Continuously updated forecasting of SARS-CoV-2 in a regional health system.

OBJECTIVES: To build a model of local hospital utilization resulting from SARS-CoV-2 and to continuously update it with new data. STUDY DESIGN: Retrospective analysis of real performance resulting from a model deployed in a major regional health system. METHODS: Using hospitalization data from the Kaiser Permanente Mid-Atlantic States integrated care system during the period from March 10, 2020, through December 31, 2020, and a custom-developed genetic particle filtering algorithm, we modeled the SARS-CoV-2 outbreak in the mid-Atlantic region. This model produced weekly forecasts of COVID-19-related hospital admissions, which we then compared with actual hospital admissions over the same period. RESULTS: We found that the model was able to accurately capture the data-generating process (weekly mean absolute percentage error, 10.0%-48.8%; Anderson-Darling P value of .97 when comparing percentiles of observed admissions with the uniform distribution) once the effects of social distancing could be accurately measured in mid-April. We also found that our estimates of key parameters, including the reproductive rate, were consistent with consensus literature estimates. CONCLUSIONS: The genetic particle filtering algorithm that we have proposed is effective at modeling hospitalizations due to SARS-CoV-2. The methods used by our model can be reproduced by any major health care system for the purposes of resource planning, staffing, and population care management to create an effective forecasting regimen at scale.

A comparison of attentional neural network architectures for modeling with electronic medical records.

OBJECTIVE: Attention networks learn an intelligent weighted averaging mechanism over a series of entities, providing increases to both performance and interpretability. In this article, we propose a novel time-aware transformer-based network and compare it to another leading model with similar characteristics. We also decompose model performance along several critical axes and examine which features contribute most to our model's performance. MATERIALS AND METHODS: Using data sets representing patient records obtained between 2017 and 2019 by the Kaiser Permanente Mid-Atlantic States medical system, we construct four attentional models with varying levels of complexity on two targets (patient mortality and hospitalization). We examine how incorporating transfer learning and demographic features contribute to model success. We also test the performance of a model proposed in recent medical modeling literature. We compare these models with out-of-sample data using the area under the receiver-operator characteristic (AUROC) curve and average precision as measures of performance. We also analyze the attentional weights assigned by these models to patient diagnoses. RESULTS: We found that our model significantly outperformed the alternative on a mortality prediction task (91.96% AUROC against 73.82% AUROC). Our model also outperformed on the hospitalization task, although the models were significantly more competitive in that space (82.41% AUROC against 80.33% AUROC). Furthermore, we found that demographic features and transfer learning features which are frequently omitted from new models proposed in the EMR modeling space contributed significantly to the success of our model. DISCUSSION: We proposed an original construction of deep learning electronic medical record models which achieved very strong performance. We found that our unique model construction outperformed on several tasks in comparison to a leading literature alternative, even when input data was held constant between them. We obtained further improvements by incorporating several methods that are frequently overlooked in new model proposals, suggesting that it will be useful to explore these options further in the future.

Exploiting hierarchy in medical concept embedding.

OBJECTIVE: To construct and publicly release a set of medical concept embeddings for codes following the ICD-10 coding standard which explicitly incorporate hierarchical information from medical codes into the embedding formulation. MATERIALS AND METHODS: We trained concept embeddings using several new extensions to the Word2Vec algorithm using a dataset of approximately 600,000 patients from a major integrated healthcare organization in the Mid-Atlantic US. Our concept embeddings included additional entities to account for the medical categories assigned to codes by the Clinical Classification Software Revised (CCSR) dataset. We compare these results to sets of publicly released pretrained embeddings and alternative training methodologies. RESULTS: We found that Word2Vec models which included hierarchical data outperformed ordinary Word2Vec alternatives on tasks which compared naïve clusters to canonical ones provided by CCSR. Our Skip-Gram model with both codes and categories achieved 61.4% normalized mutual information with canonical labels in comparison to 57.5% with traditional Skip-Gram. In models operating on two different outcomes, we found that including hierarchical embedding data improved classification performance 96.2% of the time. When controlling for all other variables, we found that co-training embeddings improved classification performance 66.7% of the time. We found that all models outperformed our competitive benchmarks. DISCUSSION: We found significant evidence that our proposed algorithms can express the hierarchical structure of medical codes more fully than ordinary Word2Vec models, and that this improvement carries forward into classification tasks. As part of this publication, we have released several sets of pretrained medical concept embeddings using the ICD-10 standard which significantly outperform other well-known pretrained vectors on our tested outcomes.

Effect of delayed misoprostol dosing interval for induction of labor: a retrospective study.

BACKGROUND: Induction of labor occurs in greater than 22% of all pregnancies in the United States. Previous studies have shown that misoprostol is more effective for induction than oxytocin or dinoprostone alone. The World Health Organization recommends vaginal misoprostol 25mcg every 6 hours and the American Congress of Obstetricians and Gynecologists recommends 25mcg vaginal misoprostol every three to 6 hours. Although route of administration and dosage of misoprostol has been extensively studied, little is known about the optimal dosing interval of vaginal misoprostol. METHODS: The primary objective of this study is to determine the effect of delayed vaginal misoprostol dosing, defined as any interval longer than 4.5 h, on time to vaginal delivery. Our hypothesis is that the routine dosing interval of 4 hours shortens times to vaginal delivery compared to delayed dosing, even when adjusted for the time of delay. Secondary objectives include the effect of delayed vaginal misoprostol dosing on cesarean section rate, operative vaginal delivery rate, maternal outcomes, and neonatal outcomes. We conducted a retrospective chart review of 323 inductions of labor at one academic institution. The primary outcome was the proportion of patients who achieved a vaginal delivery within 24 h. The group who received all doses of misoprostol within a 4.5 h dosing window (Routine Dosing Interval Group) was compared with the group who had any dosing deviation (Delayed Dosing Interval Group). RESULTS: Of 133 included patients, 64 subjects received routine interval dosing and 69 subjects received delayed interval dosing. The vaginal delivery rates within 24 h were 56% (36/64) and 20% (14/69), respectively (P < 10- 4). Spontaneous vaginal delivery rates were 86% (55/64) vs. 75% (52/69), respectively (P = .13). Kaplan Meier curves demonstrated statistically significant difference in time to vaginal delivery between groups, with a Cox Proportional Hazard ratio for routine dosing interval of 1.73 (P < 10- 5) unadjusted and 1.34 (P = .01) when adjusted for dosing delay. CONCLUSIONS: This retrospective study demonstrates a significant increase in delay-adjusted time to vaginal delivery when doses of vaginal misoprostol are delayed past 4.5 h.

Learning and Imputation for Mass-spec Bias Reduction (LIMBR).

MOTIVATION: Decreasing costs are making it feasible to perform time series proteomics and genomics experiments with more replicates and higher resolution than ever before. With more replicates and time points, proteome and genome-wide patterns of expression are more readily discernible. These larger experiments require more batches exacerbating batch effects and increasing the number of bias trends. In the case of proteomics, where methods frequently result in missing data this increasing scale is also decreasing the number of peptides observed in all samples. The sources of batch effects and missing data are incompletely understood necessitating novel techniques. RESULTS: Here we show that by exploiting the structure of time series experiments, it is possible to accurately and reproducibly model and remove batch effects. We implement Learning and Imputation for Mass-spec Bias Reduction (LIMBR) software, which builds on previous block-based models of batch effects and includes features specific to time series and circadian studies. To aid in the analysis of time series proteomics experiments, which are often plagued with missing data points, we also integrate an imputation system. By building LIMBR for imputation and time series tailored bias modeling into one straightforward software package, we expect that the quality and ease of large-scale proteomics and genomics time series experiments will be significantly increased. AVAILABILITY AND IMPLEMENTATION: Python code and documentation is available for download at https://github.com/aleccrowell/LIMBR and LIMBR can be downloaded and installed with dependencies using 'pip install limbr'. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Circadian Proteomic Analysis Uncovers Mechanisms of Post-Transcriptional Regulation in Metabolic Pathways.

Transcriptional and translational feedback loops in fungi and animals drive circadian rhythms in transcript levels that provide output from the clock, but post-transcriptional mechanisms also contribute. To determine the extent and underlying source of this regulation, we applied newly developed analytical tools to a long-duration, deeply sampled, circadian proteomics time course comprising half of the proteome. We found a quarter of expressed proteins are clock regulated, but >40% of these do not arise from clock-regulated transcripts, and our analysis predicts that these protein rhythms arise from oscillations in translational rates. Our data highlighted the impact of the clock on metabolic regulation, with central carbon metabolism reflecting both transcriptional and post-transcriptional control and opposing metabolic pathways showing peak activities at different times of day. The transcription factor CSP-1 plays a role in this metabolic regulation, contributing to the rhythmicity and phase of clock-regulated proteins.

Guidelines for Genome-Scale Analysis of Biological Rhythms.

Genome biology approaches have made enormous contributions to our understanding of biological rhythms, particularly in identifying outputs of the clock, including RNAs, proteins, and metabolites, whose abundance oscillates throughout the day. These methods hold significant promise for future discovery, particularly when combined with computational modeling. However, genome-scale experiments are costly and laborious, yielding "big data" that are conceptually and statistically difficult to analyze. There is no obvious consensus regarding design or analysis. Here we discuss the relevant technical considerations to generate reproducible, statistically sound, and broadly useful genome-scale data. Rather than suggest a set of rigid rules, we aim to codify principles by which investigators, reviewers, and readers of the primary literature can evaluate the suitability of different experimental designs for measuring different aspects of biological rhythms. We introduce CircaInSilico, a web-based application for generating synthetic genome biology data to benchmark statistical methods for studying biological rhythms. Finally, we discuss several unmet analytical needs, including applications to clinical medicine, and suggest productive avenues to address them.

A Tool Set for the Genome-Wide Analysis of Neurospora crassa by RT-PCR.

Neurospora crassa is an important model organism for filamentous fungi as well as for circadian biology and photobiology. Although the community-accumulated tool set for the molecular analysis of Neurospora is extensive, two components are missing: (1) dependable reference genes whose level of expression are relatively constant across light/dark cycles and as a function of time of day and (2) a catalog of primers specifically designed for real-time PCR (RT-PCR). To address the first of these we have identified genes that are optimal for use as reference genes in RT-PCR across a wide range of expression levels; the mRNA/transcripts from these genes have potential for use as reference noncycling transcripts outside of Neurospora. In addition, we have generated a genome-wide set of RT-PCR primers, thereby streamlining the analysis of gene expression. In validation studies these primers successfully identified target mRNAs arising from 70% (34 of 49) of all tested genes and from all (28) of the moderately to highly expressed tested genes.

PhytoBeta imager: a positron imager for plant biology.

Several positron emitting radioisotopes such as (11)C and (13)N can be used in plant biology research. The (11)CO(2) tracer is used to facilitate plant biology research toward optimization of plant productivity, biofuel development and carbon sequestration in biomass. Positron emission tomography (PET) imaging has been used to study carbon transport in live plants using (11)CO(2). Because plants typically have very thin leaves, little medium is present for the emitted positrons to undergo an annihilation event. The emitted positrons from (11)C (maximum energy 960 keV) could require up to approximately 4 mm of water equivalent material for positron annihilation. Thus many of the positrons do not annihilate inside the leaf, resulting in limited sensitivity for PET imaging. To address this problem we have developed a compact beta-positive, beta-minus particle imager (PhytoBeta imager) for (11)CO(2) leaf imaging. The detector is based on a Hamamatsu H8500 position sensitive photomultiplier tube optically coupled via optical grease to a 0.5 mm thick Eljen EJ-212 plastic scintillator. The detector is equipped with a flexible arm to allow its placement and orientation over or under the leaf to be studied while maintaining the leaf's original orientation. To test the utility of the system the detector was used to measure carbon translocation in a leaf of the spicebush (Lindera benzoin) under two transient light conditions.

A feasibility study using radiochromic films for fast neutron 2D passive dosimetry.

The objective of this paper is threefold: (1) to establish sensitivity of XRQA and EBT radiochromic films to fast neutron exposure; (2) to develop a film response to radiation dose calibration curve and (3) to investigate a two-dimensional (2D) film dosimetry technique for use in establishing an experimental setup for a radiobiological irradiation of mice and to assess the dose to the mice in this setup. The films were exposed to a 10 MeV neutron beam via the (2)H(d,n)(3)He reaction. The XRQA film response was a factor of 1.39 greater than EBT film response to the 10 MeV neutron beam when exposed to a neutron dose of 165 cGy. A film response-to-soft tissue dose calibration function was established over a range of 0-10 Gy and had a goodness of fit of 0.9926 with the calibration data. The 2D film dosimetry technique estimated the neutron dose to the mice by measuring the dose using a mouse phantom and by placing a piece of film on the exterior of the experimental mouse setup. The film results were benchmarked using Monte Carlo and aluminum (Al) foil activation measurements. The radiochromic film, Monte Carlo and Al foil dose measurements were strongly correlated, and the film within the mouse phantom agreed to better than 7% of the externally mounted films. These results demonstrated the potential application of radiochromic films for passive 2D neutron dosimetry.

Neutron-stimulated emission computed tomography of a multi-element phantom.

This paper describes the implementation of neutron-stimulated emission computed tomography (NSECT) for non-invasive imaging and reconstruction of a multi-element phantom. The experimental apparatus and process for acquisition of multi-spectral projection data are described along with the reconstruction algorithm and images of the two elements in the phantom. Independent tomographic reconstruction of each element of the multi-element phantom was performed successfully. This reconstruction result is the first of its kind and provides encouraging proof of concept for proposed subsequent spectroscopic tomography of biological samples using NSECT.

Exploring the transport of plant metabolites using positron emitting radiotracers.

Short-lived positron-emitting radiotracer techniques provide time-dependent data that are critical for developing models of metabolite transport and resource distribution in plants and their microenvironments. Until recently these techniques were applied to measure radiotracer accumulation in coarse regions along transport pathways. The recent application of positron emission tomography (PET) techniques to plant research allows for detailed quantification of real-time metabolite dynamics on previously unexplored spatial scales. PET provides dynamic information with millimeter-scale resolution on labeled carbon, nitrogen, and water transport over a small plant-size field of view. Because details at the millimeter scale may not be required for all regions of interest, hybrid detection systems that combine high-resolution imaging with other radiotracer counting technologies offer the versatility needed to pursue wide-ranging plant physiological and ecological research. In this perspective we describe a recently developed hybrid detection system at Duke University that provides researchers with the flexibility required to carry out measurements of the dynamic responses of whole plants to environmental change using short-lived radiotracers. Following a brief historical development of radiotracer applications to plant research, the role of radiotracers is presented in the context of various applications at the leaf to the whole-plant level that integrates cellular and subcellular signals andor controls.

Introduction to neutron stimulated emission computed tomography.

Neutron stimulated emission computed tomography (NSECT) is presented as a new technique for in vivo tomographic spectroscopic imaging. A full implementation of NSECT is intended to provide an elemental spectrum of the body or part of the body being interrogated at each voxel of a three-dimensional computed tomographic image. An external neutron beam illuminates the sample and some of these neutrons scatter inelastically, producing characteristic gamma emission from the scattering nuclei. These characteristic gamma rays are acquired by a gamma spectrometer and the emitting nucleus is identified by the emitted gamma energy. The neutron beam is scanned over the body in a geometry that allows for tomographic reconstruction. Tomographic images of each element in the spectrum can be reconstructed to represent the spatial distribution of elements within the sample. Here we offer proof of concept for the NSECT method, present the first single projection spectra acquired from multi-element phantoms, and discuss potential biomedical applications.