Emily B. Fox
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Emily B. Fox has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
- Alex Tank
University of Washington
1 shared publication - Ali Shojaie
Department of Biostatistics, University of Washington, Seattle, WA 98195, U.S.A.
1 shared publication - Christopher Xie
University of Washington
1 shared publication - Colm O’dushlaine
Insitro, South San Francisco
1 shared publication - Ian Covert
Department of Computer Science, University of Washington, Seattle, WA, USA
1 shared publication - James A. Landay
Stanford University, Stanford, CA, USA
1 shared publication
Latest Publications
- Open Access
Vascular waveform analysis using Bayesian pulse deconvolution
Vascular waveforms, which measure bulk flow in blood vessels, are widely used to measure vital signs, diagnose conditions, and predict long-term health outcomes. Analyzing vascular waveforms depends on three fundamentally interdependent tasks: signal filtering, pulse timing detection, and pulse shape extraction. We hypothesized that Bayesian pulse deconvolution can achieve improved performance on all three tasks by solving them jointly. This method uses an analytical, generative model of vascular waveforms with priors informed by physical and biological domain knowledge. In simulations, Bayesian pulse deconvolution achieves better performance on all tasks compared with existing algorithms: 90% reduction of median filtering error, 60% reduction in pulse timing error, and 85% reduction in shape extraction error. The advantages in simulations extend to human recordings of photoplethysmography waveforms. Taking real time-synchronized electrocardiogram R-R intervals as a proxy ground truth, Bayesian pulse deconvolution achieves 40% lower pulse interval estimation error (RMSE = 5.1 ms) compared with typical algorithms (RMSE = 8.3 ms, p=1e–10). By extracting more accurate and informative insights from vascular waveforms, Bayesian pulse deconvolution could advance a wide array of health technologies that rely on interpreting signals from blood vessels.
Open AccessVascular waveform analysis using Bayesian pulse deconvolution
Vascular waveforms, which measure bulk flow in blood vessels, are widely used to measure vital signs, diagnose conditions, and predict long-term health outcomes. Analyzing vascular waveforms depends on three fundamentally interdependent tasks: signal filtering, pulse timing detection, and pulse shape extraction. We hypothesized that Bayesian pulse deconvolution can achieve improved performance on all three tasks by solving them jointly. This method uses an analytical, generative model of vascular waveforms with priors informed by physical and biological domain knowledge. In simulations, Bayesian pulse deconvolution achieves better performance on all tasks compared with existing algorithms: 90% reduction of median filtering error, 60% reduction in pulse timing error, and 85% reduction in shape extraction error. The advantages in simulations extend to human recordings of photoplethysmography waveforms. Taking real time-synchronized electrocardiogram R-R intervals as a proxy ground truth, Bayesian pulse deconvolution achieves 40% lower pulse interval estimation error (RMSE = 5.1 ms) compared with typical algorithms (RMSE = 8.3 ms, p=1e–10). By extracting more accurate and informative insights from vascular waveforms, Bayesian pulse deconvolution could advance a wide array of health technologies that rely on interpreting signals from blood vessels.
- Open Access
A Scalable Framework for Identifying Allelic Series from Summary Statistics
Genes for which a dose-response relationship exists between mutational severity and phenotypic impact make for logical therapeutic targets, as the effects of pharmacological modulation can be anticipated from the natural variation present in a population. We refer to genes where such a dose-response relationship exists as harboring allelic series, and have introduced the rare coding-variant allelic series test (COAST) for their identification. The original COAST required access to individual-level data. However, such data are often unavailable due to privacy concerns or logistical challenges. Meanwhile, single-variant summary statistics of the type produced by genome-wide association studies are plentiful. Here we introduce COAST-SS, an extension of COAST that accepts summary statistics as input, namely the per-variant effect sizes and standard errors, along with estimates of the minor allele frequency and local linkage disequilibrium (LD). As a running example, we consider identifying allelic series for circulating lipid traits, drawing on data from the UK Biobank, Million Veterans Program, and Trans-Omics of Precision Medicine Program. Through extensive analyses of real and simulated data, we demonstrate that COAST-SS provides p-values effectively equivalent to those from the original COAST. Interestingly, we find that when LD is low, as is expected among rare variants, COAST-SS is robust to misspecification of the LD matrix, providing valid inference even when the LD matrix is set to the identity matrix. We explore several strategies for annotating the pathogenicity of variants supplied to COAST-SS, finding that they often yield similar power for detecting candidate allelic series. Lastly, we employ COAST-SS to screen for lipid trait allelic series in a meta-analyzed cohort of up to 840K subjects. COAST-SS has been incorporated into the publicly available AllelicSeries R package.
Open AccessA Scalable Framework for Identifying Allelic Series from Summary Statistics
Genes for which a dose-response relationship exists between mutational severity and phenotypic impact make for logical therapeutic targets, as the effects of pharmacological modulation can be anticipated from the natural variation present in a population. We refer to genes where such a dose-response relationship exists as harboring allelic series, and have introduced the rare coding-variant allelic series test (COAST) for their identification. The original COAST required access to individual-level data. However, such data are often unavailable due to privacy concerns or logistical challenges. Meanwhile, single-variant summary statistics of the type produced by genome-wide association studies are plentiful. Here we introduce COAST-SS, an extension of COAST that accepts summary statistics as input, namely the per-variant effect sizes and standard errors, along with estimates of the minor allele frequency and local linkage disequilibrium (LD). As a running example, we consider identifying allelic series for circulating lipid traits, drawing on data from the UK Biobank, Million Veterans Program, and Trans-Omics of Precision Medicine Program. Through extensive analyses of real and simulated data, we demonstrate that COAST-SS provides p-values effectively equivalent to those from the original COAST. Interestingly, we find that when LD is low, as is expected among rare variants, COAST-SS is robust to misspecification of the LD matrix, providing valid inference even when the LD matrix is set to the identity matrix. We explore several strategies for annotating the pathogenicity of variants supplied to COAST-SS, finding that they often yield similar power for detecting candidate allelic series. Lastly, we employ COAST-SS to screen for lipid trait allelic series in a meta-analyzed cohort of up to 840K subjects. COAST-SS has been incorporated into the publicly available AllelicSeries R package.
- Top Cited Papers
- 11 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 44 (01628828) , 1
Neural Granger Causality
While most classical approaches to Granger causality detection assume linear dynamics, many interactions in real-world applications, like neuroscience and genomics, are inherently nonlinear. In these cases, using linear models may lead to inconsistent estimation of Granger causal interactions. We propose a class of nonlinear methods by applying structured multilayer perceptrons (MLPs) or recurrent neural networks (RNNs) combined with sparsity-inducing penalties on the weights. By encouraging specific sets of weights to be zero—in particular, through the use of convex group-lasso penalties—we can extract the Granger causal structure. To further contrast with traditional approaches, our framework naturally enables us to efficiently capture long-range dependencies between series either via our RNNs or through an automatic lag selection in the MLP. We show that our neural Granger causality methods outperform state-of-the-art nonlinear Granger causality methods on the DREAM3 challenge data. This data consists of nonlinear gene expression and regulation time courses with only a limited number of time points. The successes we show in this challenging dataset provide a powerful example of how deep learning can be useful in cases that go beyond prediction on large datasets. We likewise illustrate our methods in detecting nonlinear interactions in a human motion capture dataset.
- 11 March 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 44 (01628828) , 1
Top Cited PapersNeural Granger Causality
While most classical approaches to Granger causality detection assume linear dynamics, many interactions in real-world applications, like neuroscience and genomics, are inherently nonlinear. In these cases, using linear models may lead to inconsistent estimation of Granger causal interactions. We propose a class of nonlinear methods by applying structured multilayer perceptrons (MLPs) or recurrent neural networks (RNNs) combined with sparsity-inducing penalties on the weights. By encouraging specific sets of weights to be zero—in particular, through the use of convex group-lasso penalties—we can extract the Granger causal structure. To further contrast with traditional approaches, our framework naturally enables us to efficiently capture long-range dependencies between series either via our RNNs or through an automatic lag selection in the MLP. We show that our neural Granger causality methods outperform state-of-the-art nonlinear Granger causality methods on the DREAM3 challenge data. This data consists of nonlinear gene expression and regulation time courses with only a limited number of time points. The successes we show in this challenging dataset provide a powerful example of how deep learning can be useful in cases that go beyond prediction on large datasets. We likewise illustrate our methods in detecting nonlinear interactions in a human motion capture dataset.
- 1 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 6596-6603
A Simple Adaptive Tracker with Reminiscences
- 1 May 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 6596-6603
A Simple Adaptive Tracker with Reminiscences
- Open Access
Autoregressive Models for Variance Matrices: Stationary Inverse Wishart Processes
We introduce and explore a new class of stationary time series models for variance matrices based on a constructive definition exploiting inverse Wishart distribution theory. The main class of models explored is a novel class of stationary, first-order autoregressive (AR) processes on the cone of positive semi-definite matrices. Aspects of the theory and structure of these new models for multivariate "volatility" processes are described in detail and exemplified. We then develop approaches to model fitting via Bayesian simulation-based computations, creating a custom filtering method that relies on an efficient innovations sampler. An example is then provided in analysis of a multivariate electroencephalogram (EEG) time series in neurological studies. We conclude by discussing potential further developments of higher-order AR models and a number of connections with prior approaches.
Open AccessAutoregressive Models for Variance Matrices: Stationary Inverse Wishart Processes
We introduce and explore a new class of stationary time series models for variance matrices based on a constructive definition exploiting inverse Wishart distribution theory. The main class of models explored is a novel class of stationary, first-order autoregressive (AR) processes on the cone of positive semi-definite matrices. Aspects of the theory and structure of these new models for multivariate "volatility" processes are described in detail and exemplified. We then develop approaches to model fitting via Bayesian simulation-based computations, creating a custom filtering method that relies on an efficient innovations sampler. An example is then provided in analysis of a multivariate electroencephalogram (EEG) time series in neurological studies. We conclude by discussing potential further developments of higher-order AR models and a number of connections with prior approaches.