SONY DSC

Dr. Amar K. Das, MD, PhD is director of healthcare effectiveness research at IBM Research, where he leads initiatives in outcomes research agenda of the new division by proposing, modeling, and carrying out clinical studies of informatics solutions. Prior to joining IBM, he spent four years at The Geisel School of Medicine at Dartmouth, where he was an Associate Professor of Biomedical Data Science, Psychiatry, and The Dartmouth Institute, and where he is now Adjunct Associate Professor.

Dr. Das has also served as an Assistant Professor of Medicine and of Psychiatry at Stanford University School of Medicine’s Department of Psychiatry and Behavioral Sciences. He was a member of the Stanford Center for Biomedical Informatics Research, as well as an executive member of the Stanford Graduate Program in Biomedical Informatics. At Stanford, he taught courses focused on digital medicine and smart health.

He received his B.A. in Anthropology and Applied Math (MMSS) from Northwestern University in Evanston, Il. He then earned his MD at Stanford University School of Medicine in 1997, where he participated in health policy, and eventually his PhD in Biomedical Informatics in 2002. He completed his training in psychiatry at the Columbia University College of Physicians and Surgeons, where he also served as chief resident and participated in the research track.

I had the pleasure of talking with Dr. Das about his colorful and accomplished career that dabbles in psychiatry, health disparities, informatics, and big data.

ND: I see that you studied anthropology at Northwestern, received both your MD and PhD from Stanford, and trained in psychiatry at Columbia. Can you tell me a little bit more about your unique education?

Dr. Das: My pathway reflects my multiple interests. However, the core of it is the use of technology to shape culture. At Northwestern, I looked at how technology impacts culture and did a project on homeless people. I had a strong interest in helping underserved people and learning how they are affected by technology. I went to medical school with these interests.

In medical school, the possibility of informatics, which I learned about on my rotations, excited me. Before this, I had never known about the impact of medical informatics. I joined the PhD program at Stanford in 1990 when the internet was really in its infancy. I became excited about the possibilities of the impact technology could have on medicine (and beyond!). I finished my informatics training and built and electronic health records from scratch. We wanted to find patients with certain diagnoses so we could provide them with an individually tailored treatment plan.

In the 1990s, there was a huge interest in finding HIV positive patients to enroll them into clinical trials. At the time, we didn’t know what treatments works, and many new drugs were being developed. We used the EHR to find out which patients could partake in which of the many HIV drug trials were going on. This also allowed us to organize information on patients so they could enroll in the best possible trial.

Ultimately, when I finished my medical school degree, I decided the field of greatest interest to me is psychiatrist. This was partially because my father was a psychiatrist and my mother was a social worker, so I had strong interests in mental health. Additionally, psychiatry is one of the fields that allow doctors to have long-term relationships with their patients with frequent follow-ups as opposed to just seeing them for 30 minutes. However, at the time, psychiatry did not intersect much with my medical informatics training.

ND: Can you talk more about your transition from clinical medicine to academia and industry?

Dr. Das: After I graduated from my clinical residency, I spent a number of years practicing clinical psychiatry in two different settings. The first was in an emergency and community setting in upper Manhattan in Columbia-Presbyterian in Washington Heights with primarily undeserved minority patients. I also had a private practice where I saw patients with long-term disorders where I saw patients who needed both psychosocial and medical treatments. A couple of years later, I had the opportunity to go back to Stanford and become a medical researcher. I realized at the time I needed to commit myself 110% into the responsibilities of being a professor between getting grants for research and advising students. I was at Stanford for a total of 8 years.

Then, I was recruited at The Geisel School of Medicine, which was very interested in starting a department of medical informatics. Initially, I was hired to start a department of medical informatics. However, the person hired to create the department of biostatistics and I thought that because of the overlap in big data, that we should join forces. Thus, we started the first department of biomedical data science in the world. We also trained doctoral and master’s students in this. I spent a total of four years here.

In June of 2016, IBM recruited me. While I liked academia, I didn’t feel that I could make a broader impact as much as I would like to. Additionally, the healthcare industry has lagged behind other industries about 20 years of years, which has led to a lot of frustration. Doctors want to enter a lot of complex information about patients that needs to be seen by many other people including patients and providers. I wanted to be involved in the private industry to adopt healthcare to modern-day technologies.

IBM had a new research center to understand how medical technology can affect the healthcare field on a broader level. They wanted to provide evidence to patients, doctors, and stakeholders that medical informatics works in making healthcare more efficient, accessible, and producing health outcomes- in other words, more value. I was very interested in being part of this and saw that they had the tools and capacities to make a large impact.

ND: What are some of the potential benefits in medicine and healthcare of analyzing big data and medical informatics?

Dr. Das: The idea behind big data is not the size of it but the complexity. Oftentimes, data in healthcare sits in silos being unused, but it has a lot of potential to provide valuable medical knowledge. The data can provide a lot of information on pathologies, treatments, laboratory research, diagnosis, etiology, etc. One challenge is that big data is unstructured. About 75% is unstructured- oftentimes in clinical reports, which is some of the most helpful information.

We also like to investigate new patterns in the data. For example, we can look at how doctors have characterized the disease and understand the individual subtypes. This can help us understand which treatments are better for certain patients. For example, some patients might want to get chemotherapy, surgery, and other treatments. However, we don’t know which combination is most effective for treating which patients. Data allows us to look at which treatments work. In mental health, there are also particular subtypes of disorders. For example, in autism, there are different variations such as loss of motor control even though they all come in the same code. By getting a more granular perspective, we can tailor treatments better for patients.

Finally, the application of the knowledge is important. Once we derived insight from the data, we need to figure out a way to provide that value back to proper stakeholders. We can do data summarization to make complex information digestible. Visualization and analytics are helpful tools. This does require some sophistication by the end user to look at these patterns and make conclusions. The challenge here is to come up with a compelling visualization that clearly shows patterns. Additionally, we want to translate the findings of the data and predictions to deliver knowledge in real time at the point of care.

It can also point out outcomes and allow us to assess if we getting better value healthcare for treatments.

ND: What are some of the potential limitations of using big data and medical informatics?

Dr. Das: One big issue is privacy. The data is in silos to protect people’s privacy. People analyzing this data need to be sensitive about this. Security is huge and this data must be protected from people with harmful intentions. We have to do a tradeoff between maintaining privacy and gathering quality data.

Furthermore, not all data is good data. Sometimes, data is not accurate or frankly wrong, which can lead to incorrect conclusions.

ND: Many people often have a negative association between private industry and medicine largely because of the pharmaceutical industry. However, your role at IBM occupies a different intersection between business and healthcare. What are some of the advantages of working in the private industry in terms of improving healthcare as opposed to making profit off peoples’ sickness.

People don’t like hearing profit in the context of healthcare because it comes off as making money off sick people. Currently, our healthcare system is largely a mixture of public (government) and private funding (employer-sponsored).

Assuming capitalization is an important part of the American economy, we need to figure out how money can be used to incentivize better value healthcare. Pharmaceuticals spend a lot of money on research, sometimes with pricing issues. However, there are many other industries in the healthcare field. For example, medical devices, technologies for diagnosing, and monitoring. Additionally, technology is a big element in private funding.

We need to acknowledge that the private industry in the US has a natural role. Much of the innovation in healthcare comes from the private industry, which goes beyond US borders. At IBM, we have a division called Watson Health, which has the goal of transforming healthcare. They have various initiatives, such as collaborating with the Veterans Administration to deliver better cancer treatment. We also use technology in our daily lives, such as google to search for something health-related.

IBM works with different industries to help improve their processes. Healthcare is a fragmented industry and IBM provides a valuable perspective in improving business processes and delivering data. Finally, adoption of technology happens quickly and scales solutions. I partially left academia because of the faster pace of industry compared to academia and the fact that companies come and go if they fail to provide value to the marketplace.

ND: Medical education today is quite lacking in data science. What role do you think data analysis in its many forms (big data, biostatistics) should play in medical education and policy?

Dr. Das: Data science should play a central role in medical training, especially more than it does today. It is critical for doctors to understand how to interpret and apply data science. Today, there are different pressures from a whole bunch of fields that are being impressed into medical education, and medicine is transforming.

However, doctors in the future will benefit greatly from learning techniques of informatics in understanding research and delivering healthcare. We need to provide the tools in medical education to look at EHR, gather information about a patient, integrate the most important, quickly add on the data from wearable devices, and discern good data from bad data. These are some of the things that doctors will need to know as healthcare transforms.