Bryn Roberts calls for more data, more sharing, and very cool tools.
Decision Support in High Definition
By Allison Proffitt
June 10, 2011 | A high-definition touch screen that fosters collaboration by enabling drug developers to manipulate molecules, much like Tom Cruise sifting through a crime scene in Steven Spielberg’s Minority Report, was the high point of Bryn Roberts’ keynote on novel decision support technologies at the 2011 Bio-IT World Conference.
The informatician’s role is to help scientists avoid reductionist thinking, said Roberts, global head of informatics at F. Hoffmann-La Roche in Basel. “The reason we’re here doing what we do is to enable decision-making,” he said. “We can do our very best to make this objective, to make this rational, but when a human comes to make a decision it’s clouded, it’s prejudiced by a number of things: their previous experiences, the politics of the organization, and short term objectives.”
Illustrating the data glut that R&D teams need to process, Roberts noted that if each of the 20 million citations in PubMed was printed on a sheet of paper, the compendium would stack five times higher than the Empire State Building. “There is a temptation for scientists, particularly who come from a traditional training, to be reductionist—to try to reduce the complexity of the information and the data they have on hand so that they can make decisions.”
At Roche, Roberts said, “We want to build an environment where our scientists and clinicians have integrated access. They can efficiently work with the data and information on hand, and ultimately they can derive knowledge and predict outcomes... At the end of the day, if they can predict outcomes, they can make good decisions.”
Among several Roche tools aimed at helping scientists get into the data, the Roche Cancer Genomics Database, developed with PointCross (see, “PointCross Takes Aim at Life Science,” Bio•IT World, May 2008), brings together data from lab information management systems and historical data “using a sort of semantic integration layer—orchestration layer—that [PointCross has] built to bring together the content within its context, to normalized things like conditions, calculations, and so on,” Roberts explained.
Precompetitive Plea
Quoting Henry Ford, Roberts challenged informaticians to go beyond what people are asking for to see what they really need. For instance, “We spend the majority of our time and effort on efficiency-based informatics solutions—electronic lab notebooks, LIMS, software management systems, instrument support, data management, and what have you,” Roberts observed. “Why don’t we collaborate more on some of these major, costly precompetitive platforms and spend more of our resources on the real value added decision-enabling tools?”
The most eye-catching decision-enabling tool Roberts presented was a huge Minority Report-style multi-touch screen. Using the screen, Roche chemists and researchers can work together with a virtual card stack of chemical structures—manipulating, annotating, and assigning structures to members of the team. The team approach dilutes some of the prejudices and issues that arise from individual decision-making, Roberts said. Moreover, the multi-touch screen was undeniably cool (see p. 33).
Roberts summed up the researcher’s wish list: integrated data, seamless flows, support for semantic-based question formats, intelligent interfaces, and the ability to explore data and collaborate as a team both remotely and locally. “When I really think about it,” he said, “it’s all about decision support. If we’re going to be successful as an industry, we should be thinking about the strongest support decisions.” •