What is the Agency Framework?
Semantic modeling and search techniques offer simple tools for
knowledge management that can give SIGs, small businesses, enterprise
departments or distributed groups powerful new ways to manage incoming
information:
- Agency treats data
changes as NEWS which may
trigger ALERTS
- It also aggregates inputs to support later searches over
their content
- These powers make a fine back-end option for our NLP products
The framework gives to
each subscribing group a
small suite of web services configured to
track data sources, learn from them, and report significant changes to
selected users and/or to the software they designate.
The Agency inherits expertise
from its UMLS
domain vocabulary, so its initial corpora will support Life Science
topics and applications
1. PubMed Abstracts in an Area of
Interest
Our software can track
abstracts on almost any interesting mix of topics that NCBI
can supply with its normal query services. PubMed holds millions,
and it grows too fast for most potential users to keep up. Agency
lets them.
It semantically analyzes each new abstract more deeply, typically
overnight, matches it to similar abstracts in the local corpus, and
alerts users under rules they can improve as their own experience grows
along with their corpus. Overall, Agency users learn more in
less time by "contextually" tracking PubMed.
2. Clinical Situations in an Area
of Practice
Electronic models of patient visits or clinical histories, actual or
prototypical, can be analyzed, tracked and managed in very similar
ways. Real
patient data could trigger alerts of many kinds, including standards of
care to consider, PubMed abstracts on relevant research,
or matching clinical trials.
Health Care gets more perceptive
when clinical-situation events
trigger treatment hints or warnings; start up work flow on decision
support or billing tools; help research prior outcomes;
etc. Agency APIs let
helpful alerts specific to each user be triggered by almost any input of
controlled medical vocabulary.
3. Vocabulary Models that Integrate
Systems
Interoperation of medical software systems is today often impeded by
their use of competing internal concepts. By registering their
definitions under the Agency's more
general vocabulary, developers may quickly find the translations that
help circumvent the barrier, or even create real-time, auto-aligning
APIs.
Clinical data systems can enable such interfaces if relevant usage
specs on their access get indexed in Agency,
then queried using caller specs on needed data This
approach, for example, could potentially let many new decision support
systems use many disparate clinical data sources. A major
practical benefit of it is that better integration
takes only better specs, not costly recoding efforts.
|
