Search

BackgroundCOVID-19 has overwhelmed health systems worldwide. It is important to identify severe cases as early as possible, such that resources can be mobilized and treatment can be escalated. ObjectiveThis study aims to develop a machine learning approach for automated severity assessment of COVID-19 based on clinical and imaging data. MethodsClinical data—including demographics, signs, symptoms, comorbidities, and blood test results—and chest computed tomography scans of 346 patients from 2 hospitals in the Hubei Province, China, were used to develop machine learning models for automated severity assessment in diagnosed COVID-19 cases. We compared the predictive power of the clinical and imaging data from multiple machine learning models and further explored the use of four oversampling methods to address the imbalanced classification issue. Features with the highest predictive power were identified using the Shapley Additive Explanations framework. ResultsImaging features had the strongest impact on the model output, while a combination of clinical and imaging features yielded the best performance overall. The identified predictive features were consistent with those reported previously. Although oversampling yielded mixed results, it achieved the best model performance in our study. Logistic regression models differentiating between mild and severe cases achieved the best performance for clinical features (area under the curve [AUC] 0.848; sensitivity 0.455; specificity 0.906), imaging features (AUC 0.926; sensitivity 0.818; specificity 0.901), and a combination of clinical and imaging features (AUC 0.950; sensitivity 0.764; specificity 0.919). The synthetic minority oversampling method further improved the performance of the model using combined features (AUC 0.960; sensitivity 0.845; specificity 0.929). ConclusionsClinical and imaging features can be used for automated severity assessment of COVID-19 and can potentially help triage patients with COVID-19 and prioritize care delivery to those at a higher risk of severe disease.

Non-line-of-sight (NLOS) imaging is a novel radar sensing technology that enables the reconstruction of hidden targets. However, it may suffer from synthetic aperture length reduction caused by ambient occlusion. In this study, a complex total variation (CTV) regularization-based sparse reconstruction method for NLOS three-dimensional (3-D) imaging by millimeter-wave (mmW) radar, named RM-CSTV method, is proposed to improve imaging quality and speed. In this scheme, the NLOS imaging model is first introduced, and associated geometric constraints for NLOS objects are established. Second, an effective high-resolution NLOS imaging method based on the range migration (RM) kernel and complex sparse joint total variation constraint, dubbed as modified RM-CSTV, is proposed for 3-D high-resolution imaging with edge information. The experiments with multi-type NLOS targets show that the proposed RM-CSTV method can provide effective and high-resolution NLOS targets 3-D imaging.

Sludge anaerobic digestion promoted by enzymes has the advantages of mild reaction, high efficiency and environmental protection etc. However, the difficulty of recovery and poor catalytic stability seriously restrict its application. Enzyme immobilization is the key to solve the above problems. A new type of polyvinyl alcohol(PVA) carrier was prepared by Noria and polyethylenimide(PEI) co-deposition. The effect of lysozyme immobilized with Noria-PEI modified PVA carriers on the efficiency of sludge anaerobic fermentation for VFAs production was studied. The results showed that immobilized lysozyme could enhance sludge fermentation to promote SCOD, and SCOD reached the highest concentration of 2 892.4 mg/L within 36 h, which was 3.4 times higher than in the blank(day 6). VFAs concentration peaked at 2 130.8 mg/L on day 4, mainly comprised of acetic acid and propionic acid, accounting for 61.8%-80.7%. The average particle size and specific surface area of sludge in the immobilized enzyme system were 42.957 μm and 0.413 m2/g, indicating that immobilized enzyme enhanced the hydrolysis of granular sludge. Microbial community analysis showed that the hydrolytic acid-producing phyla of Firmicutes, Proteobacteria and Bacteroides accounted for high proportion in the immobilized enzyme system. At the genus level, Macellibacteroides, Petrimona, Lactobacillu, and Clostridium_sensu_stricto_1 were the main genera, which revealed the mechanism of enhanced VFAs production by the immobilized enzyme at microscopic level.

Live streaming is revolutionizing the landscape of e-commerce, creating new opportunities for platforms and e-tailers to improve their performance. However, little is known about the underlying mechanisms that shape consumer behavior in this burgeoning business phenomenon. This study aims to shed light on the relationships between environmental cues generated by live streaming and online impulse buying. Drawing upon the Stimulus-Organism-Response framework, a comprehensive model was formulated to explore how social cues (streamer interaction, peer interaction) and media cues (vividness, realness) impact pleasure, arousal, perceived uncertainty, and subsequently induce consumers’ urge to buy impulsively. The model was tested by survey data from 403 consumers. SPSS and PLS are employed to verify the model. The findings revealed that realness and streamer interaction can reduce perceived uncertainty and foster a pleasant consumer experience, while vividness and peer interaction serve to awaken and delight consumers. Pleasure, arousal, and perceived uncertainty mediate antecedent variables’ effects on urge to buy impulsively in a parallel and reverse way, and emotions exert a more powerful influence. This study enriched the research on the influence mechanisms of impulse buying driven by live streaming and provided suggestions for platforms and streamers to optimize product display and guide interaction, which is conducive to leveraging the advantages of live streaming and creating greater commercial value.