Sept. 18, 2006 | Spotfire’s DecisionSite is popular in many industries: energy, semiconductors, financial services, and soft drinks, to name a few. But the heart of the company’s user base remains in the life sciences, says Christian Marcazzo, Spotfire’s senior director, life sciences marketing. The company is based in Somerville, Mass., but its programmers are in Sweden.
“We become the first place that people try to analyze new data,” Marcazzo says. “They don’t know what they’re looking for. We make it easy to bring that information in and explore those patterns and those trends.”
|Spotfire's DecisionSite system offers a new window on |
It’s a misconception to think Spotfire users only work in drug discovery. Clinical teams are also using Spotfire, Marcazzo says, in big pharmas, biotechs, and one or two contract research organizations. “Our customers are concerned about attrition rates in clinical trials, particularly Phase II and Phase III trials. They would like to be better at predicting which trials are going to fail. It’s that late-stage attrition that is driving a more rigorous analysis of what’s happening in Phase I.”
There is not so much a rivalry between Microsoft’s Excel spreadsheet and DecisionSite, he says, as an appreciation that almost every customer has Excel and takes a stab at using the spreadsheet for something it wasn’t designed for. Says Marcazzo: “They put it in Excel, they see something, and then they cut and paste it into Spotfire to analyze it.”
In a demo, Marcazzo put Spotfire through its paces with dummy clinical data. As he works, there is a wonderful quality of serious play — of purposeful fooling around with data to learn what it means. “I am bringing data from a Phase I trial,” he begins. At first, just a standard bar chart is on the screen. But then he clicks on a tabular column of lab values, revealing whether the patients were on drug or placebo. “That’s a sanity check,” he says. “Are we seeing anything unusual?”
Spotfire customers typically customize the application in many ways, but out of the box Spotfire can flag medically “interesting” patients or batches of records. His demo gets especially impressive when Marcazzo wants to work with subsets of patients. “I want to see the patient who failed six or more tests,” says Marcazzo, typing a few keystrokes and selecting the patients with a cursor. Each time he selects a new subset of patients, the changes ripple through several graphs, plots, and windows displaying all or part of the data.
“Maybe I want the histogram,” Marcazzo muses, “but I take out the baseline and end-of-study data. Based on that, I can see how the distributions change.” As he clicks further, he finds more outliers. Up pops one patient’s chest X-ray. This does not mean DecisionSite needs to store thousands of images. “We don’t need to have the data in our data set,” says Marcazzo. “We just need an identifier and a URL to go off and fetch that on demand.”
Data from SAS, ArisG, Phase Forward, or Oracle Clinical can be loaded into DecisionSite as well, but that may require a customization project. Spotfire has a services arm and can help customers use the application to support specialized needs and integration projects. Says Marcazzo: “We see where their data is. We see who their [human] analysts are and what they need to do. And we map the two together.” The process might take two months.
In some cases, Spotfire can unite two distinct streams of data — perhaps adverse events data and clinical data management system data. If the systems to report on each type of data are already well developed and understood by the customer, Marcazzo says, “It becomes a relatively simple exercise to bring the two together.”
In more standard cases, “Guide” macros allow a customer’s in-house subject-matter experts to record each task in a complex series of data-analysis or visualization steps. When a less-sophisticated colleague wants to try the same technique, it’s as simple as hitting a button.
For some customers, Marcazzo says, visualization is a collaborative as well as a visual experience. One life science customer has a 12-foot-wide screen that allows a cross-disciplinary team to convene in a conference room and literally stand up and walk over to a particular cluster on a scatter plot and ask the person “driving” the program to investigate further.
“They’ll say, ‘Here are some patients that have weird things happening.’ And a physician can say, ‘yes but....’ The whole point is about creating an interactive experience where there is a lot of domain expertise in a variety of end users.”
Email Mark Uehling at: firstname.lastname@example.org.