Mining the Bibliome - Translational Informatics: Realizing the Promise of Knowledge-Driven Healthcare - page 86

identify that there are gaps in current knowledge that prevent the generation of new
hypotheses.
The potential limited short-term utility of bibliome mining is also its greatest
asset in the context of a learning healthcare system. The point of a learning health-
care system is not that it is all knowing, but instead that it is capable of learning .
Thus, the ability to identify gaps in current knowledge or evaluate new knowledge
based on previously accepted reference standards (that may have evolved from
accepted gold standards) is the biggest potential contribution of bibliome mining.
Similarly, the process of bibliome mining within a particular context (e.g . , what is
the current wisdom about type 2 diabetes mellitus?) is an essential aspect of level-
setting the present status of the healthcare system. The next major task of a bibliome
mining process is then to identify potential opportunities (e.g . , by addressing the
question: “what are common versus uncommon medications that have been
described in the context of clinical trials associated with the top three common
comorbidities associated with type 2 diabetes mellitus?”).
Within learning healthcare system, bibliome mining plays an essential support-
ing role. The raw data that are generated (e.g . , standardized clinical outcomes) are
the primary source of evaluation. Bibliome mining would allow one to develop
benchmarks to identify how innovations that are implemented within a healthcare
environment are changing (either positively or negatively) versus the existing state.
However, the most signifi cant limitation of bibliome mining is that it is not, in and
of itself, a process to identify primary data. Primary level evaluations need to be
done within the learning healthcare system constraints. Nonetheless, bibliome min-
ing offers an ability to provide general understanding of the implications of applied
innovations whilst offering a sobering “reality check” of how primary data are being
interpreted - either by those within the specifi c learning healthcare environment
(e.g . , health care delivery researchers) or by those in a larger context (e.g . ,
epidemiologists).

5.3.2 Leveraging Data as Actionable Knowledge

Within any biomedical context that involves mining for new knowledge, the most
sought after type of knowledge is that which is “actionable.” Referencing the earlier
mentioned Baconian method, a major goal of bibliome mining is the reduction of
data interpretations into what can be distilled to defi nable hypotheses that can be
subsequently subjected to the standard Scientifi c Method. This is the essence of
actionable knowledge - that which provides the underpinning for some action that
can be tested and evaluated using accepted methodologies. With the increased fl ow
of data from across the biomedical spectrum, the potential to identify coalesced data
that can be used to form the basis of testable hypotheses is unprecedented. However,
the great volume, variety, and velocity of the data being generated pose signifi cant
challenges in ascertaining the potential value towards identifying knowledge. Of
particular note is the challenge of data quality. Just because the volume of data may