Your browser doesn't support javascript.
Show: 20 | 50 | 100
Results 1 - 8 de 8
Filter
1.
2nd International Conference on Artificial Intelligence and Computer Engineering, ICAICE 2021 ; : 216-220, 2021.
Article in English | Scopus | ID: covidwho-1948770

ABSTRACT

China is the world's largest pork production and consumption country, with the improvement of people's living standards and consumption upgrade, people's demand for fresh pork and other fresh products is stronger. With the outbreak of African Swine Fever and COVID-19 in China in the past two years, cold chain transportation of pork will replace live pigs as the main mode of pork supply chain. As one of the most important branches of machine learning, deep learning has developed rapidly in recent years and attracted extensive attention at home and abroad. In order to improve the real-time detection of pork freshness, this paper experimented with a variety of deep learning frameworks to achieve pork freshness classification. In this paper, pork freshness is divided into 5 levels according to TVB-N content, and the pictures taken are trained by different deep learning networks, including VGG, GoogLeNet and RestNet. After analyzing the training situation of each network, the advantages of different networks are absorbed and a new improved neural network is built to predict pork freshness. The final classification accuracy reached 97%, Indicating that this is a very efficient and accurate pork freshness classification method. © 2021 IEEE.

2.
Neurology ; 98(18 SUPPL), 2022.
Article in English | EMBASE | ID: covidwho-1925289

ABSTRACT

Objective: To further characterize the relationship between markers of inflammation and outcome in patients undergoing mechanical thrombectomy for acute stroke. Background: Inflammation and infection after ischemic stroke are known to exacerbate tissue injury and worsen clinical outcome. Thrombectomy has become standard of care in stroke, but little data exist regarding how inflammation affects outcome after thrombectomy. Design/Methods: We performed retrospective chart review of stroke patients who underwent mechanical thrombectomy at 2 tertiary academic centers between December 2018 and November 2020. The relationship between discharge mortality, admission WBC count, admission neutrophil percentage, peak WBC count, and fever (peak temperature >38°C) were analyzed using the Wilcoxon rank sum test, Student's t-test, and Fisher's exact test. Multivariable analysis was performed to test for independent predictors of discharge mortality. Analyses were performed for the entire cohort, then repeated in a cohort excluding COVIDpositive patients. Results: Of 254 patients who had thrombectomy for acute stroke, 42 (16.5%) died prior to discharge. Mortality was associated with admission WBC count (10.7 [8.9-14] vs. 8.6 [7-12], p=0.0064), admission neutrophil percentage (78% ± 11 vs. 70% ± 14, p=0.0001), peak WBC count (17 [13-22] vs. 12 [8.9-15], p<0.0001), and fever (71% vs. 29%, p<0.0001). In multivariable analysis, admission WBC count (OR 14, CI 1.5-158, p=0.024), neutrophil percentage (OR 1.04, CI 1.0-1.1, p=0.039), peak WBC count (OR 343, CI 27-5702, p<0.0001) and fever (OR 8.6, CI 3.6-23, p<0.0001) were significantly predictive of discharge mortality after controlling for age, admission NIHSS and post-thrombectomy ASPECTS score. Fifteen patients tested positive for COVID-19. In analyses excluding these patients, peak WBC count and fever remained independent predictors of discharge mortality. Conclusions: Elevated markers of inflammation during hospitalization predict discharge mortality in patients who undergo mechanical thrombectomy for acute stroke. Further study is warranted to investigate causation and identify opportunities to improve quality of care in this patient population.

3.
International Transactions in Operational Research ; 2022.
Article in English | Scopus | ID: covidwho-1874435

ABSTRACT

The spread of COVID-19 outbreak has promoted truck-drone delivery from trials to commercial applications in end-to-end contactless solutions. To fully integrate truck-drone delivery in contactless solutions, we introduce the robust traveling salesman problem with a drone, in which a drone makes deliveries and returns to the truck that is moving on its route under uncertainty. The challenge is to find, for each customer location in truck-drone routing, an assignment to minimize the expected makespan. Apart from the complexity of this problem, the risk of synchronization failure associated with uncertain travel time should be also considered. The problem is first formulated as a robust model, and a novel efficient frontier heuristic is proposed to solve this model. By coupling the implicit adaptive weighting with epsilon-constraint methods, the heuristic generates a series of scalarized single-objective problems, where the goal is to minimize expected makespan under the constraint of synchronization risk. The experiment results show that the robust (near-)optimal solutions offer a considerable reduction in risk, yet only hint at a small increase in makespan. The heuristic in the present study is effective to construct approximations of Pareto frontier and allows for assignment decisions in a priori or a posteriori manner. © 2022 The Authors. International Transactions in Operational Research © 2022 International Federation of Operational Research Societies.

4.
2021 IEEE Canadian Conference on Electrical and Computer Engineering, CCECE 2021 ; 2021-September, 2021.
Article in English | Scopus | ID: covidwho-1511202

ABSTRACT

As the COVID-19 pandemic continues to devastate globally, one promising field of research is machine learning-driven computer vision to streamline various parts of the COVID-19 clinical workflow. These machine learning methods are typically stand-alone models designed without consideration for the integration necessary for real-world application workflows. In this study, we take a machine learning and systems (MLSys) perspective to design a system for COVID-19 patient screening with the clinical workflow in mind. The COVID-Net system is comprised of the continuously evolving COVIDx dataset, COVID-Net deep neural network for COVID-19 patient detection, and COVID-Net S deep neural networks for disease severity scoring for COVID-19 positive patient cases. The deep neural networks within the COVID-Net system possess state-of-the-art performance, and are designed to be integrated within a user interface (UI) for clinical decision support with automatic report generation to assist clinicians in their treatment decisions. © 2021 IEEE.

5.
3rd MICCAI Workshop on Domain Adaptation and Representation Transfer, DART 2021, and the 1st MICCAI Workshop on Affordable Healthcare and AI for Resource Diverse Global Health, FAIR 2021, held in conjunction with 24th International Conference on Medical Image Computing and Computer Assisted Intervention, MICCAI 2021 ; 12968 LNCS:191-202, 2021.
Article in English | Scopus | ID: covidwho-1469665

ABSTRACT

The Coronavirus Disease 2019 (COVID-19) pandemic has impacted many aspects of life globally, and a critical factor in mitigating its effects is screening individuals for infections, thereby allowing for both proper treatment for those individuals as well as action to be taken to prevent further spread of the virus. Point-of-care ultrasound (POCUS) imaging has been proposed as a screening tool as it is a much cheaper and easier to apply imaging modality than others that are traditionally used for pulmonary examinations, namely chest x-ray and computed tomography. Given the scarcity of expert radiologists for interpreting POCUS examinations in many highly affected regions around the world, low-cost deep learning-driven clinical decision support solutions can have a large impact during the on-going pandemic. Motivated by this, we introduce COVID-Net US, a highly efficient, self-attention deep convolutional neural network design tailored for COVID-19 screening from lung POCUS images. Experimental results show that the proposed COVID-Net US can achieve an AUC of over 0.98 while achieving 353 × lower architectural complexity, 62 × lower computational complexity, and 14.3 × faster inference times on a Raspberry Pi. Clinical validation was also conducted, where select cases were reviewed and reported on by a practicing clinician (20 years of clinical practice) specializing in intensive care (ICU) and 15 years of expertise in POCUS interpretation. To advocate affordable healthcare and artificial intelligence for resource-constrained environments, we have made COVID-Net US open source and publicly available (https://github.com/maclean-alexander/COVID-Net-US/ ) as part of the COVID-Net open source initiative. © 2021, Crown.

7.
Value in Health ; 24:S58, 2021.
Article in English | EMBASE | ID: covidwho-1284275

ABSTRACT

Objective: To understand patient experiences with FL disease and treatment through SML. Method: Social media data were extracted between February 2019 and July 2020 using “Follicular Lymphoma” and related keywords via Social Studio®, an online aggregator tool for social media posts. English as well as local language posts were extracted from five countries including United States (US), Canada, United Kingdom (UK), Germany and France. Patient conversations were identified, synthesized, mapped, and analyzed to understand different concerns. Results: 487 patient posts discussing 1324 topics of conversation were identified. In most countries, top discussed topics included patient concerns such as quality of life (QoL) changes, and disease and/or treatment management. Multiple patient concerns (n=554) were observed across all geographies: impact on QoL (198), curability (73), fear of relapse/progression (64), disease/treatment information need (50), lack of emotional support (43), FL transformation to diffuse large B-cell lymphoma (42), and cost of treatment (30) were notable concerns. To assess QoL impact, patient conversations (198) were mapped to the statements in the Functional Assessment of Cancer Therapy – Lymphoma questionnaire (FACT-Lym). Pain and lack of energy (57), swollen nodes/lumps (47), and side effects of treatment (31) had impacted physical wellbeing, while support from family/friends (41) helped patients cope emotionally. A few patients (18) said that they were able to return to work after treatment. As for inter-country differences, conversations were mostly from the US (43%) and UK (20%);male patients in Germany were more active social media participants than female patients, which was different from other four countries;only patients in the UK had expressed concerns about COVID-19 impact. Conclusion: Insights from international SML research indicated concerns related to disease- and/or treatment-related impact on QoL and interest about potential cure for the disease.

8.
American Journal of Respiratory and Critical Care Medicine ; 203(9), 2021.
Article in English | EMBASE | ID: covidwho-1277756

ABSTRACT

Rationale: Similar to other human coronaviruses like MERS and SARS, severe manifestations of COVID-19 are associated with acute lung injury and sustained pulmonary dysfunction. A recent single-cell study of lung tissue from severe COVID-19 and idiopathic pulmonary fibrosis (IPF) patients suggested these diseases share common pro-fibrotic molecular pathways. To determine whether similar changes can be detected in the blood, we compared single-cell RNA-seq profiles of peripheral blood mononuclear cells (PBMCs) from patients with IPF or COVID-19, using influenza and healthy individuals as controls. Methods: 25 IPF, 18 COVID-19, and 13 healthy control PBMC samples were sequenced in our lab using 10X Genomics 5' single-cell technology. This data was processed using CellRanger and integrated with publicly available datasets of Covid-19, influenza, and healthy PBMC samples, yielding ∼300,000 single cells. Severe COVID-19 patients were treated in the ICU and succumbed to the disease, while severe IPF had transplant-free survival of fewer than three years. Downstream analysis was performed with the R package Seurat. The Louvain clustering algorithm generated 28 distinct cell clusters. Wilcoxon rank-sum test was used to determine significant cell type proportion differences and differentially expressed genes (DEGs). Significantly enriched pathways were found using EnrichR and Gene Set Enrichment Analysis (GSEA). Results: We report significantly increased platelets as a proportion of total cells in patients with severe COVID-19 (p = 0.0047) and severe IPF (p = 0.05) compared to healthy patients. Stable IPF and severe COVID-19 shared similar cell proportions of platelets (p=0.15) and monocytes (p=0.42). Across most cell types, COVID-19 and influenza patients had gene expression changes consistent with type I interferon activation while IPF patients exhibited changes in ribosomal upregulation and pro-fibrotic pathways relative to healthy controls. Using a composite pro-fibrotic score of TGFB1 targets and effectors, hierarchical clustering markedly differentiates between IPF and controls versus COVID-19 and influenza, perhaps distinctly highlighting mechanisms of disease. Within monocytes, we did not observe a significant pro-fibrotic phenotype (SPP1, MMP9, CHI3L1, PLA2G7) in samples of patients with any disease;hierarchical clustering of these genes again segregated IPF and controls from COVID-19 and influenza. Conclusions: Pro-fibrotic gene expression patterns could not be seen in PBMCs from patients with acute severe COVID-19 infection. More studies are needed in distinct COVID-19 patient populations, such as those with prolonged respiratory failure or with sustained respiratory dysfunction after recovery.

SELECTION OF CITATIONS
SEARCH DETAIL