Citations

Background and motivation: Lung computed tomography (CT) techniques are high-resolution and are well adopted in the intensive care unit (ICU) for COVID-19 disease control classification. Most artificial intelligence (AI) systems do not undergo generalization and are typically overfitted. Such trained AI systems are not practical for clinical settings and therefore do not give accurate results when executed on unseen data sets. We hypothesize that ensemble deep learning (EDL) is superior to deep transfer learning (TL) in both non-augmented and augmented frameworks. Methodology: The system consists of a cascade of quality control, ResNet–UNet-based hybrid deep learning for lung segmentation, and seven models using TL-based classification followed by five types of EDL’s. To prove our hypothesis, five different kinds of data combinations (DC) were designed using a combination of two multicenter cohorts—Croatia (80 COVID) and Italy (72 COVID and 30 controls)—leading to 12,000 CT slices. As part of generalization, the system was tested on unseen data and statistically tested for reliability/stability. Results: Using the K5 (80:20) cross-validation protocol on the balanced and augmented dataset, the five DC datasets improved TL mean accuracy by 3.32%, 6.56%, 12.96%, 47.1%, and 2.78%, respectively. The five EDL systems showed improvements in accuracy of 2.12%, 5.78%, 6.72%, 32.05%, and 2.40%, thus validating our hypothesis. All statistical tests proved positive for reliability and stability. Conclusion: EDL showed superior performance to TL systems for both (a) unbalanced and unaugmented and (b) balanced and augmented datasets for both (i) seen and (ii) unseen paradigms, validating both our hypotheses.

Abstract Objectives We aimed to explore current practice and interregional differences in the treatment of idiopathic inflammatory myopathies (IIMs). We triangulated these observations considering countries' gross national income (GNI), disease subtypes, and symptoms using patient-reported information. Methods A cross-sectional ancillary analysis of the ‘COVID-19 vaccination in auto-immune disease’ (COVAD) e-survey containing demographic characteristics, IIM subtypes (DM, PM, IBM, anti-synthetase syndrome [ASSD], immune-mediated necrotizing myopathy [IMNM], overlap myopathies [OM]), current symptoms (surrogate for organ involvement) and treatments (corticosteroids [CS], immunomodulators [IM], i.e. antimalarials, immunosuppressants [IS], IVIG, biologic treatments and targeted-synthetic small molecules). Treatments were presented descriptively according to continents, GNI, IIM and organ involvement, and associated factors were analysed using multivariable binary logistic regressions. Results Of 18 851 respondents from 94 countries, 1418 with IIM were analysed (age 61 years, 62.5% females). DM (32.4%), IBM (24.5%) and OM (15.8%) were the most common subtypes. Treatment categories included IS (49.4%), CS (38.5%), IM (13.8%) and IVIG (9.4%). Notably, treatments varied across regions, GNI categories (IS mostly used in higher-middle income, IM in lower-middle income, IVIG and biologics largely limited to high-income countries), IIM subtypes (IS and CS associated with ASSD, IM with OM and DM, IVIG with IMNM, and biologic treatments with OM and ASSD) and disease manifestations (IS and CS with dyspnoea). Most inter-regional treatment disparities persisted after multivariable analysis. Conclusion We identified marked regional treatment disparities in a global cohort of IIM. These observations highlight the need for international consensus-driven management guidelines considering patient-centred care and available resources.

Abstract Pericardial effusions can result in acute haemodynamic compromise and require rapid intervention. Understanding pericardial restraint is essential to determine the approach to newly identified pericardial effusions in the intensive care unit. As pericardial effusions stretch the pericardium, the pericardial compliance reserve is eventually exhausted, with an exponential rise in compressive pericardial pressure. The severity of pericardial pressure increase depends on both the rapidity and volume of pericardial fluid accumulation. This increase in pericardial pressure is reflected in an increase in measured left- and right-sided ‘filling’ pressures, but paradoxically left ventricular end-diastolic volume (the true left ventricular preload) is decreased. This uncoupling of filling pressures and preload is the hallmark of pericardial restraint. When this occurs acutely from a pericardial effusion, rapid recognition and pericardiocentesis can be lifesaving. In this review, we will discuss the haemodynamics and pathophysiology of acute pericardial effusions, provide a physiological guide to determine the need for pericardiocentesis in acute care, and discuss important caveats to management.

AbstractTo develop a prognostic model for patients undergoing allogeneic hematopoietic cell transplantation (allo-HCT) for myelofibrosis (MF), we examined the data of 623 patients undergoing allo-HCT between 2000 and 2016 in the United States (the Center for International Blood and Marrow Transplant Research [CIBMTR] cohort). A Cox multivariable model was used to identify factors prognostic of mortality. A weighted score using these factors was assigned to patients who received transplantation in Europe (the European Bone Marrow Transplant [EBMT] cohort; n = 623). Patient age >50 years (hazard ratio [HR], 1.39; 95% confidence interval [CI], 0.98-1.96), and HLA-matched unrelated donor (HR, 1.29; 95% CI, 0.98-1.7) were associated with an increased hazard of death and were assigned 1 point. Hemoglobin levels <100 g/L at time of transplantation (HR, 1.63; 95% CI, 1.2-2.19) and a mismatched unrelated donor (HR, 1.78; 95% CI, 1.25-2.52) were assigned 2 points. The 3-year overall survival (OS) in patients with a low (1-2 points), intermediate (3-4 points), and high score (5 points) were 69% (95% CI, 61-76), 51% (95% CI, 46-56.4), and 34% (95% CI, 21-49), respectively (P < .001). Increasing score was predictive of increased transplant-related mortality (TRM; P = .0017) but not of relapse (P = .12). The derived score was predictive of OS (P < .001) and TRM (P = .002) but not of relapse (P = .17) in the EBMT cohort as well. The proposed system was prognostic of survival in 2 large cohorts, CIBMTR and EBMT, and can easily be applied by clinicians consulting patients with MF about the transplantation outcomes.

Background: Obstructive sleep apnea (OSA) increases the risk of stroke and cardiovascular diseases. However, its impact on geriatric patients with a prior history of stroke/transient ischemic attack (TIA) has not been adequately studied. Methods: We utilized the 2019 National Inpatient Sample in the US to identify geriatric patients with OSA (G-OSA) who had a prior history of stroke/TIA. We then compared subsequent stroke (SS) rates among sex and race subgroups. We also compared the demographics and comorbidities of SS+ and SS− groups and utilized logistic regression models to assess outcomes. Results: Out of 133,545 G-OSA patients admitted with a prior history of stroke/TIA, 4.9% (6520) had SS. Males had a higher prevalence of SS, while Asian-Pacific Islanders and Native Americans had the highest prevalence of SS, followed by Whites, Blacks, and Hispanics. The SS+ group had higher all-cause in-hospital mortality rates, with Hispanics showing the highest rate compared to Whites and Blacks (10.6% vs. 4.9% vs. 4.4%, p < 0.001), respectively. Adjusted analysis for covariates showed that complicated and uncomplicated hypertension (aOR 2.17 [95% CI 1.78–2.64]; 3.18 [95% CI 2.58–3.92]), diabetes with chronic complications (aOR 1.28 [95% CI 1.08–1.51]), hyperlipidemia (aOR 1.24 [95% CI 1.08–1.43]), and thyroid disorders (aOR 1.69 [95% CI 1.14–2.49]) were independent predictors of SS. The SS+ group had fewer routine discharges and higher healthcare costs. Conclusions: Our study shows that about 5% of G-OSA patients with a prior history of stroke/TIA are at risk of hospitalization due to SS, which is associated with higher mortality and healthcare utilization. Complicated and uncomplicated hypertension, diabetes with chronic complications, hyperlipidemia, thyroid disorders, and admission to rural hospitals predict subsequent stroke.

Abstract Objectives Disease flares in the post–coronavirus disease 2019 (COVID-19) vaccination period represent a prominent concern, though risk factors are poorly understood. We studied these flares among patients with idiopathic inflammatory myopathies (IIMs) and other autoimmune rheumatic diseases (AIRDs). Methods The COVAD-1 and -2 global surveys were circulated in early 2021 and 2022, respectively, and we captured demographics, comorbidities, AIRDs details, COVID-19 infection history and vaccination details. Flares of IIMs were defined as (a) patient self-reported, (b) immunosuppression (IS) denoted, (c) clinical sign directed and (d) with &gt;7.9-point minimal clinically significant improvement difference worsening of Patient-Reported Outcomes Measurement Information System (PROMIS) PROMISPF10a score. Risk factors of flares were analysed using regression models. Results Of 15 165 total respondents, 1278 IIMs (age 63 years, 70.3% female, 80.8% Caucasians) and 3453 AIRDs were included. Flares of IIM were seen in 9.6%, 12.7%, 8.7% and 19.6% patients by definitions (a) to (d), respectively, with a median time to flare of 71.5 (10.7–235) days, similar to AIRDs. Patients with active IIMs pre-vaccination (OR 1.2; 95% CI 1.03, 1.6, P = 0.025) were prone to flares, while those receiving rituximab (OR 0.3; 95% CI 0.1, 0.7, P = 0.010) and AZA (OR 0.3, 95% CI 0.1, 0.8, P = 0.016) were at lower risk. Female gender and comorbidities predisposed to flares requiring changes in IS. Asthma (OR 1.62; 95% CI 1.05, 2.50, P = 0.028) and higher pain visual analogue score (OR 1.19; 95% CI 1.11, 1.27, P &lt; 0.001) were associated with disparity between self-reported and IS-denoted flares. Conclusion A diagnosis of IIMs confers an equal risk of flares in the post–COVID-19 vaccination period to AIRDs, with active disease, female gender and comorbidities conferring a higher risk. Disparity between patient- and physician-reported outcomes represents a future avenue for exploration.

INTRODUCTION: High-resolution manometry (HRM) and functional lumen imaging probe (FLIP) are primary and/or complementary diagnostic tools for the evaluation of esophageal motility. We aimed to assess the interrater agreement and accuracy of HRM and FLIP interpretations. METHODS: Esophageal motility specialists from multiple institutions completed the interpretation of 40 consecutive HRM and 40 FLIP studies. Interrater agreement was assessed using intraclass correlation coefficient (ICC) for continuous variables and Fleiss' κ statistics for nominal variables. Accuracies of rater interpretation were assessed using the consensus of 3 experienced raters as the reference standard. RESULTS: Fifteen raters completed the HRM and FLIP studies. An excellent interrater agreement was seen in supine median integral relaxation pressure (ICC 0.96, 95% confidence interval 0.95–0.98), and a good agreement was seen with the assessment of esophagogastric junction (EGJ) outflow, peristalsis, and assignment of a Chicago Classification version 4.0 diagnosis using HRM (κ = 0.71, 0.75, and 0.70, respectively). An excellent interrater agreement for EGJ distensibility index and maximum diameter (0.91 [0.90–0.94], 0.92 [0.89–0.95]) was seen, and a moderate-to-good agreement was seen in the assignment of EGJ opening classification, contractile response pattern, and motility classification (κ = 0.68, 0.56, and 0.59, respectively) on FLIP. Rater accuracy for Chicago Classification version 4.0 diagnosis on HRM was 82% (95% confidence interval 78%–84%) and for motility diagnosis on FLIP Panometry was 78% (95% confidence interval 72%–81%). DISCUSSION: Our study demonstrates high levels of interrater agreement and accuracy in the interpretation of HRM and FLIP metrics and moderate-to-high levels for motility classification in FLIP, supporting the use of these approaches for primary or complementary evaluation of esophageal motility disorders.

Abstract Background Treatment patterns for intrahepatic cholangiocarcinoma (ICC) and extrahepatic cholangiocarcinoma (ECC) differ, but limited studies exist comparing them. This study examines differences in molecular profiling rates and treatment patterns in these populations, focusing on use of adjuvant, liver-directed, targeted, and investigational therapies. Methods This multicenter collaboration included patients with ICC or ECC treated at 1 of 8 participating institutions. Retrospective data were collected on risk factors, pathology, treatments, and survival. Comparative statistical tests were 2-sided. Results Among 1039 patients screened, 847 patients met eligibility (ICC = 611, ECC = 236). Patients with ECC were more likely than those with ICC to present with early stage disease (53.8% vs 28.0%), undergo surgical resection (55.1% vs 29.8%), and receive adjuvant chemoradiation (36.5% vs 4.2%) (all P &lt; .00001). However, they were less likely to undergo molecular profiling (50.3% vs 64.3%) or receive liver-directed therapy (17.9% vs 35.7%), targeted therapy (4.7% vs 18.9%), and clinical trial therapy (10.6% vs 24.8%) (all P &lt; .001). In patients with recurrent ECC after surgery, the molecular profiling rate was 64.5%. Patients with advanced ECC had a shorter median overall survival than those with advanced ICC (11.8 vs 15.1 months; P &lt; .001). Conclusions Patients with advanced ECC have low rates of molecular profiling, possibly in part because of insufficient tissue. They also have low rates of targeted therapy use and clinical trial enrollment. While these rates are higher in advanced ICC, the prognosis for both subtypes of cholangiocarcinoma remains poor, and a pressing need exists for new effective targeted therapies and broader access to clinical trials.