Here at MM&M, we try not to veer into the realm of overstatement. That said, given recent conversations with HR execs and leaders of companies big and small, we feel comfortable saying there is not a single pharma company, payer, device maker, or health-first agency that has enough data scientists on staff to begin to sate the demand for such services.

At the same time, those organizations continue to grapple with the taxonomy of the data scientists they have on hand. Should they be siloed with the tech folks or the traditional scientists? And what of the data scientists themselves? What educational and professional backgrounds are useful? How should they divvy up their time among various internal and external constituencies?

Four A-list data scientists provide some answers. It’s time to get to know your data scientist.

David Rosen
Executive director, forecasting, analytics, and market research
Purdue Pharma

Tell us about your professional history. While in graduate school, I did an internship in managed care marketing for Becton Dickinson and another in benefits consulting at Hewitt. I then went to Bristol-Myers Squibb and started in the management development program through the IT department. When the program dissolved, I went into management consulting for Easton Associates, which was swallowed up by Navigant.

That was when I began new product evaluations for pharma, diagnostics, and devices. I joined Purdue in Q4 2003 and have been here since in a variety of expanding roles.

What was your “eureka” moment? The first was when I worked for Easton Associates and realized I was excited about evaluating new products. There was a chance to find innovative patient benefits and real clinical differences, and to learn about each new disease.

The other moment happened when I came to Purdue and started teaching myself technical forecasting. It hit me how much I enjoyed looking at and working with data.

In the broadest possible sense, what do you do? I head all the commercial science functions at the company. That includes market research, forecasting, competitive intelligence, analytics, targeting and segmentation, field force alignments, and incentive compensation.

There are lots of different responsibilities and they’re all tangentially related.

Describe a typical day. There are no similarities from one day to the next. One of the things I love about my job is there’s so much opportunity to immerse myself in different things.

One day it’s a product forecast in a new market, another it’s generating insights in company budgets and forecasts for the next five years.

What’s something people don’t know or understand about the data scientist role? I interact with every department — I talk to finance about planning models and top management about market overviews.

Also, data scientists aren’t knee-deep in data all the time. I am not a trained data scientist. I haven’t used some of the traditional tools such as SAS since grad school. Some data scientists help interpret data. That’s me — I’m a sounding board.

What’s something people don’t appreciate about the role? By no means am I always declaring, “Well, the numbers say this.” It’s more of, “The numbers say this, but my recommendation is to do something different.”

See also: Get real: The moment has finally arrived for real-world evidence

How will the role be different a few years from now? When I first started, our analytics department was two people mostly focused on forecasting. As the tools got more sophisticated and the demands became more intense, our need for a higher level of technical skill grew. As tech improves and our therapeutic areas evolve, we’re going to need different types of skill sets.

Who or what has been your biggest inspiration, professionally or otherwise? I had a couple of teachers in high school — a history teacher and a math teacher — who took so much pride in what they did. I find that inspiring.

Yi-Lin Chiu
Head of discovery and early pipeline statistics

Tell us about your professional history. Following my graduation from Northwestern University in 1997, I joined the pharma research team at AbbVie. My role allows me to put my education, background, and passion into practice and improve decision-making in medicine.

What was your “eureka” moment? About two years ago, I realized how intuitive Bayes’ Theorem is. It’s the probability of an event occurring based on prior knowledge of conditions that might be related.

When it comes to Bayes’ Theorem, some are enthusiasts, but many are skeptics, because it can appear complicated. The Bayesian school did not academically train me, but after my discovery, I have appreciation for the approach.

In the broadest possible sense, what do you do? I work with data to help scientists make the best decisions possible for patients. We begin by designing a study to collect the data from which we then extract the relevant information, make a decision, and draw conclusions. We always have the end in mind — is this a potential treatment that could improve patient outcomes?

Describe a typical day. I am responsible for a team of 35 statisticians. Like any good statistician, I analyze data for all administrative tasks to improve my team’s efficiency.

I work with my group and other cross-functional teams to come up with plans for data collection, study design, statistical methodology, and data presentation on various projects. I work hard to foster a culture of collaboration, and we get together often to discuss different aspects of current projects. The most meaningful scientific advances are made in environments where people can be bold in their solutions.

What’s something people don’t know or understand about the data scientist role? People think we only deal with numbers, but that is not the case. We deal with people all day long.

Every day, we work to strike a balance between being accurate and making sure people understand the implications of our findings.

How will the role be different a few years from now? The future will be much more dynamic. Automation will be widely implemented for many common statistical analyses and more people or AI will perform these analyses on their own.

A traditional statistician must shift from being a routine data analyst to a disciplined, idea-generating powerhouse in order to thrive in the future.

Who or what has been your biggest inspiration, professionally or otherwise? Marie Curie. Her character, passion for science, hard work, and the impact she had to shape our world always inspire me — especially when you consider how much more challenging things were for women in science back then.

Alfred Whitehead
SVP, data science, Klick Labs
Klick Health

Tell us about your professional history. I earned my bachelor’s in physics at the University of Alberta and joined a startup called Klick in 2004, where I built the software testing, IT, and compliance practices. While doing that, I got a master’s in astronomy at James Cook University. I took a break in 2008 to get a master’s in physics at Drexel University and most of a Ph.D., using high-performance computing to simulate the lives of star clusters.

Then, I returned to Klick and managed every discipline in our tech department over a series of roles, picking up a CISSP certification along the way. In 2015, we established our data science team, and the two streams of my career finally merged into a job that combines both my industry and academic experience.

What was your “eureka” moment? I haven’t had one — and I don’t believe in them. I’ve seen quite a bit of science and it never happens in a flash of brilliance. It’s a process of trying out ideas, trying to disprove them, and keeping the very few that remain.

In the broadest possible sense, what do you do? At Klick Labs, I focus on innovation in data science. My job is to build a team that can take the latest advances in data science, machine learning, and AI, then figure out how they can improve the lives of patients, physicians, and others involved in the healthcare system.

We do applied science — we try out the lovely models and techniques coming out of the more theoretical end of the scientific community and see if we can adapt them to practical problems that exist right now.

Describe a typical day. It starts with a meeting with my team. Each data scientist is working on one to three projects at any given time — a mix of healthcare experiments, marketing experiments, and direct client work. I usually end up sitting down with at least one member of the team and help them with something they’re stuck with. It could be a technical detail — but given that every one of them is already an expert, it’s usually a challenge in understanding the business side of the equation: what to do next, who to talk to, or how to navigate an organizational maze.

I always try to make some time for reading, too — I digest five or six published academic papers each week. If I’m lucky, I find time to hammer out some code on an experiment of my own.

What’s something people don’t know or understand about the data scientist role? Most of the effort in any project goes into getting the data together in a clean form that can be processed. We refer to it as data engineering, and it’s the secret sauce that makes everything work. It usually involves negotiating with multiple data holders, syncing up many files and databases, matching up records, handling all of the weird edge cases, and swearing a whole lot about date formats and time zones.

What’s something people don’t appreciate about the role? Data scientists are only as good as the data they have. Many people seem to think we can magically get insights out of next to nothing. What we actually do is refine and distill what’s already there.

We can turn wine into brandy, but if we want to turn water into wine, we need a vineyard, a hydrological cycle, some sunlight, a lot of hard work, and a few years in the cellar.

How will the role be different a few years from now? At some point, the hype is going to die down. AI tech is driving all of the hype and, while deep learning has started to produce some amazing results, it’s just one tool in the toolbox.

The role will become less “sexy” but probably more essential, as most businesses will have gotten hooked on the impact data-driven decision-making can have.

See also: MM&M’s Data Week: Demystifying one of pharma’s biggest pain points

What are your must-have work items? First and foremost, I need a whiteboard or blackboard. It’s probably my physics education, but I can barely think or speak without scribbling as I go.

Beyond that, I need a laptop with a Unix-y OS on it (MacBook Pro is my current), an internet connection, and some public cloud accounts to get more computing power when it’s needed.

Who or what has been your biggest inspiration, professionally or otherwise? My dad. He’s an engineer-turned-exec-turned-consultant who started out bringing computer systems into business for industrial control in the 1970s when the idea was brand new.

I’m happy to be carrying on that pioneering spirit today.

Patrick Richard
Managing director, data science
Syneos Health Communications

Tell us about your professional history. I started as a designer and UX pro in the online pharmacy space, and then moved to digital pharma marketing and design for a few years. In the middle of my career, I spent about six years at both online travel and retail organizations, where I focused on strategy and digital product development. I’ve been back in pharma marcomms for about seven years now, focusing on data-driven strategy, data science, and digital strategy.

What was your “eureka” moment? When it comes to communication in any market, everything is interconnected. It can begin with one person overhearing someone talk about a product that changed his or her life and then going home and researching it online, to sharing their own research with friends, to the socioeconomic impact of a decision potentially being made.

I realized you can recreate an environment or experience before a person sees it — more quickly, more smartly, and in a way they can understand — all through data. It used to be that something was created based on intuition plus some financial modeling, or by piecing together a number of inputs in a manual way.

In the broadest possible sense, what do you do? I’m not coding, but I’m leading our overarching framework along with our approach to the market using data. This includes thought leadership, strategic consulting with clients, and supporting our agencies.

Describe a typical day. There is no typical day. I’ve had days where I’m with clients in Los Angeles to discuss how they can better use data across multiple brands. I might be in New York for an internal brainstorm session with colleagues across our clinical and commercial divisions. Another week, I may be deep in a project that is just kicking off and spending most my time with internal teams.

What’s something people don’t know or understand about the data scientist role? The average non-data geek may not know that while obtaining the skills to run models with languages such as R and Python are incredibly important, the most value comes from being strategic and curious with the data. As a data scientist, if you understand how to look for key insights out of data queries and tell a story where action can be derived and brought to market, you are of top value.

If you understand the data science best practice from a fundamentals perspective — but lack business acumen or interest in driving results through some level of creativity — the value is much lower. You can learn the languages of data science, but the ability to take action off the results is what makes a difference at the end of the day.

What’s something people don’t appreciate about the role? Data science is not magic. It requires rigor, creativity, and continuous iteration with the models. There needs to be time built in for a data scientist to be creative, just as time is built in for creative pros. Although the tactics used to get there are different, there are similarities between the two jobs.

How will the role be different a few years from now? The data scientist will be asked to use more automated tools that clean and aggregate data much better than manual coding techniques. This will create a shift in the type of people who might aspire to be data scientists, because the field will draw more heavily on creativity and strategy.

These types of pros will need to be able to show specific actions resulting from a person experiencing something designed from a data model.

Imagine a data model that tested the concept of a virtual doctor built with AI that proves out the engagement metrics before the experience is designed. That’s really creative, but if you have some simulation of what might happen after that experience is built prior to spending the dollars to build it, that changes the game.

Who or what has been your biggest inspiration, professionally or otherwise? I’m inspired by anyone who is openly talking about doing things that challenge the status quo. It has to be effective, but I think we all conform at some level and it’s not usually the best way to innovate.

It calls to mind a Chicago-based company called Basecamp (formerly 37signals). The company has never conformed to the traditional way of working (long meetings or everyone in one location working in the same time frame). This keeps its people fresh and always evolving the various digital products they produce that provide real-world value.