResNet101 architecture was used to extract deep learning features (DLFs) from BUS and CEUS images. The Genetic Programming-based Symbolic Regression (GPSR) algorithm was applied to integrate radiomic features (RadFs) and DLFs, generating deep learning radiomic features (DLRFs). DLR models were subsequently constructed using the eXtreme Gradient Boosting (XGBoost) algorithm. RESULTS:Lymph node metastasis was identified in 48 out of 110 patients (43.64%). No significant differences in clinical characteristics were observed between LNM-positive and LNM-negative groups (P-values ranging from 0.14 to 0.98). A total of 837 RadFs and 4095 DLFs were initially extracted from each tumor region of interest (ROI). After feature selection, 10 RadFs (4 from BUS, 6 from CEUS) and 27 DLFs (5 from BUS, 22 from CEUS) were retained. Using the GPSR algorithm, 5 BUS-DLRFs, 10 CEUS-DLRFs, and 15 Combination-DLRFs were generated, leading to the development of three corresponding DLR models. In internal validation, the AUC values were 0.70 for the BUS-DLR model, 0.77 for the CEUS-DLR model, and 0.83 for the Combination-DLR model. In external validation, the AUC values were 0.66, 0.68, and 0.72, respectively. These results indicate that the integration of multiphasic CEUS and BUS data is essential for more comprehensively identifying LNM and achieving precise preoperative staging. CONCLUSIONS:The DLR models based on DLRFs demonstrated an enhanced ability to preoperatively predict LNM in HCCA patients, indicating that the integration of deep learning and RadFs from BUS and CEUS may offer improved predictive performance for LNM.
