Web Extras

Biomedical informatics researcher joins Geisel

Amar Das

Amar Das, M.D., Ph.D., a researcher in biomedical informatics, joined the faculty at Geisel School of Medicine on May 1, 2012. Das came to Geisel from Stanford, where he was assistant professor of medicine and of psychiatry and behavioral science. At Geisel, Das is director of the new Center for Biomedical Informatics. He is also working with Synergy, Dartmouth's center for clinical and translational science, to bring together datasets from DHMC, the Dartmouth Atlas, and other areas to create a central data resource for clinical research. Dartmouth Medicine sat down with Das to talk more about his background and upcoming projects at Geisel.

What are your primary research interests?
My research focuses on biomedical informatics, which is the application of computational technologies and methods to biomedical problems. I've been focused particularly on the informatics challenges faced by clinical researchers—supporting their work on database design, data analysis, and data visualizations. In this area, my primary work deals with the complexity of querying and analyzing longitudinal clinical data.

Can you describe some research projects you have worked on?
I worked with the Stanford HIV database group to look at patterns of drug resistance and to create tools for researchers to be able to discover these patterns and figure out which mutations are associated with certain new drugs. One of the primary reasons for treatment failure in HIV care is drug resistance, but which HIV mutations may be responsible are not clear until you collect large amounts of data so you can see which clinical patterns are emerging from new mutations.

I also worked on a project at Stanford that looked at electronic medical records to discover which patterns of care women were receiving in breast cancer treatment. Some women with certain early-stage diagnoses will get surgery, some will get surgery followed by chemotherapy, others might get chemotherapy before surgery—each of these are acceptable practices but it's not clear which ones are being put into practice and what's driving those patterns of care. Is it women's preferences, institutional factors, or other clinical factors? The goal of the project was to answer those questions.

You and your team have developed web-based visualization tools to help researchers study large sets of data. Can you give an example of this?
One project is SWEETInfo. It's a web-based visualization tool that colleagues and I developed at Stanford, and is the only kind of web accessible, freely available tool available to visualize temporal data. In SWEETInfo, we turn every patient's record into a timeline, and the researcher can use the tool to filter out patients that meet their criteria and can extract certain patterns to focus on, so if they see certain trends they can focus on those areas. This allows researchers to look at large sets of data and identify cohorts and patterns of response for those cohorts, which would be very hard to do if you didn't look visually at the data.

In informatics research, even though we're motivated by specific clinical challenges, we try to create general tools that we can use across different areas. For example, the machine-learning tool that we developed to study drug resistance in HIV can be used to study treatment patterns in breast cancer.

Why did you decide to come to the Geisel School of Medicine?
The opportunity is an exciting one. Dartmouth is committed to developing a leading academic program in biomedical informatics. It is a critical area in terms of clinical research and translational research. Having people who can learn to work with data, analyze the data, and manage the data is all essential because the size of databases is growing rapidly and becoming more complex. We need the expertise to do that.

Dartmouth already has a number of initiatives in place that will support this effort. There is Jason Moore's group focused on the computational biology and bioinformatics components. And Andy Gettinger is running the clinical informatics program within Dartmouth-Hitchcock and taking advantage of the new implementation of Epic. Synergy, which I will be involved in, will also provide core facilities for clinical researchers who are looking to use informatics tools to mine and analyze data. I will be working very closely with the biostatisticians and epidemiologists who as part of Synergy will be advising on study design and statistical analysis. At Synergy I am hoping to look at how to use SWEETIinfo or other tools to make data much more accessible to clinical researchers so they can visualize and query the data.

You are teaching as well?
Yes, I will be involved in the new Quantitative Biomedical Sciences Program. Ultimately we're planning that the Center for Bioinformatics, which I'll be heading up, will become a department and have an informatics degree-granting program.

What is a research question you are burning to figure out at this point in your career?
I think one of the big challenges that we have, as we are collecting all this data, is to understand what the practice of medicine is in the real world. Medicine is changing rapidly due to a variety of factors, so we want to be able to discover practice variation automatically from the data, and associate them with outcomes, and then give feedback to clinical researchers in real time. This requires a lot of different types of informatics tools, infrastructure, and computational algorithms.

How would you describe the culture at Dartmouth?
In upper New England, people get to know each other pretty quickly, it's very friendly. Collaborations seem very easy to make happen. There's a sense of everyone being in the same boat and trying to get things done. Often getting access to data is not just about the technological challenges, it's also about the social factors and policies that go into place. One thing I'm excited about is the initiative that Dartmouth is taking in the area of health-care delivery science. At Dartmouth, not only can we can help discover what patterns and outcomes are in the data but we can also make changes to the health-care system and deliver something that's better. The data is essential to understanding what's happening, but then we need tools to be able to act upon that, technologies that will help to make changes to the health-care environment to provide better care.