We demonstrate use of iteratively pruned deep learning model ensembles for detecting pulmonary manifestation of COVID-19 with chest X-rays. This disease is caused by the novel Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) virus, also known as the novel Coronavirus (2019-nCoV). A custom convolutional neural network and a selection of ImageNet pretrained models are trained and evaluated at patient-level on publicly available CXR collections to learn modality-specific feature representations. The learned knowledge is transferred and fine-tuned to improve performance and generalization in the related task of classifying CXRs as normal, showing bacterial pneumonia, or COVID-19-viral abnormalities. The best performing models are iteratively pruned to reduce complexity and improve memory efficiency. The predictions of the best-performing pruned models are combined through different ensemble strategies to improve classification performance. Empirical evaluations demonstrate that the weighted average of the best-performing pruned models significantly improves performance resulting in an accuracy of 99.01% and area under the curve of 0.9972 in detecting COVID-19 findings on CXRs. The combined use of modality-specific knowledge transfer, iterative model pruning, and ensemble learning resulted in improved predictions. We expect that this model can be quickly adopted for COVID-19 screening using chest radiographs.
BACKGROUND & AIMS: Vedolizumab is a gut-selective monoclonal antibody for the treatment of moderately to severely active Crohn's disease (CD). We performed a prospective study of endoscopic, radiologic, and histologic healing in patients with CD who received vedolizumab therapy. METHODS: We performed a phase 3b, open-label, single-group study of 101 patients with at least 3 months of active CD (a CD Activity Index [CDAI] score of 220-450, a simple endoscopic score for CD [SES-CD] of 7 or more, 1 or more mucosal ulcerations [identified by endoscopy], and failure of conventional therapy) from March 2015 through December 2017. Among the patients enrolled, 54.5% had previous failure of 1 or more tumor necrosis factor (TNF) antagonists and 44.6% had severe endoscopic disease activity (SES-CD scores above 15) at baseline. Participants received vedolizumab (300 mg intravenously) at weeks 0, 2, and 6, and then every 8 weeks thereafter, for 26 weeks (primary study) or 52 weeks (substudy, 56 patients). The primary endpoint at week 26 was endoscopic remission (SES-CD score of 4 or less); other endpoints included endoscopic response (50% reduction in SES-CD), radiologic remission (magnetic resonance index of activity score below 7), and histologic response (modified global histologic disease activity score of 4 or less). RESULTS: At week 26, 11.9% of patients were in endoscopic remission (95% confidence interval [CI] 6.3-9.8); at week 52, 17.9% of the patients were in endoscopic remission (95% CI 8.9-30.4). Higher proportions of patients naïve to TNF antagonists achieved endoscopic remission than patients with TNF-antagonist-failure at weeks 26 and 52. Higher proportion of patients with moderate CD (SES-CD scores, 7-15) achieved endoscopic remission at weeks 26 and 52 than patients with severe CD (SES-CD scores above 15). The proportion of patients with complete mucosal healing increased over time, with greater rates of healing in the colon than in the ileum. Remission was detected by magnetic resonance enterography in 21.9% of patients at week 26 (95% CI 9.3-40.0) and in 38.1% at week 52 (95% CI 18.1-61.6). At week 26, 24.4% of patients had a histologic response in the colon (95% CI 15.3-35.4) and 28.3% of patients had a histologic response in the ileum (95% CI 17.5-41.4). At week 52, 20.5% of patients had a histologic response in the colon (95% CI 9.8-35.3) and 34.3% of patients had a histologic response in the ileum (95% CI 19.1-52.2). There were no notable safety issues, including worsening of extraintestinal manifestations. CONCLUSIONS: In a phase 3b trial, we found that 26 and 52 weeks of treatment with vedolizumab (300 mg, at weeks 0, 2, and 6, and then every 8 weeks thereafter) induces endoscopic, radiologic, and histologic healing in patients with moderately to severely active CD. ClinicalTrials.gov no: NCT02425111.
In some patients with treatment-refractory pulmonary nontuberculous mycobacterial disease, the addition of inhaled amikacin was associated with microbiologic and/or symptomatic improvement; however, toxicity was common. Prospective evaluation of inhaled amikacin for mycobacterial disease is warranted.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.