Streamlined empirical Bayes estimation for contextual bandits with applications in mobile health

Frontiers in Biostatistics Webinar

Marianne Menictas
Postdoctoral Fellow, Department of Statistics
Harvard University

Mobile health (mHealth) technologies are increasingly being employed to deliver interventions to users in their natural environments. With the advent of increasingly sophisticated sensing devices (e.g., GPS) and phone-based EMA, it is becoming possible to deliver interventions at moments when they can most readily influence a person’s behavior. For example, for someone trying to increase physical activity, moments when the person can be active are critical decision points when a well-timed intervention could make a difference. The promise of mHealth hinges on the ability to provide interventions at times when users need the support and are receptive to it. Thus, our goal is to learn the optimal time and intervention for a given user and context. A significant challenge to learning is that there are often only a few opportunities per day to provide treatment. Additionally, when there is limited time to engage users, a slow learning rate can pose problems, potentially raising the risk that users will abandon the intervention. To prevent disengagement, a learning algorithm should learn quickly in spite of noisy measurements. To accelerate learning, information may be pooled across users and time in a dynamic manner, combining a contextual bandit algorithm with a Bayesian random effects model for the reward function. As information accumulates, however, tuning user and time specific hyperparameters becomes computationally intractable. In this talk, we focus on solving this computational bottleneck.

noah simon

Reframing proportional-hazards modeling for large time-to-event datasets with applications to deep learning

Frontiers in Biostatistics Seminar

Noah Simon, Ph.D.
Associate Professor
Department of Biostatistics
University of Washington

To build inferential or predictive survival models, it is common to assume proportionality of hazards and fit a model by maximizing the partial likelihood. This has been combined with non-parametric and high dimensional techniques, eg. spline expansions and penalties, to flexibly build survival models.

New challenges require extension and modification of that approach. In a number of modern applications there is interest in using complex features such as images to predict survival. In these cases, it is necessary to connect more modern backends to the partial likelihood (such as deep learning infrastructures based on eg. convolutional/recurrent neural networks). In such scenarios, large numbers of observations are needed to train the model. However, in cases where those observations are available, the structure of the partial likelihood makes optimization difficult (if not completely intractable).

In this talk we show how the partial likelihood can be simply modified to easily deal with large amounts of data. In particular, with this modification, stochastic gradient-based methods, commonly applied in deep learning, are simple to employ. This simplicity holds even in the presence of left truncation/right censoring, and time-varying covariates. This can also be applied relatively simply with data stored in a distributed manner.

Data Science Zoominar: Increasing Diversity in Data Science

A conversation with Emma Benn, DrPh
Associate Professor, Center for Biostatistics and Department of Population Health Science and Policy
Icahn School of Medicine at Mount Sinai

Moderator: Rafael Irizarry
RSVP at https://bit.ly/DSAugust11

Data Science Zoominar: Data-driven Movement Monitoring During the Pandemic

A conversation with Caroline Buckee, Associate Professor of Epidemiology, Harvard TH Chan School of Public Health
Moderator: Rafael Irizarry
RSVP at https://bit.ly/DSAugust4

Collin Tokheim Headshot

Collin Tokheim, PhD Wins New Damon Runyon Cancer Research Foundation Quantitative Biology Fellowship

Collin Tokheim, PhD, a postdoc in the X. Shirley Liu lab, has been awarded a Damon Runyon Cancer Research Foundation Quantitative Biology Fellowship. Dr. Tokheim’s mentors are X. Shirley Liu, PhD, and Eric S. Fischer, PhD, at Dana-Farber Cancer Institute, Boston.

“The first class of Damon Runyon Quantitative Biology Fellowship Awardees launched their research in novel directions that may lead to the next breakthroughs in cancer research. Nine brilliant young scientists will apply their quantitative skills to design innovative experiments and interpret massive data sets that may help solve important biological and clinical problems. The awardees were selected by a distinguished committee of leaders in the field.

Dr. Tokheim is developing computational models that can identify degradable proteins that are linked to the development of human cancers. Unlike traditional drugs that bind and block the activity of key proteins in cancer cells, a new generation of drugs can eliminate proteins by hijacking the protein degradation machinery within cells. By leveraging big data from thousands of tumor profiles and a novel statistical and deep learning model, he will conduct an unbiased search for candidate proteins that can be verified experimentally. This research may lead to the development of drugs targeting protein degradation as a potent and selective way to treat a variety of human cancers.

Each postdoctoral scientist selected for this unique three-year award will receive independent funding ($240,000 total) to train under the joint mentorship of an established computational scientist and a cancer biologist. Damon Runyon has created this new funding mechanism to encourage quantitative scientists (from fields such as mathematics, physics, computer science and engineering) to pursue careers in cancer research. This support will help create an elite cadre of computational biology leaders trained in quantitative and biological sciences—scientists who are capable of traversing both worlds with ease and are comfortable speaking both languages fluently.

“Because this is in essence a new field at the nexus of traditional cancer research and data science, it is critical to draw fearless and brilliant young computational scientists to these problems to drive the field forward,” said Aviv Regev, PhD, of the Broad Institute and inaugural Chair of the Quantitative Biology Fellowship Award Committee.”

Full release: Damon Runyon Cancer Research Foundation

rafael irizarry dfci headshot

Rafael Irizarry, PhD Elected International Society of Computational Biology Fellow

Rafa Irizarry, PhD, Professor of Biostatistics at Dana-Farber Cancer Institute and Harvard T.H. Chan School of Public Health, has been elected a 2020 International Society of Computational Biology Fellow.

From ISCB:

“The ISCB Fellows program was created to honor members who have distinguished themselves through outstanding contributions to the fields of computational biology and bioinformatics. Begun in 2009, 2020 marks the 11th anniversary of the program. In early December of each year, ISCB has sought Fellow’s nominations from our members, with eligibility restrictions based on selection criteria focused most heavily on the significance of scientific contributions.

Dr. Irizarry was selected for his pioneering work in expression analysis and development of statistical methods for expression analysis, which are some of the most impactful in the entire field.”