By Shraddha Chakradhar
May 1, 2012 | BOSTON—Eric Perakslis, recently installed as the chief information officer and chief scientist (informatics) at the U.S. Food and Drug Administration (FDA), gave delegates at the Bio-IT World Conference a snapshot of his new role at the agency.
Despite having worked with the FDA for several years with his previous employer, Johnson & Johnson, Perakslis admitted to being quite ignorant about the FDA’s influence beyond the drug industry. “The FDA controls 22-23 cents for every dollar spent in the United States,” he said, “and I had no idea it was that extensive.”
The biggest component of Perakslis’ job is data gathering and organization. Given the sheer volume of information received at the FDA, one of Perakslis’ main challenges is to establish solutions to consolidate, manage, and use data meaningfully.
When he visualized this goal—as he put it, “What does GREAT look like?”—there were several requisite qualities. For one, the technology had to be reliable, predictable, fast and available. “IT is about decreasing barriers to productivity, not increasing them or creating them,” said Perakslis.
In addition, Perakslis emphasized the need for something new. “You need to introduce 10-15 per cent new stuff every year,” he recommended. “That also means you’re flushing out, hopefully, 10-15 per cent that has become base cost in your infrastructure.”
He also said he would favor projects that have a five-year life cycle with no more than 18 months in development. “Break it down, get it done, make it simple, move on,” would be the Perakslis mantra. This mindset is particularly useful for services and component-based architecture because replacement, repair or starting over is relatively easy. And that, according to Perakslis, is a better model than a monolithic enterprise that breaks down completely when one part breaks down.
“Part of what I’m trying to do is take IT, process, and people out of a lot of processes that, in this day and age, can really be consumer-driven. If the Center for Devices, which has a different set of laws than the Center for Drugs, wants four or five components for their things, [I’d say to them] ‘take them and go.’ Make it consumer-based, lightweight and get out of it. Focus on something else,” said Perakslis.
Underscoring everything Perakslis—a kidney cancer survivor—is doing at FDA is the fact that, as he puts it, patients are waiting. “And I’m a data guy,” he added. “When I was diagnosed, I went home and looked up information. And what [that experience] has done for me professionally is that it has allowed me to be extremely value-focused. Patients are waiting. That’s what we [at the FDA] have to keep telling ourselves to stay focused.”
An area where Perakslis is close to a breakthrough is improving collaboration between the FDA and other agencies, especially in the area of consumer safety. When dealing with such a sensitive issue that puts a premium on collaboration and data exchange, it’s important to keep in mind what is considered safe.
“We’re building a model to enable direct connections, to enable data sharing, to have an objective set of rules to determine what data can be shared [and] what data needs to be kept private,” he said. For example, websites such as clinicaltrials.gov are low in data privacy (as the information is open and available to the public) but high in network security, enabling only a few people to control the information published. Perakslis said that most websites concerning consumer safety are somewhere in the moderate range. “We think of ourselves as custodians of data. We have all this data but a lot of it is not ours to share. We are not giving away confidential data,” he said.
One thing Perakslis has learned already from Washington, D.C. was that, “We all need to talk more, we need real-time feedback of where our file is, and vice versa. [Now], how do you set that up and how do you make that happen?”
Perakslis also offered an overview of import shipments of the FDA, growing at a fast rate of 13% per year. “Our primary goal: we really are supposed to be protecting the public. Stuff happens. How do you very quickly figure out it was the melon in Colorado [that was responsible for a food poisoning outbreak]?”
His team is working on using next-generation sequencing to figuring out that conundrum. The problem, at this point, is prioritization: when thousands of shipments arrive, for instance, finding an efficient way to inspect and monitor the products is critical—but tough. Finding solutions to that problem are additional problems on Perakslis’ plate, in addition to improving infrastructure for drug trials, liver toxicity, and vaccine safety.