YouTube Facebook LinkedIn Google+ Twitter Xinginstagram rss  

The Architect of Data Visualization

Visiual i|o makes decision-making a question for the eye.

Sept. 5, 2008 | So what is a nice architectural graduate like Angela Shen-Hsieh doing running a data visualization software company targeting big pharma? We’ll let her explain that shortly. Shen-Hsieh founded Visual i|o (the i|o originally stood for input/output) back in 2002 with CTO Mark Schindler as a spin-off from consulting firm Schindler and Associates, in order to focus specifically on data visualization for decision support. She’s been working with pharma since 1998, mostly on the drug development pipeline (though with some sales and marketing, and manufacturing). Visual i|o is based in Newton, Mass. Here, Shen-Hsieh tells Kevin Davies about the attributes of data visualization that are helping big pharma portfolio management.

Bio•IT World: Angela, how did an architect get into data visualization?

Shen-Hsieh: Architects don’t actually build buildings—we make pictures of buildings. We’re experts in graphical representation. There’s a whole language in architecture—plans, sections, elevations, perspective—in order to understand the condition of a building. That’s the same approach we take to data. A representational language, a graphical language, around looking at different kinds of data, to essentially put the story into the disparate pieces that have gone into a database.

Architects also have a kind of left brain/right brain thing, where you take a softer approach to things like how someone’s going through an experience, or a need to understand something, and they have to translate it into a technical medium. That’s very similar to what we do in data visualization and software. What are people trying to do? What is their experience? How does that translate into the bits and bytes of software?

The more practical answer is, I came out of Harvard architect school in 1991, at the bottom of the economy. I looked around and said, maybe I should get some other skills. I found that I liked designing businesses, particularly around visual communication, that’s always been my interest.

What kinds of data does your software handle for pharma clients?

The space in which we live is really the business intelligence space that sits on top of all of the data that goes into portfolio and investment decision making. That ranges from clinical data to project planning data to research data, budgeting, risk—that’s where the science comes in—marketing data, potential value, and how those all need to be looked at in order to make better investment decisions.

Are you just targeting big pharma companies?

Our focus right now is mainly the bigger pharma companies, but our technology is very horizontal. It can go across the R&D process and the information life cycle looking at clinical trials, portfolio assets. Regulatory submissions, submission management, safety, sales and marketing. There’s a lot of different areas that can use visualization. We don’t really focus too much on the visualization of scientific data. We focus on the decision-making around the business of R&D.

Who within the pharma organizations is actually using your software?

We have five of the top 12 pharma companies…  It’s most of the functions that report up to senior decision makers. Those functions range from finance, resource management, project management, project planning, trial managers, project managers, and probably one of the biggest areas is loosely called business support—a wide swath of people who feed information up into the decision makers.

How is the software deployed within those organizations?

The software is all web based. It essentially leverages existing data aggregation. We usually sit on top of a data warehouse or business intelligence platform. We don’t create another platform, and that’s a strategic decision on our part—nobody wants another platform. Visual i|o provides a light weight layer that can act as a lens so that, across an organization or a dataset, you have common ways of looking at the data. So everyone is looking at the data in the same way.

[People] have their own perspectives, so if I’m looking at resource utilization, and I’m responsible for all of chemistry, I’ve got a particular interest in where the project plans are different from the interests of a project planner. But it’s becoming more and more important that people across the organization see the data in the same way. It can be a tremendous tool for strategic alignment.

What are some of your favorite applications of the software?

Our latest product—DecisionIris—is currently used in three of the big pharmas for project portfolio analysis. The big ROI of this product is around the speed of decision making. In big pharma, it’s so difficult to get all that data around the entire portfolio into a form where your questions are readily answered and you know what to do. The tools at hand make it hard and slow to generate all the analysis you’d want to be confident about decisions, so a lot of companies don’t. If you think about the total investment in R&D, you’re making adjustments infrequently and so not “replacing winners with losers” and optimizing the use of precious funds and resources.

Then there’s the more tactical ROI: automation of the tedious and manual process of producing these reports, and then responding to the ad-hoc requests that inevitably follow. A lot of effort that goes into making analysis, then asking the next logical question, that going back into a whole reporting process that could take days or weeks to complete. A customer of ours says their senior people have essentially trained themselves not to ask too many questions, because a question could send people off for weeks and it may not be worth the time in the end.

Another of our customers produces a report for review twice a year. One person alone in this process spends about 100 hours manually pulling things together, creating all the charts and reports. And it can be a hit or miss process. It took a while for them to figure out what views resonated with people at all. They say, ‘We’re not looking at the whole picture, because it’s so hard to get any picture!’

So we need some efficiencies here, and interactive visual analysis allows decision makers to move along a line of thought, to be able to optimize what the human brain does well—pattern matching, like recognizing human faces—with what the computer does well—perfect memory... Most big companies are quite separate, they’re separate systems. What everyone’s trying to enable is faster decision making—and that’s what our products help with.

Is the upside with your software captured in the phrase, “kill early, kill often”?

“Kill early, kill often” is [one type of] project level decision making. There’s a theory around this, we’re not necessarily tied to any particular portfolio theory. We’ve worked with quite a few…  So on the one hand, there’s “kill early, kill often,” which is crucial—you can significantly increase the throughput against limited resources. Putting more winners in, pulling losers out faster—we help enable that, with the ability to review a portfolio any time you want.

The other piece of it is, you look at project level decision making, that’s where most pharmas start with portfolio management… We’re going to invest in this next phase… then usually that gets thrown over into operations you know how to staff it, it may or may not show up in context of all the other investments.

For instance, our DecisionIris product enables monitoring of how well we’re aligned on whatever our strategic goals are. We’re pulling out in this therapeutic area; we’re investing more in this area; we have a consistent pipeline here, bottlenecks here, gaps there; we’re aligned around strategies, platforms, technologies, partnerships.

There’re a lot of factors that go into portfolio balancing. That’s something visualization is very good for—the ability to very quickly slice and dice around different dimensions and see multiple dimensions at the same time. You get a much richer picture if you’re investing for the goals you thought you were.

Is your product unique or are you competing head-to-head with others?

We don’t often compete head-to-head. There’s a lot of different ways to slice and dice this problem. There are desktop tools to create visualizations, and there are tools for IT departments to generate dashboards.

At Visual i|o, we have a different approach—we’re not a toolkit, we’re specifically configuring our product, a flexible rendering engine, with our domain expertise around visualization, to expose and explore different questions. We’re intended as an enterprise layer, not simply a desktop tool, and this enterprise layer helps enable faster decision making for problems like project and portfolio management.


 Business Intelligence Mash-Up

It was almost inevitable that Visual i|o CEO Angela Shen-Hsieh would have trouble booting up the projector to show some examples of her company’s latest data visualization tool, DecisionIris. But when the image on the whitewashed wall of her conference room finally falls into focus, it’s easy to see how it might appeal to a harried pharma executive.

The June 2008 launch of DecisionIris by Visual i|o provides pharma companies with a new enterprise solution to collect disparate corporate R&D data, including strategic portfolio management, cross-project resource management, and project tracking and analysis.  The visual representations themselves are not overly elaborate—simple colored two-dimensional bar graphs, Gantt charts, and so on—but their simplicity facilitates interpretation. 

The resulting mash-up presents a visually striking overview of a pipeline that might reveal glaring holes in a company’s future product portfolio, or productivity issues in a particular therapeutic area, or high attrition rates in a phase of clinical development. For example, an executive can review the company’s entire drug pipeline broken out by therapeutic area, with an interactive query facility that provides a wealth of customization options, including resource utilization, speed, throughput, and probability of success. Another neat feature is the ability to project the state of the pipeline months or years into the future. 

The same level of analysis could be applied to supply chain, drug supply, or sales force analytics, depicting customers, fields, sales reps, region, type of deal, and so on.


This article appeared in Bio-IT World Magazine.

Subscriptions are free for qualifying individuals.  Apply Today.




Click here to login and leave a comment.  


Add Comment

Text Only 2000 character limit

Page 1 of 1

For reprints and/or copyright permission, please contact Angela Parsons, 781.972.5467.