Skip to main content
eScholarship
Open Access Publications from the University of California

Scholarly Works (55 results)

  • Article
  • Peer Reviewed
UC Berkeley Previously Published Works (2022)
To analyze species count data when detection is imperfect, ecologists need models to estimate relative abundance in the presence of unknown sources of heterogeneity. Two candidate models are generalized linear mixed models (GLMMs) and hierarchical N-mixture models. GLMMs are computationally robust but do not explicitly separate detection from abundance patterns. N-mixture models separately estimate detection and abundance via a latent state but are sensitive to violations in assumptions and subject to practical estimation issues. When one can assume that detection is not systematically confounded with ecological patterns of interest, these two models can be viewed as sharing a heuristic framework for relative abundance estimation. Model selection can then determine which predicts observed counts best, for example by AIC. We compared four N-mixture model variants and two GLMM variants for predicting bird counts in local subsets of a citizen science dataset, eBird, based on model selection and goodness-of-fit measures. We found that both GLMMs and N-mixture models-especially N-mixtures with beta-binomial detection submodels-were supported in a moderate number of datasets, suggesting that both tools are useful and that relative fit is context-dependent. We provide faster software implementations of N-mixture likelihood calculations and a reparameterization to interpret unstable estimates for N-mixture models.
    Cover page: Comparing N-mixture models and GLMMs for relative abundance estimation in a citizen science dataset.
    • Thesis
    • Peer Reviewed
    UC Berkeley Electronic Theses and Dissertations (2011)

    This dissertation consists of the analyses of three separate genetic association datasets. Each represents a unique data structure with a different question of interest that therefore require distinct approaches and methodologies. As such, the three substantive chapters (2-4) can each stand on their own. However, the over-arching question in each of these studies is the same: which genes (or genetic material) are related to the disease or outcome being studied. Moreover, while the methodologies are each distinct, they all incorporate statistical learning methodologies to obtain some modicum of inference.

    Study 1 - As computational power has improved the application of statistical learning algorithms to finding SNPs related to disease has become more ubiquitous. The hope is that these algorithms will be more capable than typical marginal testing in detecting SNPs with higher order effects. The Random Forests (RF) algorithm is one such algorithm that has seen increased use with genetic data. As part of its output, RF ranks the predictor variables (SNPs) on their relative importance.

    The present study represents the first application of the RF algorithm to Genome Wide Association (GWA) data and investigates how best to use the algorithm for this unique data structure. A multiple sclerosis (MS) GWA data set is used for the analysis. Results indicate the typical tuning parameter settings need to be adjusted for the high degree of sparsity in the data. Furthermore, most meaningful results were obtained when both unimportant and overly important SNPs were removed. RF was able to replicate some previous findings using the same data. Moreover, four genes not previously associated with MS were identified.

    Study 2 - In many analyses, one has data on one level but desires to draw inference on another level. For example, in genetic association studies, one observes units of DNA referred to as SNPs, but wants to determine whether genes that are comprised of SNPs are associated with disease. While there are some available approaches for addressing this issue, they usually involve making parametric assumptions and are not easily generalizable. A statistical test is proposed for testing the association of a set of variables with an outcome of interest. No assumptions are made about the functional form relating the variables to the outcome. A general function is fit using any statistical learning algorithm, with the SuperLearner algorithm suggested. The parameter of interest is the cross-validated risk and this is compared to an expected risk. A Wald test is proposed using the influence curve of the cross-validated risk to obtain the variance. It is shown both theoretically and via simulation that the test maintains appropriate type I error control and is more powerful than parametric tests under more general alternatives. The test is applied to an MS candidate gene study. Three separate analyses are performed highlighting the flexibility of the approach.

    Study 3 - Secondary analyses, such as Gene Ontology and Motif analysis, have become central components of gene expression experiments, allowing researchers to derive biological understanding from the set of genes that are differentially expressed. An important statistical task is determining which genes should be passed on to such programs and how the genes should be grouped for analysis. The typical approach is to cluster the set of differentially expressed genes, and pass these clusters on to the secondary analyses. However, many expression experiments have specific hypotheses which allow one to analyze the genes and group them in a more targeted approach. To illustrate the utility of being more specific, a gene expression study of C. elegans is used where a particular outcome was observed and hoped to be explained. A general

    model is fit and analyzed to estimate the parameters corresponding to the specific hypothesis, leading to four natural groupings of the differentially expressed genes. These groupings lead to meaningful results in the secondary analyses that allow for the biologist to make robust hypotheses that are experimentally confirmed. It is shown that a traditional approach would not have yielded such robust findings.

      Cover page: Finding Genes Related to Disease Using Statistical Learning
      • Article
      • Peer Reviewed
      UC Berkeley Previously Published Works (2022)
      Identifying rates at which birders engage with different species can inform the impact and efficacy of conservation outreach and the scientific use of community-collected biodiversity data. Species that are thought to be “charismatic” are often prioritized in conservation, and previous researchers have used sociological experiments and digital records to estimate charisma indirectly. In this study, we take advantage of community science efforts as another record of human engagement with animals that can reveal observer biases directly, which are in part driven by observer preference. We apply a multistage analysis to ask whether opportunistic birders contributing to iNaturalist engage more with larger, more colorful, and rarer birds relative to a baseline approximated from eBird contributors. We find that body mass, color contrast, and range size all predict overrepresentation in the opportunistic dataset. We also find evidence that, across 472 modeled species, 52 species are significantly overreported and 158 are significantly underreported, indicating a wide variety of species-specific effects. Understanding which birds are highly engaging can aid conservationists in creating impactful outreach materials and engaging new naturalists. The quantified differences between two prominent community science efforts may also be of use for researchers leveraging the data from one or both of them to answer scientific questions of interest.
        Cover page: Identifying engaging bird species and traits with community science observations.
        • Thesis
        • Peer Reviewed
        UC Berkeley Electronic Theses and Dissertations (2023)

        Estimating the spatiotemporal distributions of species and understanding how variation in those distributions is explained by the environment are central goals in ecology. Observations of animals generated by participatory science (or "citizen science") are an increasingly important resource for ecologists interested in estimating species distributions because they are high-volume and high-resolution. However, statistical inference with these data is more challenging than inference with data collected under standardized sampling, because participatory science observations contain substantial unmeasured variation in sampling effort and observer behavior. Ecologists need tools and methodological guidance that support the estimation of computationally efficient, flexible statistical models useful for robust inference with participatory science data. In this dissertation, I advance the field of species distribution modeling with participatory science data via contributions across three chapters. First, I present a new software tool, nimbleEcology, that supports the efficient and flexible estimation of hierarchical ecological models, alongside a brief review of the use of such models in ecology and three worked examples of model estimation. Second, I undertake a comparison of two modeling approaches useful for estimating relative abundance from participatory science data, making practical recommendations for model selection. Finally, I apply these methodological developments to data obtained from an important participatory science dataset, eBird, to investigate how common birds respond to drought in California's Central Valley ecoregion. This project demonstrates the application of modeling principles to an important ecological case study and produces new evidence to characterize critical dimensions of birds' drought responses.

          Cover page: Inferring species distributions from semi-structured biodiversity observations
          • Article
          • Peer Reviewed
          UC San Francisco Previously Published Works (2012)

          Background

          Current regulations require that all cardiac allograft offers for transplantation must include an interpreted 12-lead electrocardiogram (ECG). However, little is known about the expected ECG findings in potential organ donors or the clinical significance of any identified abnormalities in terms of cardiac allograft function and suitability for transplantation.

          Methods and results

          A single experienced reviewer interpreted the first ECG obtained after brain stem herniation in 980 potential organ donors managed by the California Transplant Donor Network from 2002 to 2007. ECG abnormalities were summarized, and associations between specific ECG findings and cardiac allograft use for transplantation were studied. ECG abnormalities were present in 51% of all cases reviewed. The most common abnormalities included voltage criteria for left ventricular hypertrophy, prolongation of the corrected QT interval, and repolarization changes (ST/T wave abnormalities). Fifty-seven percent of potential cardiac allografts in this cohort were accepted for transplantation. Left ventricular hypertrophy on ECG was a strong predictor of allograft nonuse. No significant associations were seen among corrected QT interval prolongation, repolarization changes, and allograft use for transplantation after adjusting for donor clinical variables and echocardiographic findings.

          Conclusions

          We have performed the first comprehensive study of ECG findings in potential donors for cardiac transplantation. Many of the common ECG abnormalities seen in organ donors may result from the heightened state of sympathetic activation that occurs after brain stem herniation and are not associated with allograft use for transplantation.
            Cover page: Electrocardiographic Characteristics of Potential Organ Donors and Associations With Cardiac Allograft Use
            • Article
            • Peer Reviewed
            UC Berkeley Previously Published Works (2016)
            Comparing the relative fit of competing models can be used to address many different scientific questions. In classical statistics one can, if appropriate, use likelihood ratio tests and information based criterion, whereas clinical medicine has tended to rely on comparisons of fit metrics like C-statistics. However, for many data adaptive modelling procedures such approaches are not suitable. In these cases, statisticians have used cross-validation, which can make inference challenging. In this paper we propose a general approach that focuses on the "conditional" risk difference (conditional on the model fits being fixed) for the improvement in prediction risk. Specifically, we derive a Wald-type test statistic and associated confidence intervals for cross-validated test sets utilizing the independent validation within cross-validation in conjunction with a test for multiple comparisons. We show that this test maintains proper Type I Error under the null fit, and can be used as a general test of relative fit for any semi-parametric model alternative. We apply the test to a candidate gene study to test for the association of a set of genes in a genetic pathway.
              Cover page: Testing the Relative Performance of Data Adaptive Prediction Algorithms: A Generalized Test of Conditional Risk Differences
              • Article
              • Peer Reviewed
              UC San Francisco Previously Published Works (2013)

              Objective

              Mortality associated with acute lung injury remains high. Early identification of acute lung injury prior to onset of respiratory failure may provide a therapeutic window to target in future clinical trials. The recently validated Lung Injury Prediction Score identifies patients at risk for acute lung injury but may be limited for routine clinical use. We sought to empirically derive clinical criteria for a pragmatic definition of early acute lung injury to identify patients with lung injury prior to the need for positive pressure ventilation.

              Design

              Prospective observational cohort study.

              Setting

              Stanford University Hospital.

              Patients

              We prospectively evaluated 256 patients admitted to Stanford University Hospital with bilateral opacities on chest radiograph without isolated left atrial hypertension.

              Interventions

              None.

              Measurements and main results

              Of the 256 patients enrolled, 62 patients (25%) progressed to acute lung injury requiring positive pressure ventilation. Clinical variables (through first 72 hr or up to 6 hr prior to acute lung injury) associated with progression to acute lung injury were analyzed by backward regression. Oxygen requirement, maximal respiratory rate, and baseline immune suppression were independent predictors of progression to acute lung injury. A simple three-component early acute lung injury score (1 point for oxygen requirement > 2-6 L/min or 2 points for > 6 L/min; 1 point each for a respiratory rate ≥ 30 and immune suppression) accurately identified patients who progressed to acute lung injury requiring positive pressure ventilation (area under the receiver-operator characteristic curve, 0.86) and performed similarly to the Lung Injury Prediction Score. An early acute lung injury score greater than or equal to 2 identified patients who progressed to acute lung injury with 89% sensitivity and 75% specificity. Median time of progression from early acute lung injury criteria to acute lung injury requiring positive pressure ventilation was 20 hours.

              Conclusions

              This pragmatic definition of early acute lung injury accurately identified patients who progressed to acute lung injury prior to requiring positive pressure ventilation. Pending further validation, these criteria could be useful for future clinical trials targeting early treatment of acute lung injury.
                Cover page: Early Acute Lung Injury
                • Article
                • Peer Reviewed
                UC Berkeley Previously Published Works (2013)
                Inhibition of DAF-2 (insulin-like growth factor 1 [IGF-1] receptor) or RSKS-1 (S6K), key molecules in the insulin/IGF-1 signaling (IIS) and target of rapamycin (TOR) pathways, respectively, extend lifespan in Caenorhabditis elegans. However, it has not been clear how and in which tissues they interact with each other to modulate longevity. Here, we demonstrate that a combination of mutations in daf-2 and rsks-1 produces a nearly 5-fold increase in longevity that is much greater than the sum of single mutations. This synergistic lifespan extension requires positive feedback regulation of DAF-16 (FOXO) via the AMP-activated protein kinase (AMPK) complex. Furthermore, we identify germline as the key tissue for this synergistic longevity. Moreover, germline-specific inhibition of rsks-1 activates DAF-16 in the intestine. Together, our findings highlight the importance of the germline in the significantly increased longevity produced by daf-2 rsks-1, which has important implications for interactions between the two major conserved longevity pathways in more complex organisms.
                  Cover page: Germline signaling mediates the synergistically prolonged longevity produced by double mutations in daf-2 and rsks-1 in C. elegans.
                  • Article
                  • Peer Reviewed
                  UC Santa Cruz Previously Published Works (2017)

                  Background

                  Differences in NICU quality of care provided to very low birth weight (<1500 g) infants may contribute to the persistence of racial and/or ethnic disparity. An examination of such disparities in a population-based sample across multiple dimensions of care and outcomes is lacking.

                  Methods

                  Prospective observational analysis of 18 616 very low birth weight infants in 134 California NICUs between January 1, 2010, and December 31, 2014. We assessed quality of care via the Baby-MONITOR, a composite indicator consisting of 9 process and outcome measures of quality. For each NICU, we calculated a risk-adjusted composite and individual component quality score for each race and/or ethnicity. We standardized each score to the overall population to compare quality of care between and within NICUs.

                  Results

                  We found clinically and statistically significant racial and/or ethnic variation in quality of care between NICUs as well as within NICUs. Composite quality scores ranged by 5.26 standard units (range: -2.30 to 2.96). Adjustment of Baby-MONITOR scores by race and/or ethnicity had only minimal effect on comparative assessments of NICU performance. Among subcomponents of the Baby-MONITOR, non-Hispanic white infants scored higher on measures of process compared with African Americans and Hispanics. Compared with whites, African Americans scored higher on measures of outcome; Hispanics scored lower on 7 of the 9 Baby-MONITOR subcomponents.

                  Conclusions

                  Significant racial and/or ethnic variation in quality of care exists between and within NICUs. Providing feedback of disparity scores to NICUs could serve as an important starting point for promoting improvement and reducing disparities.
                    Cover page: Racial/Ethnic Disparity in NICU Quality of Care Delivery
                    • Article
                    • Peer Reviewed
                    UCLA Previously Published Works (2020)

                    Background

                    Despite the well-documented negative impact of untreated bipolar illness, approaches to early intervention in childhood-onset bipolar and related disorders are not well delineated.

                    Methods

                    We reviewed the extant treatment literature on children at high risk for bipolar disorder, with definitions based on family history, childhood adversity, and prodromal symptoms.

                    Results

                    A panoply of approaches have been described, but most interventions are based on an inadequate database to support their routine implementation. We classify early stage interventions as a function of their safety and tolerability with the hope that these might generate more rigorous study and a stronger database.

                    Limitations

                    Critics may rightly argue that identifying viable treatment methods is premature given our lack of ability to reliably predict illness trajectory in very young children. However, many of the psychosocial and pharmacological interventions we present could have nonspecific positive effects across a variety of symptoms, syndromes, and diagnoses, further enhancing the rationale for more rigorous study.

                    Conclusions

                    Early stage interventions have the potential to improve functioning in prodromal illness and exert long-term positive effects on the course of illness. Many of the safest interventions deserve consideration for implementation and dissemination studies.
                      Cover page: Toward prevention of bipolar disorder in at-risk children: Potential strategies ahead of the data
                      Top