SEARCH · Search NASA

Results for “Generalized Linear Model”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

SourceResearch connectionFrom yearThrough year

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 37 records · Page 2

All sources NASA NTRS DOE OSTI

DOE OSTI2015

Posterior propriety for hierarchical models with log-likelihoods that have norm bounds

Statisticians often use improper priors to express ignorance or to provide good frequency properties, requiring that posterior propriety be verified. Our paper addresses generalized linear mixed models, GLMMs, when Level I parameters have Normal distributions, with many commonly-used hyperpriors. It provides easy-to-verify sufficient posterior propriety conditions based on dimensions, matrix ranks, and exponentiated norm bounds, ENBs, for the Level I likelihood. Since many familiar likelihoods have ENBs, which is often verifiable via log-concavity and MLE finiteness, our novel use of ENBs permits unification of posterior propriety results and posterior MGF/moment results for many useful Level I distributions, including those commonly used with multilevel generalized linear models, e.g., GLMMs and hierarchical generalized linear models, HGLMs. Furthermore, those who need to verify existence of posterior distributions or of posterior MGFs/moments for a multilevel generalized linear model given a proper or improper multivariate F prior as in Section 1 should find the required results in Sections 1 and 2 and Theorem 3 (GLMMs), Theorem 4 (HGLMs), or Theorem 5 (posterior MGFs/moments).

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2018

Projection of temperature-related mortality due to cardiovascular disease in beijing under different climate change, population, and adaptation scenarios

Highlights: • The population change can bring 5 times growth to temperature-related CVD deaths. • Increase of heat-related CVD deaths is higher than decrease of cold-related ones. • The adaptation to climate change may cause more temperature-related CVD deaths. Human health faces unprecedented challenges caused by climate change. Thus, studies of the effect of temperature change on total mortality have been conducted in numerous countries. However, few of those studies focused on temperature-related mortality due to cardiovascular disease (CVD) or considered future population changes and adaptation to climate change. We present herein a projection of temperature-related mortality due to CVD under different climate change, population, and adaptation scenarios in Beijing, a megacity in China. To this end, 19 global circulation models (GCMs), 3 representative concentration pathways (RCPs), 3 socioeconomic pathways, together with generalized linear models and distributed lag non-linear models, were used to project future temperature-related CVD mortality during periods centered around the years 2050 and 2070. The number of temperature-related CVD deaths in Beijing is projected to increase by 3.5–10.2% under different RCP scenarios compared with that during the baseline period. Using the same GCM, the future daily maximum temperatures projected using the RCP2.6, RCP4.5, and RCP8.5 scenarios showed a gradually increasing trend. When population change is considered, the annual rate of increase in temperature-related CVD deaths was up to fivefold greater than that under no-population-change scenarios. The decrease in the number of cold-related deaths did not compensate for the increase in that of heat-related deaths, leading to a general increase in the number of temperature-related deaths due to CVD in Beijing. In addition, adaptation to climate change may enhance rather than ameliorate the effect of climate change, as the increase in cold-related CVD mortality greater than the decrease in heat-related CVD mortality in the adaptation scenarios will result in an increase in the total number of temperature-related CVD mortalities.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2019

Effects of Omitting Non-confounding Predictors From General Relative-Risk Models for Binary Outcomes

The effects, in terms of bias and precision, of omitting non-confounding predictive covariates from generalized linear models have been well studied, and it is known that such omission results in attenuation bias but increased precision with logistic regression. However, many epidemiologic risk analyses utilize alternative models that are not based on a linear predictor, and the effect of omitting non-confounding predictive covariates from such models has not been characterized. We employed simulation to study the effects on risk estimation of omitting non-confounding predictive covariates from an excess relative risk (ERR) model and a general additive-multiplicative relative-risk mixture model for binary outcome data in a case-control setting. We also compared the results to the effects with ordinary logistic regression. For these commonly employed alternative relative-risk models, the bias was similar to that with logistic regression when the risk was small. More generally, the bias and standard error of the risk-parameter estimates demonstrated patterns that are similar to those with logistic regression, but with greater magnitude depending on the true value of the risk. The magnitude of bias and standard error had little relation to study size or underlying disease prevalence. Prior conclusions regarding omitted covariates in logistic regression models can be qualitatively applied to the ERR and the general additive-multiplicative relative-risk mixture model without substantial change. Quantitatively, however, these alternative models may have slightly greater omitted-covariate bias, depending on the magnitude of the true risk being estimated.

60 APPLIED LIFE SCIENCES↗

DOE OSTI2013

Semisupervised Classification of Texts Using Particle Learning for Probabilistic Automata

This chapter presents a novel online learning system for classification of email texts based on particle learning for composite mixture models involving probabilistic automata. The composite mixture structure allows specification of a joint probability model for heterogeneous collections of independent variables without requiring complex embeddings via generalized linear models or copula techniques. The chapter is organized as follows. Section 12.2 presents a hierarchical model representation of a class of probabilistic automata and derives the particle learning algorithm for obtaining parameter estimates. Section 12.3 describes a framework for semisupervised text classification problems based on a composite mixture model formulation. Section 12.4 discusses an application of the classifier to a spam detection dataset. Section 12.5 concludes with a discussion of computational considerations and future work.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2020

Differential credibility assessment for statistical downscaling

Climate science is increasingly using (i) ensembles of climate projections from multiple models derived using different assumptions and/or scenarios and (ii) process-oriented diagnostics of model fidelity. Efforts to assign differential credibility to projections and/or models are also rapidly advancing. A framework to quantify and depict the credibility of statistically downscaled model output is presented and demonstrated. Here, the approach employs transfer functions in the form of robust and resilient generalized linear models applied to downscale daily minimum and maximum temperature anomalies at 10 locations using predictors drawn from ERA-Interim reanalysis and two global climate models (GCM; GFDL-ESM2M and MPI-ESM-LR). The downscaled time series are used to derive several impact relevant CLIMDEX temperature indices that are assigned credibility based on (1) the reproduction of relevant large-scale predictors by the GCMs (i.e. fraction of regression beta-weights derived from predictors that are well-reproduced) and (2) the degree of variance in the observations reproduced in the downscaled series following application of a new variance inflation technique. Credibility of the downscaled predictands varies across locations, between the two GCM and is generally higher for minimum temperature than maximum temperature. The differential credibility assessment framework demonstrated here is easy to use and flexible. It can be applied as is to inform decision makers regarding projection confidence, and/or extended to include other components of the transfer functions, and/or used to weight members of a statistically downscaled ensemble.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2011

Normal Tissue Complication Probability Modeling of Acute Hematologic Toxicity in Cervical Cancer Patients Treated With Chemoradiotherapy

Purpose: To test the hypothesis that increased pelvic bone marrow (BM) irradiation is associated with increased hematologic toxicity (HT) in cervical cancer patients undergoing chemoradiotherapy and to develop a normal tissue complication probability (NTCP) model for HT. Methods and Materials: We tested associations between hematologic nadirs during chemoradiotherapy and the volume of BM receiving {>=}10 and 20 Gy (V{sub 10} and V{sub 20}) using a previously developed linear regression model. The validation cohort consisted of 44 cervical cancer patients treated with concurrent cisplatin and pelvic radiotherapy. Subsequently, these data were pooled with data from 37 identically treated patients from a previous study, forming a cohort of 81 patients for normal tissue complication probability analysis. Generalized linear modeling was used to test associations between hematologic nadirs and dosimetric parameters, adjusting for body mass index. Receiver operating characteristic curves were used to derive optimal dosimetric planning constraints. Results: In the validation cohort, significant negative correlations were observed between white blood cell count nadir and V{sub 10} (regression coefficient ({beta}) = -0.060, p = 0.009) and V{sub 20} ({beta} = -0.044, p = 0.010). In the combined cohort, the (adjusted) {beta} estimates for log (white blood cell) vs. V{sub 10} and V{sub 20} were as follows: -0.022 (p = 0.025) and -0.021 (p = 0.002), respectively. Patients with V{sub 10} {>=} 95% were more likely to experience Grade {>=}3 leukopenia (68.8% vs. 24.6%, p < 0.001) than were patients with V{sub 20} > 76% (57.7% vs. 21.8%, p = 0.001). Conclusions: These findings support the hypothesis that HT increases with increasing pelvic BM volume irradiated. Efforts to maintain V{sub 10} < 95% and V{sub 20} < 76% may reduce HT.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

DOE OSTI2015

Parametric behaviors of CLUBB in simulations of low clouds in the C ommunity A tmosphere M odel ( CAM )

Abstract In this study, we investigate the sensitivity of simulated low clouds to 14 selected tunable parameters of Cloud Layers Unified By Binormals (CLUBB), a higher‐order closure (HOC) scheme, and four parameters of the Zhang‐McFarlane (ZM) deep convection scheme in the Community Atmosphere Model version 5 (CAM5). A Quasi‐Monte Carlo (QMC) sampling approach is adopted to effectively explore the high‐dimensional parameter space and a generalized linear model is applied to study the responses of simulated cloud fields to tunable parameters. Our results show that the variance in simulated low‐cloud properties (cloud fraction and liquid water path) can be explained by the selected tunable parameters in two different ways: macrophysics itself and its interaction with microphysics. First, the parameters related to dynamic and thermodynamic turbulent structure and double Gaussian closure are found to be the most influential parameters for simulating low clouds. The spatial distributions of the parameter contributions show clear cloud‐regime dependence. Second, because of the coupling between cloud macrophysics and cloud microphysics, the coefficient of the dissipation term in the total water variance equation is influential. This parameter affects the variance of in‐cloud cloud water, which further influences microphysical process rates, such as autoconversion, and eventually low‐cloud fraction. This study improves understanding of HOC behavior associated with parameter uncertainties and provides valuable insights for the interaction of macrophysics and microphysics.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2014

A sensitivity analysis of cloud properties to CLUBB parameters in the single-column Community Atmosphere Model (SCAM5)

In this study, we investigate the sensitivity of simulated shallow cumulus and stratocumulus clouds to selected tunable parameters of Cloud Layers Unified by Binormals (CLUBB) in the single column version of Community Atmosphere Model version 5 (SCAM5). A quasi-Monte Carlo (QMC) sampling approach is adopted to effectively explore the high-dimensional parameter space and a generalized linear model is adopted to study the responses of simulated cloud fields to tunable parameters. One stratocumulus and two shallow convection cases are configured at both coarse and fine vertical resolutions in this study.. Our results show that most of the variance in simulated cloud fields can be explained by a small number of tunable parameters. The parameters related to Newtonian and buoyancy-damping terms of total water flux are found to be the most influential parameters for stratocumulus. For shallow cumulus, the most influential parameters are those related to skewness of vertical velocity, reflecting the strong coupling between cloud properties and dynamics in this regime. The influential parameters in the stratocumulus case are sensitive to the choice of the vertical resolution while little sensitivity is found for the shallow convection cases, as eddy mixing length (or dissipation time scale) plays a more important role and depends more strongly on the vertical resolution in stratocumulus than in shallow convections. The influential parameters remain almost unchanged when the number of tunable parameters increases from 16 to 35. This study improves understanding of the CLUBB behavior associated with parameter uncertainties.

58 GEOSCIENCES↗

DOE OSTI2016

Sensitivity analysis of simulated SOA loadings using a variance-based statistical approach: SENSITIVITY ANALYSIS OF SOA

We investigate the sensitivity of secondary organic aerosol (SOA) loadings simulated by a regional chemical transport model to 7 selected tunable model parameters: 4 involving emissions of anthropogenic and biogenic volatile organic compounds, anthropogenic semi-volatile and intermediate volatility organics (SIVOCs), and NOx, 2 involving dry deposition of SOA precursor gases, and one involving particle-phase transformation of SOA to low volatility. We adopt a quasi-Monte Carlo sampling approach to effectively sample the high-dimensional parameter space, and perform a 250 member ensemble of simulations using a regional model, accounting for some of the latest advances in SOA treatments based on our recent work. We then conduct a variance-based sensitivity analysis using the generalized linear model method to study the responses of simulated SOA loadings to the tunable parameters. Analysis of SOA variance from all 250 simulations shows that the volatility transformation parameter, which controls whether particle-phase transformation of SOA from semi-volatile SOA to non-volatile is on or off, is the dominant contributor to variance of simulated surface-level daytime SOA (65% domain average contribution). We also split the simulations into 2 subsets of 125 each, depending on whether the volatility transformation is turned on/off. For each subset, the SOA variances are dominated by the parameters involving biogenic VOC and anthropogenic SIVOC emissions. Furthermore, biogenic VOC emissions have a larger contribution to SOA variance when the SOA transformation to non-volatile is on, while anthropogenic SIVOC emissions have a larger contribution when the transformation is off. NOx contributes less than 4.3% to SOA variance, and this low contribution is mainly attributed to dominance of intermediate to high NOx conditions throughout the simulated domain. The two parameters related to dry deposition of SOA precursor gases also have very low contributions to SOA variance. This study highlights the large sensitivity of SOA loadings to the particle-phase transformation of SOA volatility, which is neglected in most previous models.

Shrivastava, Manish↗

DOE OSTI2018

Comparing theory based and higher-order reduced models for fusion simulation data

We consider using regression to fit a theory-based log-linear ansatz, as well as higher order approximations, for the thermal energy confinement of a Tokamak as a function of device features. We use general linear models based on total order polynomials, as well as deep neural networks. The results indicate that the theory-based model fits the data almost as well as the more sophisticated machines, within the support of the data set. The conclusion we arrive at is that only negligible improvements can be made to the theoretical model, for input data of this type.

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

DOE OSTI2021

Machine Learning-Based Classification of Lignocellulosic Biomass from Pyrolysis-Molecular Beam Mass Spectrometry Data

High-throughput analysis of biomass is necessary to ensure consistent and uniform feedstocks for agricultural and bioenergy applications and is needed to inform genomics and systems biology models. Pyrolysis followed by mass spectrometry such as molecular beam mass spectrometry (py-MBMS) analyses are becoming increasingly popular for the rapid analysis of biomass cell wall composition and typically require the use of different data analysis tools depending on the need and application. Here, the authors report the py-MBMS analysis of several types of lignocellulosic biomass to gain an understanding of spectral patterns and variation with associated biomass composition and use machine learning approaches to classify, differentiate, and predict biomass types on the basis of py-MBMS spectra. Py-MBMS spectra were also corrected for instrumental variance using generalized linear modeling (GLM) based on the use of select ions relative abundances as spike-in controls. Machine learning classification algorithms e.g., random forest, k-nearest neighbor, decision tree, Gaussian Naïve Bayes, gradient boosting, and multilayer perceptron classifiers were used. The k-nearest neighbors (k-NN) classifier generally performed the best for classifications using raw spectral data, and the decision tree classifier performed the worst. After normalization of spectra to account for instrumental variance, all the classifiers had comparable and generally acceptable performance for predicting the biomass types, although the k-NN and decision tree classifiers were not as accurate for prediction of specific sample types. Gaussian Naïve Bayes (GNB) and extreme gradient boosting (XGB) classifiers performed better than the k-NN and the decision tree classifiers for the prediction of biomass mixtures. The data analysis workflow reported here could be applied and extended for comparison of biomass samples of varying types, species, phenotypes, and/or genotypes or subjected to different treatments, environments, etc. to further elucidate the sources of spectral variance, patterns, and to infer compositional information based on spectral analysis, particularly for analysis of data without a priori knowledge of the feedstock composition or identity.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2014

Summer outdoor temperature and occupational heat-related illnesses in Quebec (Canada)

Background: Predicted rise in global mean temperature and intensification of heat waves associated with climate change present an increasing challenge for occupational health and safety. Although important scientific knowledge has been gathered on the health effects of heat, very few studies have focused on quantifying the association between outdoor heat and mortality or morbidity among workers. Objective: To quantify the association between occupational heat-related illnesses and exposure to summer outdoor temperatures. Methods: We modeled 259 heat-related illnesses compensated by the Workers' Compensation Board of Quebec between May and September, from 1998 to 2010, with maximum daily summer outdoor temperatures in 16 health regions of Quebec (Canada) using generalized linear models with negative binomial distributions, and estimated the pooled effect sizes for all regions combined, by sex and age groups, and for different time lags with random-effect models for meta-analyses. Results: The mean daily compensation count was 0.13 for all regions of Quebec combined. The relationship between daily counts of compensations and maximum daily temperatures was log-linear; the pooled incidence rate ratio (IRR) of daily heat-related compensations per 1 °C increase in daily maximum temperatures was 1.419 (95% CI 1.326 to 1.520). Associations were similar for men and women and by age groups. Increases in daily maximum temperatures at lags 1 and 2 and for two and three-day lag averages were also associated with increases in daily counts of compensations (IRRs of 1.206 to 1.471 for every 1 °C increase in temperature). Conclusion: This study is the first to quantify the association between occupational heat-related illnesses and exposure to summer temperatures in Canada. The model (risk function) developed in this study could be useful to improve the assessment of future impacts of predicted summer outdoor temperatures on workers and vulnerable groups, particularly in colder temperate zones. - Highlights: • 259 heat-related compensated illnesses were modeled with ambient temperature • An overall risk ratio of 1.419 (95% CI 1.326–1.520) for every 1 °C increase was found • Risk estimates were similar for men and women and by large age groups. • There were little lag effects (IRRs of 1.206 to 1.471 for every 1 °C increase)

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2019

Bayesian probability of agreement for comparing the similarity of response surfaces

For a variety of linear and generalized linear models, there may be interest in evaluating whether the expected response values in two different populations are sufficiently similar to be considered practically equivalent. The probability of agreement has been introduced as a strategy for quantifying the similarity between two groups. We propose a Bayesian version of the probability of agreement to quantify the similarity between linear and generalized linear response surfaces. The proposed methodology is based on Markov chain Monte Carlo estimation and it allows for an intuitive interpretation based directly on population parameters. As demonstrated, the methodology can be flexibly applied to a variety of different models. We illustrate its use with three examples for which the response depends on predictor variables through a linear, logistic, or Poisson regression model. The computation associated with this approach has been automated with a freely available R Shiny app.

97 MATHEMATICS AND COMPUTING↗

DOE OSTI2017

Effects of hydrokinetic turbine sound on the behavior of four species of fish within an experimental mesocosm

The development of hydrokinetic energy technologies (e.g., tidal turbines) has raised concern over the potential impacts of underwater sound produced by hydrokinetic turbines on fish species likely to encounter these turbines. To assess the potential for behavioral impacts, we exposed four species of fish to varying intensities of recorded hydrokinetic turbine sound in a semi-natural environment. Although we tested freshwater species (redhorse suckers [Moxostoma spp], freshwater drum [Aplondinotus grunniens], largemouth bass [Micropterus salmoides], and rainbow trout [Oncorhynchus mykiss]), these species are also representative of the hearing physiology and sensitivity of estuarine species that would be affected at tidal energy sites. Here, we evaluated changes in fish position relative to different intensities of turbine sound as well as trends in location over time with linear mixed-effects and generalized additive mixed models. We also evaluated changes in the proportion of near-source detections relative to sound intensity and exposure time with generalized linear mixed models and generalized additive models. Models indicated that redhorse suckers may respond to sustained turbine sound by increasing distance from the sound source. Freshwater drum models suggested a mixed response to turbine sound, and largemouth bass and rainbow trout models did not indicate any likely responses to turbine sound. Lastly, findings highlight the importance for future research to utilize accurate localization systems, different species, validated sound transmission distances, and to consider different types of behavioral responses to different turbine designs and to the cumulative sound of arrays of multiple turbines.

16 TIDAL AND WAVE POWER↗

DOE OSTI2014

Soil Dust Aerosols and Wind as Predictors of Seasonal Meningitis Incidence in Niger

Epidemics of meningococcal meningitis are concentrated in sub-Saharan Africa during the dry season, a period when the region is affected by the Harmattan, a dry and dusty northeasterly trade wind blowing from the Sahara into the Gulf of Guinea. We examined the potential of climate-based statistical forecasting models to predict seasonal incidence of meningitis in Niger at both the national and district levels. We used time series of meningitis incidence from 1986 through 2006 for 38 districts in Niger. We tested models based on data that would be readily available in an operational framework, such as climate and dust, population, and the incidence of early cases before the onset of the meningitis season in January–May. Incidence was used as a proxy for immunological state, susceptibility, and carriage in the population. We compared a range of negative binomial generalized linear models fitted to the meningitis data. At the national level, a model using early incidence in December and averaged November–December zonal wind provided the best fit (pseudo-R2 = 0.57), with zonal wind having the greatest impact. A model with surface dust concentration as a predictive variable performed indistinguishably well. At the district level, the best spatiotemporal model included zonal wind, dust concentration, early incidence in December, and population density (pseudo-R2 = 0.41). We showed that wind and dust information and incidence in the early dry season predict part of the year-to-year variability of the seasonal incidence of meningitis at both national and district levels in Niger. Models of this form could provide an early-season alert that wind, dust, and other conditions are potentially conducive to an epidemic.

59 BASIC BIOLOGICAL SCIENCES↗

DOE OSTI2020

Modeling regional precipitation over the Indus River basin of Pakistan using statistical downscaling

Complex processes govern spatiotemporal distribution of precipitation within the high-mountainous headwater regions (commonly known as the upper Indus basin (UIB)), of the Indus River basin of Pakistan. Reliable precipitation simulations particularly over the UIB present a major scientific challenge due to regional complexity and inadequate observational coverage. Here, we present a statistical downscaling approach to model observed precipitation of the entire Indus basin, with a focus on UIB within available data constraints. Taking advantage of recent high altitude (HA) observatories, we perform precipitation regionalization using K-means cluster analysis to demonstrate effectiveness of low-altitude stations to provide useful precipitation inferences over more uncertain and hydrologically important HA of the UIB. We further employ generalized linear models (GLM) with gamma and Tweedie distributions to identify major dynamic and thermodynamic drivers from a reanalysis dataset within a robust cross-validation framework that explain observed spatiotemporal precipitation patterns across the Indus basin. Final statistical models demonstrate higher predictability to resolve precipitation variability over wetter southern Himalayans and different lower Indus regions, by mainly using different dynamic predictors. The modeling framework also shows an adequate performance over more complex and uncertain trans-Himalayans and the northwestern regions of the UIB, particularly during the seasons dominated by the westerly circulations. However, the cryosphere-dominated trans-Himalayan regions, which largely govern the basin hydrology, require relatively complex models that contain dynamic and thermodynamic circulations. Furthermore, we also analyzed relevant atmospheric circulations during precipitation anomalies over the UIB, to evaluate physical consistency of the statistical models, as an additional measure of reliability. Overall, our results suggest that such circulation-based statistical downscaling has the potential to improve our understanding towards distinct features of the regional-scale precipitation across the upper and lower Indus basin. Additionally, such understanding should help to assess the response of this complex, data-scarce, and climate-sensitive river basin amid future climatic changes, to serve communal and scientific interests.

54 ENVIRONMENTAL SCIENCES↗

DOE OSTI2015

Dose Response for Radiation Cataractogenesis: A Meta-Regression of Hematopoietic Stem Cell Transplantation Regimens

Purpose/Objective(s): To perform a meta-regression on published data and to model the 5-year probability of cataract development after hematopoietic stem cell transplantation (HSCT) with and without total body irradiation (TBI). Methods and Materials: Eligible studies reporting cataract incidence after HSCT with TBI were identified by a PubMed search. Seventeen publications provided complete information on radiation dose schedule, fractionation, dose rate, and actuarial cataract incidence. Chemotherapy-only regimens were included as zero radiation dose regimens. Multivariate meta-regression with a weighted generalized linear model was used to model the 5-year cataract incidence and contributory factors. Results: Data from 1386 patients in 21 series were included for analysis. TBI was administered to a total dose of 0 to 15.75 Gy with single or fractionated schedules with a dose rate of 0.04 to 0.16 Gy/min. Factors significantly associated with 5-year cataract incidence were dose, dose times dose per fraction (D•dpf), pediatric versus adult status, and the absence of an ophthalmologist as an author. Dose rate, graft versus host disease, steroid use, hyperfractionation, and number of fractions were not significant. Five-fold internal cross-validation showed a model validity of 83% ± 8%. Regression diagnostics showed no evidence of lack-of-fit and no patterns in the studentized residuals. The α/β ratio from the linear quadratic model, estimated as the ratio of the coefficients for dose and D•dpf, was 0.76 Gy (95% confidence interval [CI], 0.05-1.55). The odds ratio for pediatric patients was 2.8 (95% CI, 1.7-4.6) relative to adults. Conclusions: Dose, D•dpf, pediatric status, and regimented follow-up care by an ophthalmologist were predictive of 5-year cataract incidence after HSCT. The low α/β ratio indicates the importance of fractionation in reducing cataracts. Dose rate effects have been observed in single institution studies but not in the combined data analyzed here. Although data were limited to articles with 5-year actuarial estimates, the development of radiation-induced cataracts extends beyond this time.

62 RADIOLOGY AND NUCLEAR MEDICINE↗

DOE OSTI2017

The weighted priors approach for combining expert opinions in logistic regression experiments

When modeling the reliability of a system or component, it is not uncommon for more than one expert to provide very different prior estimates of the expected reliability as a function of an explanatory variable such as age or temperature. Our goal in this paper is to incorporate all information from the experts when choosing a design about which units to test. Bayesian design of experiments has been shown to be very successful for generalized linear models, including logistic regression models. We use this approach to develop methodology for the case where there are several potentially non-overlapping priors under consideration. While multiple priors have been used for analysis in the past, they have never been used in a design context. The Weighted Priors method performs well for a broad range of true underlying model parameter choices and is more robust when compared to other reasonable design choices. Finally, we illustrate the method through multiple scenarios and a motivating example. Additional figures for this article are available in the online supplementary information.

97 MATHEMATICS AND COMPUTING↗