Assessing Variable Importance Nonparametrically using Machine Learning Techniques
Published:
This talk (on a preliminary version of my R-squared variable importance paper published in Biometrics) was selected as the Most Outstanding Oral Paper.
Posts by Collection
portfolio
publications
talks
Nonparametric Variable Importance using an Augmented Neural Network with Multi-Task Learning
Published:
Contributed talk at the Thirty-fifth International Conference on Machine Learning.
Assessing Variable Importance Nonparametrically using Machine Learning Techniques
Published:
This talk (on a preliminary version of my R-squared variable importance paper published in Biometrics) was selected for an ASA Biometrics Section Travel Award.
Assessing Variable Importance Nonparametrically using Machine Learning Techniques
Published:
Invited talk on my dissertation research given at the UW Biostatistics Colloquium.
A Unified Approach to Nonparametric Variable Importance Assessment
Published:
This talk (on a preliminary version of my general variable importance paper published in Journal of the American Statistical Association) was selected for an ASA Nonparametrics Section Travel Award.
Guiding HIV-1 Antibody Regimen Down-selection and Prevention Efficacy Trial Design using Machine Learning
Published:
Contributed talk at the 27th International Dynamics and Evolution of HIV and Other Human Viruses Meeting on SLAPNAP (see publications).
Efficient Nonparametric Statistical Inference on Population Feature Importance using Shapley Values
Published:
We discuss our paper to be published in the Proceedings of the Thirty-seventh International Conference on Machine Learning.
A Unified Approach for Inference for Model-Agnostic Variable Importance
Published:
I discuss a framework for inference on general model-agnostic variable importance measures.
Statistics in Infectious Disease Research
Published:
Guest lecture on some of the ways that statistics is used in infectious disease research in the Computational Biology course at Roanoke Valley Governor’s School.
Statistical Inference and Containerization in Computational Pipelines
Published:
I discuss three approaches towards a more principled use of machine learning: inference on the goodness of fit, inference on variable importance, and containerization.
Inference for Model-Agnostic Variable Importance
Published:
I discuss a framework for inference on general model-agnostic variable importance measures.
Inference for Model-Agnostic Variable Importance
Published:
Keynote talk at the 3rd annual Hutch United Symposium.
Flexible Variable Selection in the Presence of Missing Data
Published:
I gave a talk on work in progress, developing methods for variable selection in settings with missing data that do not rely on (generalized) linear models.
Inference for Model-Agnostic Variable Importance
Published:
I discuss a framework for inference on general model-agnostic variable importance measures.
Model-Agnostic Variable Importance and Selection
Published:
I discuss a framework for inference on general model-agnostic variable importance measures, and how this framework can be used to perform variable selection. I also briefly discuss several directions of current work, including longitudinal variable importance; a measure of how important variables are for tailoring treatment; and fairness-aware variable importance.
Flexible Variable Selection with Missing Data
Published:
I discuss a method for model-agnostic variable selection that is robust to model misspecification and valid in settings with missing data.
Inference for Model-Agnostic Variable Importance
Published:
I discuss a framework for inference on general model-agnostic variable importance measures and possible summary measures for longitudinal variable importance.
Scalable Phenotyping for Safety Outcomes Using Electronic Health Record Data
Published:
I discuss a framework for predicting safety outcomes of interest within the FDA’s Sentinel system using data from electronic health records.
Inference for Model-Agnostic Longitudinal Variable Importance
Published:
I discuss a framework for inference on general model-agnostic variable importance measures and possible summary measures for longitudinal variable importance.
Statistics in Infectious Disease Research
Published:
Lecture on some of the ways that statistics is used in infectious disease research to students in UW Biostatistics 111.
Inference for Model-Agnostic Longitudinal Variable Importance
Published:
I discuss inference for summaries of longitudinal model-agnostic variable importance.
Inference for Model-Agnostic Longitudinal Variable Importance
Published:
I discuss inference for summaries of longitudinal model-agnostic variable importance.
Inference for Model-Agnostic Variable Importance
Published:
In many applications, it is of interest to assess the relative contribution of features (or subsets of features) toward the goal of predicting a response – in other words, to gauge the variable importance of features. In this talk, I will discuss a model-agnostic notion of variable importance and general conditions under which valid inference on the true importance can be obtained, even when machine learning-based techniques are used as part of estimation. We define variable importance as a population-level contrast between the oracle predictiveness of all available features versus all features except those under consideration. I provide several examples of predictiveness measures, including for right-censored outcomes, and illustrate the use of the proposed methods with data from a study of an antibody against HIV-1 infection.
Statistics in Infectious Disease Research
Published:
Lecture on some of the ways that statistics is used in infectious disease research to students in UW Biostatistics 111.
Design considerations for subgroup analyses in cluster-randomized trials based on aggregated individual-level predictors
Published:
In research assessing the effect of an intervention or exposure, a key secondary objective often involves assessing differential effects of this intervention or exposure in subgroups of interest; this is often referred to as assessing effect modification or heterogeneity of treatment effects (HTE). Observed HTE can have important implications for policy, including intervention strategies (e.g., will some patients benefit more from intervention than others?) and prioritizing resources (e.g., to reduce observed health disparities). Analysis of HTE is well understood in studies where the independent unit is an individual. In contrast, in studies where the independent unit is a cluster (e.g., a hospital or school) and a cluster-level outcome is used in the analysis, it is less well understood how to proceed if the HTE analysis of interest involves an individual-level characteristic (e.g., self-reported race) that must be aggregated at the cluster level. Through simulations, we show that only individual-level models have power to detect HTE by individual-level variables; if outcomes must be defined at the cluster level, then there is often low power to detect HTE by the corresponding aggregated variables. We illustrate the challenges inherent to this type of analysis in a study assessing the effect of an intervention on increasing COVID-19 booster vaccination rates at long-term care centers.