Search

Background To assess the possible impact of antiretroviral therapy improvements, aging, and comorbidities, we examined trends in all-cause and cause-specific hospitalization rates among persons with HIV (PWH) from 2005 to 2015. Methods In 6 clinical cohorts, we followed PWH in care (≥1 outpatient CD4 count or HIV load [VL] every 12 months) and categorized ICD codes of primary discharge diagnoses using modified Clinical Classifications Software. Poisson regression estimated hospitalization rate ratios for calendar time trends, adjusted for demographics, HIV risk factor, and annually updated age, CD4, and VL. Results Among 28 057 patients (125 724 person-years), from 2005 to 2015, the median CD4 increased from 389 to 580 cells/µL and virologic suppression from 55% to 85% of patients. Unadjusted all-cause hospitalization rates decreased from 22.3 per 100 person-years in 2005 (95% confidence interval [CI], 20.6–24.1) to 13.0 in 2015 (95% CI, 12.2–14.0). Unadjusted rates decreased for almost all diagnostic categories. Adjusted rates decreased for all-cause, cardiovascular, and AIDS-defining conditions, increased for non-AIDS–defining infection, and were stable for most other categories. Conclusions Among PWH with increasing CD4 counts and viral suppression, unadjusted hospitalization rates decreased for all-cause and most cause-specific hospitalizations, despite the potential effects of aging, comorbidities, and cumulative exposure to HIV and antiretrovirals.

This document was prepared by the MicroBooNE Collaboration using the resources of the Fermi National Accelerator Laboratory (Fermilab), a U.S. Department of Energy, Office of Science, HEP User Facility. Fermilab is managed by Fermi Research Alliance, LLC (FRA), acting under Contract No. DE-AC02-07CH11359. MicroBooNE is supported by the following: the U.S. Department of Energy, Office of Science, Offices of High Energy Physics and Nuclear Physics; the U.S. National Science Foundation; the Swiss National Science Foundation; the Science and Technology Facilities Council (STFC), part of the United Kingdom Research and Innovation; and The Royal Society (United Kingdom). Additional support for the laser calibration system and cosmic ray tagger was provided by the Albert Einstein Center for Fundamental Physics, Bern, Switzerland., We present the multiple particle identification (MPID) network, a convolutional neural network for multiple object classification, developed by MicroBooNE. MPID provides the probabilities that an interaction includes an e(-), gamma, mu(-), pi(+/-), and protons in a liquid argon time projection chamber single readout plane. The network extends the single particle identification network previously developed by MicroBooNE [Convolutional neural networks applied to neutrino events in a liquid argon time projection chamber, R. Acciarri et al. J. Instrum. 12, P03011 (2017)]. MPID takes as input an image either cropped around a reconstructed interaction vertex or containing only activity connected to a reconstructed vertex, therefore relieving the tool from inefficiencies in vertex finding and particle clustering. The network serves as an important component in MicroBooNE's deep-learning-based.e search analysis. In this paper, we present the network's design, training, and performance on simulation and data from the MicroBooNE detector., Fermi Research Alliance, LLC (FRA) DE-AC02-07CH11359, United States Department of Energy (DOE), National Science Foundation (NSF), Swiss National Science Foundation (SNSF) European Commission, Science and Technology Facilities Council (STFC), United Kingdom Research and Innovation, Royal Society of London

We present the performance of a semantic segmentation network, SparseSSNet, that provides pixel-level classification of MicroBooNE data. The MicroBooNE experiment employs a liquid argon time projection chamber for the study of neutrino properties and interactions. SparseSSNet is a submanifold sparse convolutional neural network, which provides the initial machine learning based algorithm utilized in one of MicroBooNE's $\nu_e$-appearance oscillation analyses. The network is trained to categorize pixels into five classes, which are re-classified into two classes more relevant to the current analysis. The output of SparseSSNet is a key input in further analysis steps. This technique, used for the first time in liquid argon time projection chambers data and is an improvement compared to a previously used convolutional neural network, both in accuracy and computing resource utilization. The accuracy achieved on the test sample is $\geq 99\%$. For full neutrino interaction simulations, the time for processing one image is $\approx$ 0.5 sec, the memory usage is at 1 GB level, which allows utilization of most typical CPU worker machine.

Epigenetic differences may help to distinguish between PTSD cases and trauma-exposed controls. Here, we describe the results of the largest DNA methylation meta-analysis of PTSD to date. Ten cohorts, military and civilian, contribute blood-derived DNA methylation data from 1,896 PTSD cases and trauma-exposed controls. Four CpG sites within the aryl-hydrocarbon receptor repressor (AHRR) associate with PTSD after adjustment for multiple comparisons, with lower DNA methylation in PTSD cases relative to controls. Although AHRR methylation is known to associate with smoking, the AHRR association with PTSD is most pronounced in non-smokers, suggesting the result was independent of smoking status. Evaluation of metabolomics data reveals that AHRR methylation associated with kynurenine levels, which are lower among subjects with PTSD. This study supports epigenetic differences in those with PTSD and suggests a role for decreased kynurenine as a contributor to immune dysregulation in PTSD., PTSD has been associated with DNA methylation of specific loci in the genome, but studies have been limited by small sample sizes. Here, the authors perform a meta-analysis of DNA methylation data from 10 different cohorts and identify CpGs in AHRR that are associated with PTSD.

The NASA Advanced Composites Project (ACP) was formed to create new and innovative solutions for reducing the timeline for developing and certifying state-of-the-art composite structures and materials. Under the ACP, the manufacturing tasks focus on reducing the occurrence of defects during Automated Fiber Placement (AFP) and helping close the design-analysis-manufacturing loop through Design for Manufacturing (DFM). One goal of this task is to leverage and enhance commercial off-the-shelf AFP path simulation software (PSS) by creating a link to the digital design process and establishing a feedback loop for composite designers. The implementation of these enhancement criteria targets two main categories – analysis tools and improvements to the path generation algorithms. To improve the manufacturing process, there must be a transition in the mindset behind programming a composite part from the traditional ply-by-ply approach to a laminate-focused strategy. Having a clear understanding of the precise defect placement through the thickness of a laminate can lead to a reduction in the required strength margins and potential optimization of the strength and weight of the final part. Additionally, a better understanding of the stack-up of defects through the thickness of the part, and the optimization of their placement can lead to improved surface quality that can be critical for bonded or fastened surfaces. For this task, CGTech’s VERICUT Composite Programming (VCP) was selected through this consortium as the AFP PSS of choice to support the DFM methodology. The results of this development have included the creation of several path and defect analysis, automation, and export routines, ply- and laminate-level analysis reports, and the formation of the Laminate Manager (LM), a powerful tool for visualizing, programming, and analyzing a composite laminate through a single interface. The LM has facilitated a broader transformation in the overall approach to programming a composite part by transitioning to a laminate-based focus and allowing a programmer/designer to visualize, analyze, and generate each ply based on its interactions with the rest of the laminate. This transformation in approach can have a tremendous impact on the future of automated composite manufacturing by not only improving the overall quality of composite parts but also significantly streamlining the development cycle, reaching optimal designs in a shorter and more efficient timeframe.

Liquid argon time projection chambers (LArTPCs) are now a standard detector technology for making accelerator neutrino measurements, due to their high material density, precise tracking, and calorimetric capabilities. An electric field (E-field) is required in such detectors to drift ionization electrons to the anode where they are collected. The E-field of a TPC is often approximated to be uniform between the anode and the cathode planes. However, significant distortions can appear from effects such as mechanical deformations, electrode failures, or the accumulation of space charge generated by cosmic rays. The latter effect is particularly relevant for detectors placed near the Earth's surface and with large drift distances and long drift time. To determine the E-field in situ, an ultraviolet (UV) laser system is installed in the MicroBooNE experiment at Fermi National Accelerator Laboratory. The purpose of this system is to provide precise measurements of the E-field, and to make it possible to correct for 3D spatial distortions due to E-field non-uniformities. Here we describe the methodology developed for deriving spatial distortions, the drift velocity and the E-field from UV-laser measurements.

Neuroevolution is a powerful and general technique for evolving the structure and weights of artificial neural networks. Though neuroevolutionary approaches such as NeuroEvolution of Augmenting Topologies (NEAT) have been successfully applied to various problems including classification, regression, and reinforcement learning problems, little work has explored application of these techniques to larger-scale multiclass classification problems. In this paper, NEAT is evaluated in several multiclass classification problems, and then extended via two ensemble approaches: One-vs-All and One-vs-One. These approaches decompose multiclass classification problems into a set of binary classification problems, in which each binary problem is solved by an instance of NEAT. These ensemble models exhibit reduced variance and increasingly superior accuracy as the number of classes increases. Additionally, higher accuracy is achieved early in training, even when artificially constrained for the sake of fair comparison with standard NEAT. However, because the approach can be trivially distributed, it can be applied quickly at large scale to solve real problems. In fact, these approaches are incorporated into DarwinTM, an enterprise automatic machine learning solution that also incorporates various other algorithmic enhancements to NEAT. The resulting complete system has proven robust to a wide variety of client datasets.