Guest Commentary: Generating more knowledge from data and more value from studies
he ability to make meaningful correlations between preclinical observations and clinical outcomes can mean the difference between successfully moving a candidate drug through development and pipeline attrition. Getting an integrated perspective on the entire project—the studies, results, research citations, therapeutic and business potential—is a major challenge in translational medicine. Efforts to find relevant data, leverage information and generate new knowledge can be hampered by the exponential growth in the volume of data available.
Until recently, the role of the information scientist has largely focused on creating better, more efficient tools and ways to search, access, manage, manipulate and retrieve relevant data. These tools have evolved over time to become more intuitive and integrated, with expert systems that enable the end user to conduct more meaningful analysis, visualization and data mining.
But who is the end user? More often than not, it's the biologist, the chemist and other researchers on the project team. In principle, this makes sense. The problem is that data sources are growing in volume and complexity, with a diversity of formats that extends beyond the written word to include pictures, videos and podcasts. The diversity, complexity and volume of data also go beyond the technical expertise of the project scientist who rarely has the technical, computational or quantitative skills to extract the most meaning from the universe of available and relevant data.
The change that's needed is not just in software or systems, but in who drives those systems. Information scientists often have both the scientific training and the necessary computational skills to assist project teams with complex analysis and interpretation, yet even they often overlook how critical their function is in the excitement of using a new technology or developing a better tool. So, culture and convention work against them, and they are rarely considered for this more involved and expert user role.
Translational research, by its very nature, incorporates diverse data from multiple areas: disease information, target sequence, lead compounds, animal models, drug toxicity, medical records and drug metabolism, among many others. To further complicate the integration and standardization of data are the various types of data, parameters, formats, sources and standards—or lack of standards. The more sources there are to consider, the harder it is to make comparisons and evaluate content across all dimensions.
The tools of informatics have continued to improve, enabling better access, integration and mining of data. However, even the best tools fall short if drug project teams lack the involvement of those who know best how to use them, know how to interpret their content and know enough to understand the questions that need to be asked and answered.
This speaks to the cultural and skill-based challenges involved in getting more value from data. The nature of drug discovery and development remains an information-rich endeavor, and in order to increase success rates, we must make better use of our data—and our information science experts.
Consider the composition of drug development project teams. They likely include biologists who understand the disease area and work with assays; medicinal chemists who design and synthesize compounds; pharmacokineticists to study the bioavailability of a compound and its effect on physiology; toxicologists to study the safety or any toxicity issues; and physicians to design and lead clinical trials. What about information scientists? They're the experts in the tools, systems and the content (the data and information in those systems) to turn data and information into knowledge. They're the ones who can tap specific information and technical expertise as needed. And they're the ones who are often left off teams, except as ad hoc members, brought in to answer specific predefined questions instead of being problem-solvers, expert users of data and information and enablers of best practice.
A cultural shift needs to occur: one where information scientists are brought on board as full members of project teams, where they can be the expert users of data and help get the most out of available information, trying new tools and approaches as appropriate. They can be a unifying influence in the fragmented environment of multiple scientific disciplines, as they work across silos and clear bottlenecks in information flow.
The role of the information scientist can be likened to an advisor from a data perspective, much like the role of a financial advisor at a brokerage firm. Financial advisors listen to your goals, find out what you're trying to achieve with your investments, pull together a number of options for you to consider and then execute your agreed-upon plan. While it's possible to plan your financial future without the help of a financial advisor, having one adds the expert knowledge of someone who knows how to find and assess appropriate alternatives. Similarly, the information scientist can be a "one-stop shop" for the project team, looking at the totality of information, helping with interpretation and suggesting different research questions to pursue or even identifying questions or sub-questions that existing data can answer.
Information scientists play an evolving role in project teams. Once purely a data resource, with a narrow focus on technology, now they add more value by tapping their broad understanding of the science and of business needs. Instead of asking, "what question do you want me to answer?" the information scientists needs a different approach: "Tell me the problem you're looking to solve, so together we can formulate the most meaningful question and I can help you use the information available to answer it."
Two real-life examples come to mind that demonstrate the value of having this kind of information expert on project teams. In one case, the expert information scientist was able to circumvent the running of assays by recognizing that the answer to a question already existed. The team members had been looking at slices of the picture and didn't see the totality of inputs or that the answer to one of their questions was already available in existing data if analyzed differently, across studies. Having the answer to one question meant that a six-month study did not need to be run or repeated; instead, a different one could be run to test the hypothesis and glean new knowledge.
A second example illustrates the value of being connected within an information network and contributing to project memory. By understanding the competitive landscape and the value of learning from others' experience, the information scientist discovered a conference that would be addressing research relevant to a specific project. The conference would be held in another part of the world, without any attendance by project team members, so the information scientist worked to gain access to the proceedings and important new knowledge.
For information scientists to be fully accepted into project teams, they need to sharpen their quantitative and analytical skills to handle the complex analysis and interpretation of data—and to trust, or refresh, their scientific training. They need to be more actively engaged as core members, using their scientific and technical knowledge to determine what information can best answer a particular therapeutic or business question. Their involvement should lead the team to a greater level of confidence in making decisions, and the ability to make them sooner, in crucial aspects of drug projects.
The value of expert information skills is having both the scientific understanding to help define the questions and the technical know-how to bring all the relevant data together and use the most appropriate tools and analysis methods. Just as important as finding the sought-after information is playing a key role in interpreting the data and finding relevancy and meaning. Information scientists should not just be shepherds of data, but skilled users of data and technology systems that can help transform data into information and information into knowledge.
It takes more than expert systems to get better answers; it takes systems experts and data experts to extract meaning and generate more value from studies and assays.
Dr. Anastasia Christianson is senior director of R&D Information at AstraZeneca in Wilmington, Del. She is responsible for delivering the information needs of the Central Nervous System & Pain Innovative Medicines unit and Global Product teams, Personalized Healthcare and Biomarkers and R&D Strategy, Portfolio and Performance. Anastasia obtained her Ph.D. in Biological Chemistry from the University of Pennsylvania in 1989, followed by postdoctoral training at Harvard University in Cellular and Developmental Biology.