Objectives
Early recognition of coronavirus disease 2019 (COVID-19) severity can guide patient management. However, it is challenging to predict when COVID-19 patients will progress to critical ...illness. This study aimed to develop an artificial intelligence system to predict future deterioration to critical illness in COVID-19 patients.
Methods
An artificial intelligence (AI) system in a time-to-event analysis framework was developed to integrate chest CT and clinical data for risk prediction of future deterioration to critical illness in patients with COVID-19.
Results
A multi-institutional international cohort of 1,051 patients with RT-PCR confirmed COVID-19 and chest CT was included in this study. Of them, 282 patients developed critical illness, which was defined as requiring ICU admission and/or mechanical ventilation and/or reaching death during their hospital stay. The AI system achieved a C-index of 0.80 for predicting individual COVID-19 patients’ to critical illness. The AI system successfully stratified the patients into high-risk and low-risk groups with distinct progression risks (
p
< 0.0001).
Conclusions
Using CT imaging and clinical data, the AI system successfully predicted time to critical illness for individual patients and identified patients with high risk. AI has the potential to accurately triage patients and facilitate personalized treatment.
Key Point
• AI system can predict time to critical illness for patients with COVID-19 by using CT imaging and clinical data.
Abstract
The absence of thermalization in certain isolated many-body systems is of great fundamental interest. Many-body localization (MBL) is a widely studied mechanism for thermalization to fail in ...strongly disordered quantum systems, but it is still not understood precisely how the range of interactions affects the dynamical behavior and the existence of MBL, especially in dimensions
D
> 1. By investigating nonequilibrium dynamics in strongly disordered
D
= 2 electron systems with power-law interactions ∝ 1/
r
α
and poor coupling to a thermal bath, here we observe MBL-like, prethermal dynamics for
α
= 3. In contrast, for
α
= 1, the system thermalizes, although the dynamics is glassy. Our results provide important insights for theory, especially since we obtained them on systems that are much closer to the thermodynamic limit than synthetic quantum systems employed in previous studies of MBL. Thus, our work is a key step towards further studies of ergodicity breaking and quantum entanglement in real materials.
In recent years, the modular multilevel cascaded converters (MMCCs) have attracted interest for potential applications in utility-scale photovoltaic and battery energy storage systems. However, the ...large numbers of cascaded modules increase the difficulty of implementation in a centralized control architecture. This paper presents a distributed control technique for three-phase MMCCs, which can significantly reduce complexity and increases the flexibility to expand the system. In addition, reference voltage analysis is performed to better understand the operation of the proposed method. Experimental results carried out on a seven-level star-connected cascaded converter are included to validate the proposed approach.
Abstract
Deep learning (DL) is one of the fastest-growing topics in materials data science, with rapidly emerging applications spanning atomistic, image-based, spectral, and textual data modalities. ...DL allows analysis of unstructured data and automated identification of features. The recent development of large materials databases has fueled the application of DL methods in atomistic prediction in particular. In contrast, advances in image and spectral data have largely leveraged synthetic data enabled by high-quality forward models as well as by generative unsupervised DL methods. In this article, we present a high-level overview of deep learning methods followed by a detailed discussion of recent developments of deep learning in atomistic simulation, materials imaging, spectral analysis, and natural language processing. For each modality we discuss applications involving both theoretical and experimental data, typical modeling approaches with their strengths and limitations, and relevant publicly available software and datasets. We conclude the review with a discussion of recent cross-cutting work related to uncertainty quantification in this field and a brief perspective on limitations, challenges, and potential growth areas for DL methods in materials science.
Malnutrition is associated with poor clinical outcomes among hospitalized patients. However, studies linking malnutrition with poor clinical outcomes in the intensive care unit (ICU) often have ...conflicting findings due in part to the inappropriate diagnosis of malnutrition. We primarily aimed to determine whether malnutrition diagnosed by validated nutrition assessment tools such as the Subjective Global Assessment (SGA) or Mini Nutritional Assessment (MNA) is independently associated with poorer clinical outcomes in the ICU and if the use of nutrition screening tools demonstrate a similar association. PubMed, CINAHL, Scopus, and Cochrane Library were systematically searched for eligible studies. Search terms included were synonyms of malnutrition, nutritional status, screening, assessment, and intensive care unit. Eligible studies were case-control or cohort studies that recruited adults in the ICU; conducted the SGA, MNA, or used nutrition screening tools before or within 48 hours of ICU admission; and reported the prevalence of malnutrition and relevant clinical outcomes including mortality, length of stay (LOS), and incidence of infection (IOI). Twenty of 1168 studies were eligible. The prevalence of malnutrition ranged from 38% to 78%. Malnutrition diagnosed by nutrition assessments was independently associated with increased ICU LOS, ICU readmission, IOI, and the risk of hospital mortality. The SGA clearly had better predictive validity than the MNA. The association between malnutrition risk determined by nutrition screening was less consistent. Malnutrition is independently associated with poorer clinical outcomes in the ICU. Compared with nutrition assessment tools, the predictive validity of nutrition screening tools were less consistent.
A defining feature of resident gut macrophages is their high replenishment rate from blood monocytes attributed to tonic commensal stimulation of this site. In contrast, almost all other tissues ...contain locally maintained macrophage populations, which coexist with monocyte-replenished cells at homeostasis. In this study, we identified three transcriptionally distinct mouse gut macrophage subsets that segregate based on expression of Tim-4 and CD4. Challenging current understanding, Tim-4
CD4
gut macrophages were found to be locally maintained, while Tim-4
CD4
macrophages had a slow turnover from blood monocytes; indeed, Tim-4
CD4
macrophages were the only subset with the high monocyte-replenishment rate currently attributed to gut macrophages. Moreover, all macrophage subpopulations required live microbiota to sustain their numbers, not only those derived from blood monocytes. These findings oppose the prevailing paradigm that all macrophages in the adult mouse gut rapidly turn over from monocytes in a microbiome-dependent manner; instead, these findings supplant it with a model of ontogenetic diversity where locally maintained subsets coexist with rapidly replaced monocyte-derived populations.
Human pluripotent and trophoblast stem cells have been essential alternatives to blastocysts for understanding early human development
. However, these simple culture systems lack the complexity to ...adequately model the spatiotemporal cellular and molecular dynamics that occur during early embryonic development. Here we describe the reprogramming of fibroblasts into in vitro three-dimensional models of the human blastocyst, termed iBlastoids. Characterization of iBlastoids shows that they model the overall architecture of blastocysts, presenting an inner cell mass-like structure, with epiblast- and primitive endoderm-like cells, a blastocoel-like cavity and a trophectoderm-like outer layer of cells. Single-cell transcriptomics further confirmed the presence of epiblast-, primitive endoderm-, and trophectoderm-like cells. Moreover, iBlastoids can give rise to pluripotent and trophoblast stem cells and are capable of modelling, in vitro, several aspects of the early stage of implantation. In summary, we have developed a scalable and tractable system to model human blastocyst biology; we envision that this will facilitate the study of early human development and the effects of gene mutations and toxins during early embryogenesis, as well as aiding in the development of new therapies associated with in vitro fertilization.
ABSTRACT
We present photometric and spectroscopic observations of the unusual Type Ia supernova ASASSN-18tb, including a series of Southern African Large Telescope spectra obtained over the course of ...nearly six months and the first observations of a supernova by the Transiting Exoplanet Survey Satellite. We confirm a previous observation by Kollmeier et al. showing that ASASSN-18tb is the first relatively normal Type Ia supernova to exhibit clear broad (∼1000 km s−1) H α emission in its nebular-phase spectra. We find that this event is best explained as a sub-Chandrasekhar mass explosion producing $M_{\mathrm{ Ni}} \approx 0.3\,\, \rm {M}_\odot$. Despite the strong H α signature at late times, we find that the early rise of the supernova shows no evidence for deviations from a single-component power-law and is best fit with a moderately shallow power law of index 1.69 ± 0.04. We find that the H α luminosity remains approximately constant after its initial detection at phase +37 d, and that the H α velocity evolution does not trace that of the Fe iii λ4660 emission. These suggest that the H α emission arises from a circumstellar medium (CSM) rather than swept-up material from a non-degenerate companion. However, ASASSN-18tb is strikingly different from other known CSM-interacting Type Ia supernovae in a number of significant ways. Those objects typically show an H α luminosity two orders of magnitude higher than what is seen in ASASSN-18tb, pushing them away from the empirical light-curve relations that define ‘normal’ Type Ia supernovae. Conversely, ASASSN-18tb exhibits a fairly typical light curve and luminosity for an underluminous or transitional SN Ia, with MR ≈ −18.1 mag. Moreover, ASASSN-18tb is the only SN Ia showing H α from CSM interaction to be discovered in an early-type galaxy.
We report the discovery of ASASSN-15lh (SN 2015L), which we interpret as the most luminous supernova yet found. At redshift z = 0.2326, ASASSN-15lh reached an absolute magnitude of Mu,AB = −23.5 ± ...0.1 and bolometric luminosity Lbol = (2.2 ± 0.2) × 10⁴⁵ ergs s⁻¹, which is more than twice as luminous as any previously known supernova. It has several major features characteristic of the hydrogen-poor super-luminous supernovae (SLSNe-l), whose energy sources and progenitors are currently poorly understood. In contrast to most previously known SLSNe-l that reside in star-forming dwarf galaxies, ASASSN-15lh appears to be hosted by a luminous galaxy (MK ≈ −25.5) with little star formation. In the 4 months since first detection, ASASSN-15lh radiated (1.1 ± 0.2) × 10⁵² ergs, challenging the magnetar model for its engine.