Google Research

Katherine Heller


I work at Google Medical Brain. I am also an Assistant Professor in Statistical Science at Duke University, where I also hold secondary appointments in Computer Science, Electrical and Computer Engineering, Neurobiology, Neurology, and is affiliated with the Center for Cognitive Neuroscience and the Information Initiative at Duke. I received my PhD from the Gatsby Computational Neuroscience Unit at UCL, and was a postdoc at the University of Cambridge on an EPSRC postdoc fellowship, and at MIT on an NSF postdoc fellowship. In terms of grants I am the recent recipient of a couple of Google faculty research awards, a first round BRAIN initiative award from the NSF, a CAREER, a DARPA lifelong learning award, and an ARI award for modeling teams. My PhD thesis work was on coming up with efficient ways to do Bayesian clustering. I did a bunch of work on nonparametric Bayes, social networks, and in computational cognitive science. I am currently interested in the combination of Bayesian statistics and deep neural networks, interacting dynamical techniques, and time series and network modeling in general. I apply the work that I do to problems in medicine and in social sciences. Here are a couple of my recent projects: 1) Using a combination of Gaussian Processes and Recurrent Neural Networks to predict whether a hospital patient is at risk for sepsis. This work has been deployed in the Emergency Department at Duke University Hospital. 2) I have used Gaussian Processes, Infinite Factor Models, and logistic regression to tackle other medical problems such as chronic kidney disease, and the prediction of surgical complications. 3) I developed an app to collect data on people with multiple sclerosis. The problem when you have any chronic illness is that you spend more time outside the medical system (where data might be collected) than in it, so this is an exploration of using mobile devices and other trackers to fill in the missing data. 4) I am using time series, stochastic Bayesian methods to model social networks, ranging from conversation networks (who is most influential in a controversial supreme court case?), to disease networks (is my friend likely to get me sick?), to Duke basketball team networks (how do we construct a winning basketball team?). On a personal level: I grew up in the Bronx, and in the northwest suburbs of NYC. I went to school on long island, and in Manhattan. I lived for a decade in the UK, a year each in Oxford and Cambridge, the rest of the time in London. Over a year in Boston, and, most recently, 6 years in North Carolina. I have two children, 4 and 6 years old.

Research Areas

Learn more about how we do research

We maintain a portfolio of research projects, providing individuals and teams the freedom to emphasize specific types of work