The eRA Project Team recently explored the potential role of Knowledge Management (KM) in maximizing the utility of eRA data for promoting health research and accelerating discoveries. KM is one of the key components in resetting the eRA vision, the focus of the Third Annual eRA Retreat on October 9–10. (See full article in this issue). 

Dr. Richard Morris, eRA advocate for knowledge discovery, gave a presentation on KM concepts at the retreat. KM refers to the family of text-mining tools for examining vast quantities of data to identify patterns and establish relationships. Given the exponential rate at which the world’s information is growing (estimated at 10¹⁸  bytes yearly), KM has great potential for optimizing the knowledge assets of an organization and saving thousands of labor hours.

Among the promising benefits of KM are:

• NIH-wide level of investment analysis.        
• Timely integration of emerging scientific trends into program formulation.        
• Better access to clinical data for interpretation of research and assessment of discoveries.        
• Accurate and efficient referral of incoming proposals for review.

eRA already has demonstrated proof of concept that KM can successfully identify qualified reviewers for an incoming application. Dr. Bob Lewis of Mitretek Systems described a recent pilot during which eRA “fingerprinted” (profiled) each incoming research proposal using the National Library of Medicine Medical Subject Headings (MeSH^®) Thesaurus. MeSH is a database of hierarchical (e.g., “ankle” is subordinate to “anatomy”) and cross-referenced topics (e.g., “for vitamin C, see ascorbic acid”), which permits searching at various levels of specificity. The fingerprint for each research plan comprised the most appropriate MeSH terms.

The other sources of input for the pilot were databases (internal and external) of reviewer biosketches. Once the biosketches were fingerprinted, locating subject-matter experts consisted of comparing proposal profiles with expert profiles to produce the best matches. Thus, KM enabled NIH to transform raw text (research plans and biosketches) into useful information.

Dr. Arthur Petrosian, a scientific review administrator at the Center for Scientific Review (CSR), presented the Computerized Reviewer Assignment and Search Program (CRASP^®) that he and his son developed. They designed the program to locate reviewers for specific ad hoc diagnostic imaging study sections. CRASP searches for reviewers based on keywords in CRISP and PubMed. In addition to expertise, CRASP considers other factors such as diversity, gender, geographical balance, and potential conflicts of interest. 

eRA also can use KM techniques to ascertain similar research plans (from proposed and already funded research) and to create opportunities for scientific collaboration by identifying experts with common interests. These and other KM applications are expected to improve NIH performance and efficiency by facilitating data sharing, verifying facts, informing decisions, shortening grant cycle times, and reducing costs. NIH also hopes to realize qualitative gains.

Implementation of KM is not included in the current eRA budget. Dr. McGowan will make a presentation to NIH Director Dr. Elias Zerhouni to establish a business case for purchasing a commercial KM product for NIH-wide use.