Artificial intelligence tools show promise in fracture detection but may be impaired by hidden stratification. We aim to evaluate the diagnostic accuracy of a commercially available AI tool for detecting hip and pelvic fractures, with emphasis on clinically relevant subgroups and AO/OTA fracture classifications. This retrospective multicenter diagnostic test accuracy study included consecutive trauma patients who underwent hip or pelvic radiography. The reference standard was based on post-conference clinical radiology reports, incorporating MRI, CT, and radiography in hierarchical order. Fractures were classified according to the AO/OTA system. Studied subgroups included surgical metal, degenerative disease, old fractures, radiographically occult fractures and fracture classification. Sensitivity and specificity were calculated with 95% confidence intervals. Among 642 patients (median age 82years), 262 (42%) had fractures. Overall sensitivity was 87% [83-91%] and specificity 86% [82-89%]. Specificity was reduced in cases with old fractures (29% [13-51%]). Sensitivity was high for femoral neck (95% [88-98%]) and trochanteric fractures (92% [82-97%]) and moderate for pelvic (82% [70-91%]) and acetabular fractures (58% [28-85%]). Within each segment the classifications with most missed fractures were unilateral anterior pelvic arch (61A2.2), subcapital femur neck (31B1), and simple trochanteric fractures (31A1). In summary, while specificity was notably reduced in cases with old fractures and sensitivity was moderate for pelvic and acetabular fractures, the AI tool demonstrated high diagnostic accuracy for hip-region trauma radiographs, with most missed cases occurring among the subtle AO/OTA fracture groups.

Bright-blood late gadolinium enhancement (LGE) cardiovascular magnetic resonance (CMR) is standard for myocardial scar detection, yet subendocardial infarcts may be difficult to identify due to limited scar-blood contrast. Grayscale inversion is a simple, vendor-independent post-processing technique that may enhance visual contrast. We assessed whether grayscale inversion improves diagnostic accuracy or reader confidence, and whether its effect varies by reader experience or scanner vendor. In this single-center, multivendor study, 120 patients (90 with and 30 without ischemic scar) underwent 2D bright-blood LGE on GE, Philips, and Siemens 1.5T scanners. Three short-axis slices per patient were evaluated in standard bright-blood and grayscale-inverted visualizations by nine readers (three per EACVI experience level). Readers assessed scar presence, transmurality, and diagnostic confidence (4-point scale). Consensus of two Level-3 experts served as reference. Across 34,560 segment-level assessments, diagnostic accuracy was high (91.5%, 95% CI 91.0-91.9%) and similar for standard vs. inverted images (90.7% vs. 91.9%, p=0.27), with no differences across experience levels or in low-transmurality lesions. In contrast, reader confidence increased significantly with grayscale-inversion (3.25±0.81 vs. 2.78±0.88, p<0.0001), with the greatest relative improvement among Level-1 readers (2.63±0.89→3.00±0.83, p<0.001). Interobserver agreement and transmurality reproducibility remained unchanged. Results were consistent across vendors. Grayscale-inverted bright-blood LGE significantly increases reader confidence without compromising diagnostic accuracy or interobserver agreement. This simple, vendor-neutral technique may support training and improve interpretive certainty in routine clinical LGE assessment.

To assess the diagnostic accuracy of ultrahigh-resolution (UHR) photon-counting detector (PCD)-CT for detecting obstructive and non-obstructive coronary in-stent restenosis (ISR), using invasive coronary angiography (ICA) as the reference. Consecutive patients with coronary stents who underwent clinically indicated coronary CT angiography (CCTA) on a dual-source PCD-CT were retrospectively identified from two academic centers. All UHR scans were reconstructed at 0.2mm and 0.4mm slice thickness, the latter served as an energy-integrating detector CT-like proxy. Series were assessed for the presence of non-obstructive (<50%) and obstructive (≥50%) ISR, with diagnostic accuracy validated against ICA in the overall cohort and in a sub-analysis of stents<3 mm in diameter. Fifty patients (16.0% women; 70.9±10.8years) with 106 coronary stents were included. ICA identified 17 non-obstructive and 27 obstructive ISR. Non-obstructive ISR was detected in 10 stents on 0.2mm UHR and 14 on 0.4mm EID-like standard-resolution reconstructions.Obstructive ISR was described in 26 (0.2mm) vs. 32 (0.4mm) stents, respectively. The proportion of evaluable stents increased from 91.5% based on 0.4mm to 99.1% with 0.2mm UHR. Compared with standard-resolution images, 0.2-mm UHR improved non-obstructive ISR sensitivity (90.5%vs82.9%), specificity (100.0%vs78.5%), and accuracy (96.2%vs80.2%); and in obstructive ISR, specificity (98.7%vs82.3%), positive predictive value (96.3%vs64.1%), and accuracy (98.1%vs84.9%). Thin-slice UHR achieved higher accuracy in small-caliber stents (<3mm), for both non-obstructive (94.1vs73.5%) and obstructive ISR (94.1%vs79.4%). UHR PCD-CT demonstrated excellent diagnostic accuracy and significant improvements in ISR detection compared to the quasi-standard reconstructions, including non-obstructive and small-caliber stent ISR.

To evaluate the diagnostic accuracy and feasibility of the swirl sign, a novel bedside visual adjunct, for confirmation of epidural catheter placement. Prospective, randomized, parallel-group pilot diagnostic study. Single tertiary-care teaching hospital. Adult patients (18–65 years) with American Society of Anesthesiologists physical status I–III undergoing elective lower abdominal or lower limb surgery under combined spinal–epidural anesthesia. Patients were randomized to epidural catheter confirmation using the swirl sign combined with a conventional test dose (swirl group) or test dose alone (control group). The primary outcome was successful epidural block, defined as achievement of bilateral sensory block to T10 within 20 min without catheter reinsertion. Successful clinical block served as the pragmatic reference standard for diagnostic accuracy analysis. Secondary outcomes included diagnostic accuracy of the swirl sign (sensitivity, specificity, predictive values), onset time of analgesia, need for catheter reinsertion, and interobserver agreement for swirl sign interpretation. A total of 180 patients were randomized (90 per group). Successful epidural block occurred in 94.4% of patients in the swirl group and 89.4% in the control group ( p = 0.21). Time to onset of T10 sensory block was shorter in the swirl group (11.2 ± 2.3 vs 12.1 ± 2.6 min; p = 0.03). The swirl sign demonstrated a sensitivity of 96.6%, specificity of 80.0%, positive predictive value of 98.9%, and negative predictive value of 57.1% when compared with successful clinical block as the reference standard. Interobserver agreement for swirl sign interpretation was excellent (κ = 0.84). No serious complications were observed. The swirl sign is a feasible and highly sensitive bedside adjunct for epidural catheter confirmation, with excellent interobserver reproducibility. It did not significantly alter overall block success rates and should be considered an adjunctive diagnostic indicator rather than a replacement for standard confirmation methods. As an initial evaluation, this randomized pilot study supports further validation of the swirl sign in larger and multicenter settings.

Comparative accuracy of telecytology and in-person rapid on-site evaluation by specimen source: A multi-campus analysis.

Evaluation of large language models for clinical sign-based oral assessment in dogs compared with veterinary practitioners.

Effects of implementing universal dysphagia screening at admission in a university hospital.

Traditional medical board examinations present clinical information in static vignettes with multiple-choices (MC), fundamentally different from how physicians gather and integrate data in practice. Recent advances in large language models (LLMs) offer promising approaches to creating more realistic clinical interactive conversations. However, these approaches are limited in neurosurgery, where patient communication capacity varies significantly and diagnosis heavily relies on objective data such as imaging and neurological examinations. We aimed to develop and evaluate a multi-artificial intelligence (AI) agent conversation framework for neurosurgical case assessment that enables realistic clinical interactions through simulated patients and structured access to objective clinical data. We developed a framework to convert 608 Self-Assessment in Neurological Surgery first-order diagnosis questions into conversation sessions using 3 specialized AI agents: patient AI for subjective information, system AI for objective data, and clinical AI for diagnostic reasoning. We evaluated generative pretrained transformer 4o's (GPT-4o's) diagnostic accuracy across traditional vignettes, patient-only conversations, and patient + system AI interactions, with human benchmark testing from 10 neurosurgery residents. GPT-4o showed significant performance drops from traditional vignettes to conversational formats in both MC (89.0%-60.9%, P < .0001) and free-response scenarios (78.4%-30.3%, P < .0001). Adding access to objective data through system AI improved performance (to 67.4%, P = .0015; and 61.8%, P < .0001, respectively). Questions requiring image interpretation showed similar patterns but lower accuracy. Residents outperformed GPT-4o in free-response conversations (70.0% vs 28.3%, P = .0030) using fewer interactions and reported high educational value of the interactive format. This multi-AI agent framework provides both a more challenging evaluation method for LLMs and an engaging educational tool for neurosurgical training. The significant performance drops in conversational formats suggest that traditional MC testing may overestimate LLMs' clinical reasoning capabilities, while the framework's interactive nature offers promising applications for enhancing medical education.

Complex regional pain syndrome (CRPS) is an orphan condition with no interventions proven effective in large-scale clinical trials. In-person clinical evaluations are required to make a CRPS diagnosis and determine study eligibility, making recruitment for CRPS trials challenging. We developed and validated a self-report measure of CRPS symptoms suitable for remote administration. We evaluated diagnostic accuracy of a self-report version of the CRPS Severity Score (CSS-SR) and assessed its validity relative to the original clinician-completed CSS (CSS-Clinician) in two samples: 112 postsurgical patients with total knee arthroplasty and 80 patients with CRPS and non-CRPS limb pain attending a chronic pain clinic. All patients completed the CSS-SR and underwent a semi-structured history and physical examination from which the CSS-Clinician was scored and a criterion-based Budapest CRPS diagnosis was determined. There was good to excellent correspondence (intraclass correlations [ICCs]) between CSS-Clinician and CSS-SR total scores (postsurgical sample: ICCs = 0.86-0.92; pain clinic sample: ICC = 0.76). In the postsurgical sample, the 2 measures displayed good agreement regarding changes in CRPS features across 2 assessments 4 months apart (ICC = 0.84). The CSS-SR exhibited high diagnostic accuracy in receiver operating characteristic analyses (area under the curve = 0.84-0.94). In the pain clinic sample, a CSS-SR cutoff of ≥8 had a 79% probability of correctly predicting a Budapest CRPS diagnosis after full evaluation (sensitivity = 0.79, specificity = 0.79). The CSS-SR appears valid as a self-report measure of CRPS status and displays good accuracy as a remote diagnostic screening tool.

Pediatric skin conditions are a common reason for primary care visits, yet access to general dermatologists and pediatric dermatologists remains limited, particularly in rural areas. Primary care providers often manage these conditions with minimal formal dermatology training, which can result in diagnostic and treatment challenges. The Extension for Community Healthcare Outcomes (ECHO) model offers a telementoring framework that combines didactic learning with case-based discussions to improve provider knowledge and clinical decision-making. Seven peer-reviewed studies evaluating pediatric and general dermatology ECHO programs have been identified that include retrospective cohort studies, program evaluations using strengths, weaknesses, opportunities, and threats analysis, a prospective educational intervention, and a descriptive case report. While program structures varied, results consistently demonstrated improvements in provider confidence, diagnostic accuracy, and ability to develop more effective treatment plans. Diagnostic concordance between primary care providers and dermatology experts ranged from 40% to 54%, with the lowest agreement on treatment plans. Participants reported high satisfaction, and several programs demonstrated increased attendance over time. Pediatric dermatology ECHO programs represent a promising and scalable strategy to improve dermatologic care for children in underserved settings. Continued development of pediatric-focused curricula, sustainable delivery models, and strategies to better assess outcomes will be critical to realizing Project ECHO’s full potential.

Micro-ultrasound (micro-US; 29-MHz) offers real-time, high-resolution prostate imaging, but its stand-alone diagnostic accuracy remains uncertain. We synthesized prospective evidence to evaluate micro-US for classifying clinically significant prostate cancer (csPCa) using histopathology as the reference standard. We searched PubMed, Embase, Scopus, and Web of Science (inception-20 May 2025) for prospective studies assessing micro-US as an index test on a diagnostic pathway. Data were pooled using random-effects models on logit-transformed sensitivity and specificity, with an HSROC representation and model diagnostics. Subgroup and meta-regression analyses explored heterogeneity, including threshold (PRI-MUS) and spectrum effects. Clinical utility was appraised using Fagan nomograms and a likelihood-ratio scatter. Small-study effects were evaluated with Deeks' test. Five prospective studies met criteria. Pooled sensitivity was 0.84 (95% CI 0.65-0.94) and pooled specificity was 0.41 (95% CI 0.25-0.59), indicating moderate discrimination on HSROC. Secondary metrics were concordant (PLR 1.45, 95% CI 1.17-1.80; NLR 0.37, 95% CI 0.23-0.61; DOR 3.95, 95% CI 2.48-6.30). On a 25% pre-test probability, the Fagan nomogram showed modest shifts (~ 33% after a positive test; ~11% after a negative), supporting a triage/rule-out role. Heterogeneity was substantial and strongly influenced by threshold and clinical spectrum differences; subgroup and meta-regression suggested that spectrum-related factors were associated with lower specificity, whereas no covariate robustly altered sensitivity (exploratory given small k). Model checks were acceptable, and Deeks' test showed no evidence of small-study effects (p ≈ 0.70). As a stand-alone index test for csPCa classification, micro-US demonstrates high sensitivity but low specificity, yielding modest impact on post-test probability. These findings support micro-US as a complementary/triage (rule-out) adjunct, particularly when mpMRI is unavailable, contraindicated, or delayed, while highlighting the need for standardized PRI-MUS thresholds, reader training, and larger multicenter studies to refine specificity and clarify integration with MRI-based pathways.

Oral and maxillofacial schwannoma (OMSCH): An institutional study of 102 patients.

Blood-based circulating tumour DNA (ctDNA) tests for colorectal cancer screening: Systematic review and meta-analysis of diagnostic accuracy.

Accurate assessment of tract-specific inflammatory activity remains difficult in perianal fistulizing Crohn's disease (pfCD), as current clinical and imaging indices incompletely reflect local activity. To evaluate fistula calprotectin (FiCP) as a tract-specific biomarker, focusing on diagnostic accuracy, associations with histological, clinical, and radiological indices, longitudinal responsiveness, and local infliximab exposure. In this single-center observational study (April 2023-July 2025; n = 105), patients with pfCD provided fistula exudate samples during surgical or outpatient assessment. FiCP concentrations were quantified by ELISA and correlated with histological inflammation, the Perianal Disease Activity Index (PDAI), the MRI-based modified Van Assche Index (mVAI), and total fistula volume (TFV). Diagnostic performance was analysed using receiver operating characteristic (ROC) curves with DeLong comparison, and treatment responsiveness was evaluated longitudinally using the standardised response mean (SRM). Serum and local infliximab concentrations were measured using ELISA. FiCP levels were significantly higher in fistulas with severe versus mild histological inflammation (p = 0.021) and in clinically or radiologically active tracts (both p < 0.001). FiCP correlated with PDAI (R = 0.61), mVAI (R = 0.40), and local infliximab concentration (R = -0.64). FiCP discriminated active from quiescent fistulas (AUC 0.843; 95% CI 0.758-0.928) and retained diagnostic value in patients with normal CDAI scores. FiCP levels declined significantly after therapy and showed greater responsiveness than PDAI and TFV. FiCP represents a minimally invasive, tract-specific biomarker that mirrors local inflammatory activity and dynamically tracks treatment response in pfCD. Multicenter validation is warranted to define its clinical utility.

Diagnostic performance of urine-based versus cervical swab human papillomavirus testing among women in a high-burden region in Northeastern, India: A prospective cross-sectional study.

Sepsis is a complex pathogen-host interaction that traditional laboratories do not easily detect. A test that rapidly assesses immune dysregulation could improve early diagnosis and clinical decision making. This study evaluates ease of application of a new host-response sepsis diagnostic test (IntelliSep; Cytovale, Inc) and its ability to affect decision making, risk stratification, and diagnostic efficacy in the emergency department (ED). This was a case vignette-based, randomized, multisite decision-impact study. We conducted the study with ED clinicians from 3 health care institutions, selected among the 5 participating institutions from a previous observational study. Participants were ED-attending physician/fellows, residents, and advanced practice clinicians. The study excluded interns. Finally, 13 clinicians at site 1, 28 at site 2, and 11 at site 3 (total, n = 52) were included. The study used retrospectively de-identified data from patient cases to generate 100 case vignettes. Each participant reviewed 10 cases with IntelliSep results and 10 cases without. The study assessed outcomes including impact of the test results on clinical decision making, confidence in decision, and diagnostic accuracy. The test supported or changed/augmented clinician decisions in 86% of cases. The presence of the test significantly increased the proportion of raters with high confidence in their ratings (from 53.1% without, to 63.1% with). Additionally, diagnostic accuracy increased, with a notable increase in sensitivity for critical bands, from 73.1% without the test to 82.8% with it (95% CI, 63.5-81.3 vs 74.6-89.1); this increase was not statistically significant (P = .06). This paper-based study demonstrates that the IntelliSep host-response sepsis test results could impact clinician decisions and may improve the quality of clinical decision making.

Evaluating 18F-FDG PET-CT for Regional Lymph Node Assessment in Advanced Upper Tract Urothelial Carcinoma.

Cervical Spine Clearance in Adult and Pediatric Trauma: A Systematic Review.

The impact of next generation sequencing studies on the diagnosis of BAP1 inactivated melanocytic tumors.

Diagnostic accuracy of contrast-enhanced mammography (CEM) for the evaluation of mammographic asymmetries: A systematic review and meta-analysis.