Title: Don’t forget the notes: Why NLP is key to health care transformation
Abstract: The majority of clinical information useful for patient care and research is locked in clinical notes and only accessible with great pain and effort. Natural Language Processing has the potential to unlock the information in the notes to support phenotyping for precision medicine, quality improvement, and health services research. This talk will illustrate the potential of NLP through existing applications, will describe the challenges of making NLP a real and scalable solution, and will provide concrete suggestions for how the audience can help NLP reach its potential in health care and discovery.

Biography: Dr. Chapman earned her Bachelor’s degree in Linguistics and her PhD in Medical Informatics from the University of Utah in 2000. From 2000-2010 she was a National Library of Medicine postdoctoral fellow and then a faculty member at the University of Pittsburgh. She joined the Division of Biomedical Informatics at the University of California, San Diego in 2010. In 2013, Dr. Chapman became the chair of the University of Utah, Department of Biomedical Informatics where she continues her research on natural language processing in the context of informatics solutions to problems that vex health care.

Title: An evolutionary biologist's skeptical search for computational biology
Abstract: This talk will explain how, starting with an interest in biology, and also in computers, I gradually learned how to use computers to illuminate problems in evolutionary biology. Along the way I learned about theoretical population genetics, learned why it is not always best to write your theorems down, and how fascination with a problem may indicate that something more important is at stake. I moved from theoretical population genetics to algorithms for inferring evolutionary trees (phylogenies). The statistical viewpoint that was standard in theoretical population genetics turned out to be highly controversial among taxonomists studying evolution, and was also considered unnecessary by computer scientists. Both of these groups of people were wrong. I will argue that computer scientists and biologists should indeed communicate, but that this is best done via a statistician. I will argue that a parametric model based on evolutionary theory is crucial, but that one should beware of believing in it too much. Computation is essential in biology, but I wonder whether there really is a field called Computational Biology. Or ought to be.. In the era of Complex Systems and Big Data, a Simple Systems perspective based on Small Data has distinct advantages. As we reach limits in what genome data can tell us, a concern for efficient use of those data will become important, and an understanding of the effects of statistical noise will prove important, and it should encourage a little more humility.

Biography: Joe Felsenstein grew up in Philadelphia, and attended the University of Wisconsin, where he got involved with theoretical population genetics in the lab of James F. Crow. He went on to do his Ph.D. with Richard Lewontin at the University of Chicago, and a postdoctoral fellowship with Alan Robertson at the Institute of Animal Genetics at the University of Edinburgh. He has since then been a faculty member of the Department of Genetics at the University of Washington, Seattle, and its successor the Department of Genome Sciences, and he is also jointly appointed in the Department of Biology. Although his training was thus in theoretical population genetics, since his graduate work he has also been fascinated by the reconstruction of evolutionary trees (phylogenies). This led him to promote and develop likelihood methods for inference of phylogenies, to apply the bootstrap method to investigating which parts of them are well-supported, and to release the first general program package for inferring phylogenies, PHYLIP, in 1980. He wishes that computational biology textbooks would pay more attention to phylogenies, which are the basic structures for making sense of multispecies data. His work in this area has also led him into the extreme and byzantine conflicts in systematics -- some of his closest friendships in computational phylogenetics were cemented by shared victimization. Joe has received a number of very nice honors which are listed at his online CV, but which false modesty dictates that he not mention here.

Title: Data, Predictions, and Decisions
Abstract: I will describe several projects that highlight directions with the use of machine learning to enhance patient care and to build insights about health and wellbeing. I will first present research on leveraging large amounts of data drawn from electronic health records to predict outcomes and to guide decisions. I will focus on opportunities with reducing readmissions and identifying patients at risk for hospital-associated infection, emphasizing the promise of coupling predictive models with decision analysis. I will reflect on challenging directions with these efforts, including causal inference and transfer learning. Then, I will move to studies of health and well-being from non-traditional sources of data, including the use of anonymized logs of online activities. I will present results on pharmacovigilance, detecting the onset of illness, and building deeper understandings of episodic information needs of patients over phases of illness. I’ll wrap up by discussing several aspirational directions with data, predictions, and decisions.

Biography: Eric Horvitz is technical fellow at Microsoft, where he serves as director of the Microsoft Research lab at Redmond. His interests span theoretical and practical challenges with computing systems that learn from data and that can perceive, reason, and decide. His efforts and collaborations have led to fielded systems in the areas of transportation, healthcare, ecommerce, and operating systems. Eric received MD and PhD degrees at Stanford University. He has been elected fellow of the National Academy of Engineering (NAE), AAAI, ACM, AAAS, and the American Academy of Arts and Sciences. He received the Feigenbaum Prize and the ACM-AAAI Allen Newell Award for his research contributions. He currently serves on the Board of Regents of the National Library of Medicine, the Computer Science and Telecommunications Board (CSTB), and the advisory board for the Center for Causal Discovery at the University of Pittsburgh. More information can be found at http://research.microsoft.com/~horvitz.