Dr. Peter DeWan, PhD. is the VP of Research & Development at Decision Resources Group where he manages data scientists and analysts as well as extracts value from big data sets. Dr. Dewan is also the Founder of Activate Networks, Inc. where he designed, implemented and managed all products and projects, shepherding it through Series A & B financing followed by acquisition. Dr. DeWan completed his doctorate in sociology and has developed substantial expertise in applying social network analysis to healthcare.
Brigham Hyde, PhD., is the Chief Data Officer and SVP of Analytics at Decision Resources Group where he leads data analytics for Life Sciences. Previously, as CEO and co-founder of Relay Technology Managemen,t he developed proprietary analytics software and data aggregation tools. Dr. Hyde has over 10 years life science experience and brings extensive knowledge of both basic science research and drug development operations. Dr.Hyde received his Ph.D. in Clinical Pharmacology from the Tufts University School of Medicine, where he is currently an adjunct professor of Pharmacology and experimental therapeutics.
ML: If you’re on a plane and you have to give a brief yet informative introduction about what you do and who you are, what would you say?
PD: I’m a sociologist who studies the way that networks of relationships affect behavior. I use that to figure out how to target interventions to change that behavior.
BH: I’m the Chief Data Officer and Head of Analytics for a healthcare information company. I use data and analysis of data to advise strategy and implementation for life science companies, including pharmaceutical medtech, providers and payers.
ML: What excites both of you about these roles?
PD: To be able to solve new kinds of problems for customers and to extract information from data in ways that other people have not thought of yet.
BH: I’m in general a truth-seeker, and trying to find out the truth about what is valuable and effective in healthcare is something I’ve pursued my whole career. Gathering data and being able to analyze it in interesting ways gets me closer to that.
ML: Often people interpret data differently, or analysis is misleading because there is no proper correlation between data. How do you address that?
BH: I think there are two parts to it. One is continuously trying to build a single-source of truth data sets, particularly in healthcare in the US and globally, that contains all the relevant information. The other thing is that we apply rigor through our knowledge, data science approach, typical models of pros and cons. Plus when they should be applied, the contextual understanding of the question being asked, and the dynamics within the data. It’s knowing and acknowledging biases in designing and implementing models that are correctly chosen and correctly interpreted.
ML: Can you tell us more about the data decision group and how you ended up working there?
BH: DRG is an analytics, healthcare insights, and consulting firm. We’ve been around for about twenty-five years and have grown up through acquisition of different companies. Our data analytics help our clients work with complex data, which is a challenge in and of itself these days. Insights means we have experts, people with PhDs and MDs, who understand certain commercial dynamics and what’s happening in these areas, and can then write insights on that.
I ended up at DRG through the acquisition of the company I founded, Relay Technology Management. Relay’s focus was using high-end mathematical models and interesting data to predict the behavior of the market in the pharmaceutical industry: Which drugs should be developed, which new ideas were likely to succeed and which weren’t. That was also what made it a fit for where DRG was heading, as they were adding more technology and data infrastructure to their team.
PD: I joined Decision Resources when my startup, Activate Networks, was acquired last March. My company focused on applications of network science, which I had been developing at Harvard. As it turns out, pharma was one of the more successful applications of that, and what we were doing, as I noted, was drawing sets of relationships. We were figuring out, by using commercial data, how to measure the relationships that physicians have, and the degree to which they influence each other’s behavior, in order to then create a maximized behavior-change plan from within that network structure.
ML: Different types of organizations and industries are trying to use data to solve various business challenges. What do you think makes the healthcare field different or unique from others?
BH: I began my career as a PhD scientist working on basic biology, biochemistry, and trying to understand ways to develop new therapies. Through my involvement in that, I began to realize that many of the opportunities to improve outcomes – improve patient care, cure cancer, if you will – existed in the information that was already available. It was just being used improperly and inconsistently. One of the things I think that’s interesting about the data business, or data science in healthcare, is that you have the ability to make tangible impact, not just in the businesses we serve but on actual patients.
Whether it’s identifying inconsistent or appropriate care, helping providers operate more efficiently and more cost-effectively, or assisting pharmaceutical companies in identifying opportunity in patient populations that are undertreated or need therapeutic relief, all these things are part of daily life. I don’t know of another industry that has that type of reward system built into it. Not to say that every day is curing cancer and saving kids, but as opposed to pure retail data science, for example, there’s an extra layer that is particularly motivating and there are still a lot of opportunities for substantial impacts to be made.
ML: What future trends do you foresee happening in how analytics and technology are shaping the healthcare industry?
BH: Today, one very tangible area is that the dynamics of how care is paid for are changing. I think one thing we’re seeing in the data we look at is that there’s a shifting burden of cost which is transferring to patients. What this really means in the US, where most folks with insurance are unaware of the actual cost of therapy, is that they’re not really savvy consumers or purchasers of care. They largely rely on their doctors’ decisions. Maybe they have a co-pay, or some type of deductible, but for the most part their care is paid for elsewhere.
We think what’s happening is that payers have shifted a lot of that cost to the patient through a rise in deductible, a rise in co-pay, and constriction of care. Increasingly, consumers are going to need a way to evaluate what’s valuable for them and what’s effective. At the core of that will be the proof coming from healthcare data and healthcare data science. I think there’s still a ways to go on that, but it’s certainly a future statement. I would say that the opportunity for pharmaceutical companies is the ability to pivot their message and marketing to the real benefits of their therapy, beyond the old marketing message of generic Super Bowl ads, to real, evidence-based and value-based care. We definitely see that happening.
PD: I think the entire industry of healthcare, and not just pharma, is substantially behind the curve in terms of use of modern, cutting-edge analytics techniques and methods, despite the fact that it’s a relatively data-rich environment in comparison to other industries. What I really foresee is that over the next five to ten years, the analytic techniques that are currently being used in things like advertising or computer science , are going to start to be used more broadly in healthcare. Not just from the perspective of the pharma and the promotional side of it, but also from the perspective of actual health care outcomes.
Relative to all the trends that have been happening with data science over the last years, I think it’s very noticeable that healthcare has a lot of catch-up work to do, which I expect to be happening fairly soon. Especially as more EHR data becomes available and as that industry starts to consolidate.
BH: Similar to Pete, I’m excited about mapping the potential of social and digital information to become part of a patient story. Whether it’s behavioral information about their activities, or lifestyle habits like diet, exercise, anxiety, depression, and so on. Merging that with the purely clinical world that we look at, which involves blood tests and clinical input, really has the potential to evolve the patient experience, whereas now it’s often driven by endpoints that are purely clinical. We think that patient experience, how the patient is perceiving things, and patient-reported outcome have the potential to be exciting parts of healthcare instead of just focusing on what the clinician is looking at.
ML: In healthcare, privacy has always been tough, for obvious reasons. How do you feel privacy affects what you do with data, and how you’re able to use it? Also, there’s always the question of who actually owns the information collected about a patient. Is it theirs or is it yours to work with?
BH: I think that’s an important question, but I’m increasingly of the mind that it is not a roadblock, but simply a part of the system. We operate in an info-safe environment, but we have constructive relationships with pharms which allow us to safely identify patient information and analyze it in aggregate form to produce insights.
Long-term, I think exciting technologies like Blockchain offer the potential for more transactional environments, potentially then allowing patients to not only participate but potentially profit from the value of their participation. These are solvable IT problems, that’s my feeling. They are not insurmountable and they are not scary. They should be addressed appropriately and thoughtfully, but they are solvable. Ultimately, in an environment where we’d love to move to real-time interventions, we’d love to be in a world where there are digital interventions available, behavioral or otherwise.
I think that while regulations have slowed, when it comes to this industry’s ability to produce that type of real-time interface, I think it’s something we know how to deal with and can. I think we will catch up eventually. I don’t think privacy will be an absolute roadblock. My hope is that it becomes an evidence-based interaction as opposed to “digital snake oil.” What’s important, of course, is continuing the standard of being thoughtful and secure, and I am encouraged by the potential for the consumer and the benefit they can see.
ML: There is always a gap between what we can see and learn from analytics, and how we put it into action. A lot of organizations are interested in real-time data, but fail to apply it as fast as the data becomes available to them. How do you see that being addressed in healthcare? Also, how would you suggest organizations go about the process of changing the organizational structure in order to be better and faster-equipped at capitalizing on the data they receive?
BH: One part of your question is about the infrastructure side, and what could be done to speed up the data access and analytic model. Look at daily life and, for example, retail. Amazon is potentially an example. You can build a lot of infrastructure and analytical modeling in a way that is self-improving. I think the models exist for that. I think healthcare’s starting to have the data frequency and currency that would allow it. We receive updates every night on what’s going on with patients in the US. The only thing stopping us from making analytical recommendations on top of that, nightly, is our own code base and the demand of our clients.
The second part of your question focuses on the demand side, which comes from how people are handling this type of insight change management. Truthfully, I think physicians are not well-oriented to receive analytical insights that they can act on. While we do have a digital standard now in the US, the systems are not well-oriented to receive it.
In fact, my belief is that disruption, and the slowing of that process, has come from the patient. It started a little bit with things like Healthweb or WebMd. All of a sudden, you had patients in the office who had Googled things, saying, “Doctor, I have this.” That forced a major change in which doctors had to address certain things. I think to the extent that Silicon Valley, Boston, New York or wherever can come up with disruptive tools to empower patients, it will force behavior change in physicians, especially if it ends up impacting payment. I think there are two sides to it. The infrastructure side, I’m optimistic about, and I think we’re closer than people think. On the people and demand side, I think it’s going to require a behavioral disruption and I see the patients are driving that.
ML: Do you have any advice for someone just starting in this field? Are there underlying skills or philosophies that you found to be especially effective through your work tenure?
PD: I think what we’re doing is part of a broader process of really improving analytics across the world. Higher-level universities are starting to think about having data science departments. For many of the people that we’re hiring, data science is their aspiration. They want to be in that space. From that perspective, we look at analytical and computer science technical capabilities.
I think what we tend to find, when we hire people, is that they’re strong in one of those areas and not in the other. We work to balance those out. Were I to give somebody advice, I think the first thing I’d say is, “What are you already strong in?” and then, “let’s work on the other side,” because in order to be really good in this field, you have to both understand the numbers that you’re working with and be able to put together the computing system to actually deal with the size of data and the runtime on the analytical problems being solved.
BH: While I want you to have data science skills, like programming ability in Python or tableau experience, which are standard for the healthcare industry, what I want you also to have, more than anything, is problem-solving ability, proven within the healthcare domain. Healthcare is particularly hard for data scientists because it’s not just about knowing mathematical models, or how to write efficient code or structure queries. Folks that I’ve seen struggle in our organization, or in healthcare data science, are people who think knowing how to code and waiting to be told what to code is enough. I need people who can think and code. That’s the key.
I think I wound up in it from the fact of where I began, and learned my data science skills later on in building things. Pete’s experience is somewhat inverse to that, but I know he would say that the more he knows, the better a model he can design, with better outcomes. The only way that I ask somebody to know both sides and believe that they can handle them both, is to demonstrate problem-solving abilities. Data science, more than anything in healthcare, is about running into a possible problem and then proving you’re willing to go over that wall and figure out how to solve it.
ML: Is there a certain gold nugget of truth, aha moment, or anything you’ve discovered that has stood out to you, through your research work, that you feel like many others aren’t aware of, or have misconceptions about?
BH: Yes, I would mention what we’ve learned about oncology care in the US. Some of this is known, but I think what’s more shocking than people realize is that it’s incredibly inconsistent. There are some projects we’ve done in leukemia that look at which patients receive which drug and why. What’s clear above all else is that it’s incredibly varied. There are often details of patients which fly in the face of clear clinical evidence to the contrary. When I see that, it, A, makes my blood boil but, B, tells me that, wow, we can really make a difference for people.
When we talk about cancer, we’re talking about actually saving lives if we’re able to get this information in the right hands at the right time. Those are exciting moments for me, and also ones that I think would shock the average patient who may be trusting their physician to make that choice.
PD: The idea I proved early on is that there’s not a set number of a few influential people that determine the way everybody’s behavior goes. In this situation, there’s not some set of various prestigious researchers or physicians whom everybody just kind of follows. Instead, different people are influential in different ways at different times.
I think that recognition is that there’s not a subset of special people, like the kind of idea that people get from reading Malcolm Gladwell. I think that realization that you have to be flexible with different things at different times is one that I’m trying to always convey to people. I think it can be a hard message for a lot of people to hear, because they’re so used to thinking that we can get what we want if we just persuade a few important people, and that’s not really the way that the world works.
ML: If a billboard was given to you, anywhere in the country, and you could promote any type of message on it, what would it be?
BH: I would say: Don’t trust opinion, trust data. I’m thinking about patients when I say this. If you’re making decisions without data, I don’t know what you’re doing. I wish that patients would follow that advice more.
PD: To second what Brigham is saying: Just trust the data. That’s probably a reasonable motto for either one of us.
BH: Demand the data, maybe.
PD: That’s a good one. Demand the data.
ML: Are there any books that you not only like yourself, but find so valuable that you give them as gifts?
BH: I really like The Black Swan, by Nassim Taleb. That made a big difference in my thinking at the time I read it, which was related to my PhD. It opened my eyes to the idea that seemingly unpredictable things could be predictable. This was something I was reinforcing and seeing in my clinical research, which is that the outliers are very interesting, or maybe, the most interesting part of data.
PD: The book that I’ve passed out a lot is Nicholas Christakis’ and James Fowler’s book Connected: The Surprising Power of Our Social Networks and How They Shape Our Lives, which I think is a really excellent introduction to the ways that social networks frame and structure how social systems work.
Resources
You can find Brigham Hyde on Twitter @BrighamHyde or learn more about the work Decision Resources Group is doing with healthcare business intelligence on their blog.
