Bryn Roberts on Data Citizenship, FAIR Culture, Data Models for the Long Term
By Allison Proffitt
May 23, 2023 | Bryn Roberts has spent thirty years in informatics and data, most recently at Roche and before that at AstraZeneca. Now as Global Head of Data and Analytics at Roche Information Solutions, Roberts applies his career-long learnings to the current era of data in life sciences.
There’s a sea change in pharma right now, Roberts told Stan Gloss, host of Bio-IT World’s Trends from the Trenches podcast. Across the industry, everybody is recognizing data’s value for healthcare and life sciences. “They’re sitting on very real assets,” Roberts said. “Often those assets don’t yield any value until they are combined with other people’s data or put into a context or workflow or activated by analytics or other things.”
At Roche, Roberts said, the efforts to bring data together to curate them and make them analytics-ready has been going on for years. “We initiated a FAIR culture within Roche: data FAIR-ification, making data findable, accessible, interoperable, and reusable,” he said. “It was a recognition of the clinical trial subjects—many of them patients—and physicians and others involved in creating those data—it was a sense that these data are not realizing their full value despite the investment of people’s lives that have gone into their generation.”
Swearing in Data Citizens
Part of the changes required were changes in data culture—how team members thought of the data they produced and consumed. There was certainly a tension there, Roberts concedes. But Roche introduced the concept of a Data Citizen, which, Roberts said, aligns with country citizenship. A country’s citizens have certain rights and freedoms that are theirs for abiding by the laws and responsibilities of the country, he explained. There are, “responsibilities as well as the privileges of being able to use data that have been generated by other parts of the organization at another point in time and for a different purpose.”
“It was a bringing together of the community to agree: ‘As a producer of data, I will make my data available in the agreed depositories or formats to the agreed standards with a certain amount of metadata and annotation that enabled those data to be treated FAIR.’ And then as a consumer, ‘I will take the responsibility to—where appropriate—consult, to inform, and to involve others who have been involved in the data.’” It’s a notion that is still evolving, Roberts said.
Data Strategy in Hindsight
In fact, data strategies must always evolve, Roberts said. Gloss asked for advice on how to get data right “from the beginning” and Roberts advised, instead, to recognize that whatever you do now will be outdated soon. “You have to accept a certain degree of pragmatism,” he said.
But he still recommended as much care as possible when architecting your data. Think about the data models you plan to use and what is needed to make those models complete, he said. Consider what sort of curation and annotation is needed when the data are produced so they are reusable in the future. Create a long-term data management plan, including when and how to dispose of data either because the cost of storage is too high or the outdated instrumentation that generated the data makes them no longer valuable.
Roberts warned against proprietary data lock-in: “becoming very dependent upon a single technology or vendor for the fundamental modeling, structuring, and storing your data.” Instead, he recommended using open data standards with vendors adding additional value, but not limiting future options, he said. “There are so many great open data standards, data models, and tools out there. I think there’s no excuse for accidentally getting locked in.”