Below is a list of current projects. Please also see the students' pages for additional projects. Previous projects can also be found through the following links: censored data and causal inference, computational biology, data-adaptive learning, and multiple hypothesis testing.
Current Projects
Biomedical Informatics for Critical Care
Intel Corporation
P.I. Stuart Russell, PhD and Geoffrey Manley, MD, PhD
Here we propose to bring together a powerful multidisciplinary team from academic medicine (UCSF), computer science (UCB), and industry (Intel’s Digital Health Group) to develop a high performance information system and novel computational techniques for critical care medicine. To create the infrastructure and methods needed to achieve this goal, we will develop a scalable warehouse for critical care data to facilitate multi-institutional collaboration and knowledge discovery. In parallel, we will explore the potential of advanced informatics methods to improve patient classification, prognostic accuracy, and clinical decision-making. We believe that this multidisciplinary approach will ultimately fuel the development of a new generation of ICU information systems to improve the diagnosis, treatment and outcome of critically ill and injured patients.
Fresno Asthmatic Children’s Environment Study
National Institutes of Health/National Heart, Lung and Blood Institute
P.I. Ira Tager
The specific aims of the study are the following: 1) to evaluate the long-term health effects of exposure to air pollutants/bioaerosols on symptoms, asthma severity and growth of lung function; 2) to evaluate the extent to which genes involved in defense against oxidative stress influence short and long-term pollutant effects; 3) to evaluate the effects of interactions between exposures to traffic related pollutants and bioaerosals on both short-term and long-term pollutant effects; 4) to use methods of causal analysis (marginal structural models) and compare the quantitative inferences between causal and association analyses.
Toxic Substances in the Environment, Subproject Core D, Biostatistics and Computing
National Institutes of Health/National Institute of Environmental Health Sciences
P.I. Martyn Smith
The major goals of this project are to provide project investigators with consultative support in biostatistics, and to support computer-based communication and database needs.
Statistical Methods to Study the Epidemiology of HIV & Other Diseases
National Institutes of Health
P.I. Nicholas Jewell
Development of statistical techniques for the analysis of a variety of data sets relating to HIV disease, uterine fibroids and other diseases including STIs.
Statistical Techniques for Complex Environmental Epidemiological Studies
National Institutes of Health
P.I. Nicholas Jewell
This study is devoted to the development of statistical techniques to understand the effect of environmental exposures on the onset of disease in the presence of diagnosis data and assess the effects of multiple environmental exposures on a variety of pregnancy and child development outcomes.
Pregnancy Outcomes in Polycystic Ovary Syndrome
Kaiser Permanente Division of Research (NIH Prime)
P.I. Joan Lo
This is an epidemiologic study focused on the pregnancy outcomes of women with polycystic ovary syndrome using retrospective data obtained from Kaiser Permanente Northern California. Dr. van der Laan will provide oversight and input in regarding to modeling approaches and will supervise the work of the Assistant Researcher.
Targeted Empirical Super Learning in HIV Research
National Institutes of Health
P.I. Mark van der Laan
This project will develop a general statistical methodology, called Targeted Empirical Learning, into a practical product that can be applied to answer scientific questions. Specific applications include treatment rules for HIV-infected patients; activity levels in the elderly; mutation of HIV-virus for prediction of clinical response to drug combinations; measures of variable importance/causal effects of air pollution components on asthmatic children.
Biological Response Indicators of Environmental Stress Center: Project 1 protein Adducts as Molecular Signatures of Carcinogen Dose
National Institutes of Health
P.I. Steven Rappaport
The major goal of Project 1 is to demonstrate the viability of protein adductomics as a true omics approach.
Center for Integrative Research on Childhood Leukemia and the Environment: Project 2 - Exposure Assessment for Childhood Leukemia
National Institutes of Health/National Institute of Environmental Health Sciences
P.I. Patricia Buffler
Project Leader: Stephen Rappaport
A project to assess exposures to persistent contaminants present in homes that may cause leukemia, based upon analysis of house dust, blood collected at the time of the diagnosis of leukemia cases, and archived newborn dried blood spots collected at birth.
Contemporary Treatment and Outcomes for
Atrial Fibrillation in Clinical Practice
National Institutes of Health/National Heart, Lung and Blood Institute
P.I. Alan Go
Collaboration between two members of the NHLBI-sponsored Cardiovasuclar Research Network (KPNC, KPSC) and a clinical trialist consortium to address the following aims: Aim 1:Develop and test novel risk stratification schemes for adverse outcomes (thromboembolism/stroke, bleeding) in patients with atrial fibrillation on and off anticoagulation from a large-scale community-based cohort and additionally validate risk models against patients enrolled in randomized clinical trials. Aim 2: Establish and characterize contemporary registries of incident and prevalent atrial fibrillation within very large, diverse community-based populations to provide critical insights into current outcome event rates and practice patterns, potential health disparities, and to facilitate more rapid enrollment into future effectiveness studies and clinical trials. Aim 3: Identify and validate novel statistical approaches for conducting comparative effectiveness (CE) studies (eg, antithrombotic therapy, catheter ablation) in observational studies and clinical trials
Recent Abstracts
Methodology Development in Comparative Effectiveness Research
Jennifer Creasman, Susan Gruber, Mark van der Laan
We have three objectives for this project. First, we hope to bridge the gap between scientific progress and broad adoption by industry, academic, and government agencies by developing professional, user-friendly super learner and targeted maximum likelihood estimation (SL-TMLE) software that will allow researchers to compare patient management strategies, identify which strategies work best for specific subgroups and predict outcomes based on individual clinical and demographic characteristics. Second, we aim to demonstrate the superiority of SL-TMLE by analyzing simulated data based on a broad range of existing “real” datasets. Collaborators have pledged access to ten datasets that address some of the priority areas outlined by the Institute of Medicine including cardiovascular disease, functional limitations and HIV/AIDS. These collaborations will result in publications demonstrating SL-TMLE to solve outstanding data analysis problems in CER. Finally, we aim to increase the understanding and use of SL-TMLE throughout the research community by disseminating SL-TMLE materials via publications and presentations.
