IRC log of HCLS on 2012-03-27

Timestamps are in UTC.

15:11:23 [RRSAgent]
RRSAgent has joined #HCLS
15:11:23 [RRSAgent]
logging to
15:13:08 [ericP],
15:15:50 [mscottm]
mscottm has joined #hcls
15:15:54 [cme]
cme has joined #hcls
15:16:49 [ericP]
-> Cecil's slides
15:17:12 [Stuart]
Got them, thanks
15:17:41 [cme]
cme has joined #hcls
15:18:14 [iker]
thx ericP
15:18:18 [ericP]
ericP has changed the topic to: Cecil Lynch's SemWeb in Health Care slides:
15:20:32 [ericP]
scribenick: ericP
15:20:38 [ericP]
[slide 3]
15:20:56 [cme]
cme has joined #hcls
15:21:07 [Joanne_Luciano]
slides aren't numbered :-(
15:21:32 [Joanne_Luciano]
ah, but the browser numbers them!
15:22:00 [egombocz]
If you look at them not in show mode, you can see the numbers on the side thumbnails
15:22:00 [ericP]
Cecil: antibiotic-resistent airline passenger promted review on Tuberculosis Information Management System (TIMS)
15:22:44 [ericP]
... reporting a TB case required passing a brittle set of messaging and business rules
15:23:12 [ericP]
[slide 4: Message Processing Integration]
15:23:23 [ericP]
Joanne_Luciano: each state wanted their own standard?
15:23:33 [ericP]
Cecil: CDC wanted a standard
15:23:43 [ericP]
... states would take anything which makes reporting easier
15:23:52 [ericP]
Cecil: [re: slide 4]
15:24:21 [ericP]
... choices about how to import messages to CDC
15:24:30 [ericP]
... .. after message had some processing
15:24:38 [ericP]
... .. as a Web Service RPC
15:24:51 [ericP]
[slide 5: Deployment Architecture]
15:25:08 [Zakim]
15:25:09 [ericP]
Cecil: going with existing CDC infrastructure
15:25:22 [ericP]
... staring from left:
15:25:22 [michel]
michel has joined #hcls
15:25:47 [ericP]
... .. some source, usually state or large counties (53 jurisdictions) reports
15:25:57 [Joanne_Luciano]
is going with the CDC one of those three options on slide 4 or is it another one (not listed on slide 4)?
15:26:10 [ericP]
... .. goes into data messaging broker, which validates syntax
15:26:33 [Joanne_Luciano]
looks like it's option 1 on slide 4
15:26:36 [ericP]
... .. if a valid TB message, off to content validation queue
15:27:02 [ericP]
... .. also split into components for e.g. line listing of incoming cases
15:27:31 [ericP]
... .. after validation, email with contents of alert sent to CDC's TB group
15:27:47 [ericP]
Joanne_Luciano: this is slide 3 option 1?
15:28:09 [ericP]
Cecil: this is slide option 3 (RPC)
15:28:39 [ericP]
... we had tried driving real-time alerting from biosense
15:29:11 [ericP]
... we took messages off the first transport, never queued in DMB [slide 4 left]
15:29:25 [ericP]
s/[slide 4 left]/[slide 5 left]/
15:29:38 [ericP]
Cecil: the HL7 2.x standard is fairly loose
15:29:45 [ericP]
... flexible, can take any payload
15:29:51 [ericP]
... can be structured in any way
15:30:21 [ericP]
... segments are well-defined, but segment structure requires point to point negotiation
15:30:59 [ericP]
... p2p neg is a guideline
15:31:00 [ericP]
charlie: HL7 2.x is a syntactic standard and a semantics guideline
15:31:16 [ericP]
[slide 6: Message Content Validation Architecture]
15:31:32 [Joanne_Luciano]
15:31:38 [ericP]
Cecil: after leaving broker, falls into JMS interface
15:32:08 [ericP]
... because this has the 2.5 validation, we don't need the 2.x syntactic validation
15:32:21 [ericP]
... so we don't do the validation
15:32:39 [ericP]
... before we went live, we validated and found 2 errors in HL7 messaging
15:32:51 [ericP]
... (was a benefit of 2-tier validation)
15:33:03 [ericP]
... once live, we don't do syntacit validation
15:33:16 [ericP]
... but we do parse out components
15:33:50 [ericP]
... questions like birthday and date of problem were found via OBX extractions
15:33:51 [Zakim]
15:34:03 [Zakim]
15:34:28 [ericP]
... an OWL ontology tells us how to process a message
15:34:40 [ericP]
ack me
15:35:34 [ericP]
Cecil: the ontology links all the knowledge
15:36:01 [ericP]
... it guides parsing the message by aligning the OBX-extracted facts with an RDF graph
15:36:26 [ericP]
... we can then use the JESS reasoner for evaluating these facts
15:36:53 [ericP]
... JESS (Java Expert System Shell) is a rules FW/BW chaining rules engine
15:37:09 [ericP]
... has a protege plugin, interprets SWRL
15:37:38 [ericP]
... good commercial tool for high-volume processing
15:38:10 [ericP]
... payed for by tax dollars, only free for government use
15:38:32 [ericP]
... ($75K)
15:39:03 [ericP]
ack me
15:39:15 [Stuart]
15:39:17 [iker]
15:39:23 [mr_sticky]
Drools is from JBoss
15:39:41 [mr_sticky]
15:39:48 [ericP]
Cecil: we tried Drools, which has FW/BW chaining and similar fact structure
15:40:05 [ericP]
... use JESS if you're processing millions of facts
15:40:12 [ericP]
Joanne_Luciano: and Jena?
15:40:19 [ericP]
Cecil: no experience with it
15:40:39 [ericP]
ack me
15:42:04 [ericP]
Cecil: at OTR, we pass what we expect to see and what we got as two graphs
15:42:53 [ericP]
... the choreography of the OTR framework works out that something is a question about an e.g. resistance pattern of anitbiotic
15:43:10 [ericP]
... we have a set of "listeners" (patterns)
15:43:36 [ericP]
... we built this on V3 semantics, but mapped back to V2 syntax
15:44:09 [ericP]
... once we've matched the graph against the patterns, we pass it to jess
15:45:04 [ericP]
... we give jess the profile for an e.g. normal patient, MDR (multi drug resistant) patient, XDR (extensive drug resistant) (potential super-spreader)
15:45:33 [ericP]
... the reasoning framework decides if an event needs action
15:46:13 [ericP]
... another listener strains through alerts from JESS for outbound messaging
15:46:33 [ericP]
... we also use the output for visualization
15:47:03 [ericP]
... folks don't need to need to use SAS to extract this data from mid-tier, instead just using graph representations
15:47:44 [ericP]
... with agreement from CDC, we could have sent output messages back to reporters
15:47:53 [ericP]
... output:
15:47:58 [ericP]
... .. drug resistant
15:48:17 [ericP]
... .. appropriateness of drugging (per WHO codes)
15:48:39 [ericP]
... .. predictive analysis of whether someone is likely to fall off treatment based on patient history
15:48:54 [ericP]
[slide 7: Types of problems that could be solved by extending the TB framework]
15:49:16 [ericP]
Cecil: had to bend to time and budget limitations
15:49:44 [ericP]
... we could have added a d2rq interface to retrofit the pre-existing data
15:49:50 [ericP]
... a lot we could have done
15:50:02 [ericP]
[slide 8: The use of an OWL ontology]
15:50:05 [ericP]
15:50:53 [ericP]
[slide 9: HL7 Message Artifact Taxonomy]
15:51:11 [ericP]
Cecil: this is how we mapped the OBX structure to the ontology
15:52:13 [ericP]
[slide 11: Rule Processing]
15:52:33 [ericP]
[slide 12: Message Content Validation Rule Implementation]
15:52:49 [ericP]
Cecil: this demonstrates the advantage of using OWL
15:53:00 [ericP]
... the blue is what we deleted
15:53:08 [ericP]
... (from TIMS)
15:53:25 [ericP]
... went from 358 to 175
15:53:42 [ericP]
... reduces frustration of reporters facing conflicting rules
15:54:26 [ericP]
... beyond OWL being able to do syntax, vocabulary, rule processing, we see the advantage of declarative rules
15:54:40 [ericP]
[slde 13: Message Content Validation Rules]
15:55:23 [ericP]
Cecil: with tons of volume and response time requirements, you need a more efficient bw-chaining system (JESS)
15:55:36 [ericP]
[slide 14: Message Content Validation Results View]
15:55:43 [ericP]
Cecil: sample output
15:55:58 [ericP]
[slide 15: Processing Results]
15:56:11 [ericP]
Cecil: average processing time 3.5s round trip
15:56:26 [ericP]
... far faster than a human, and more accurate
15:56:38 [ericP]
... scales up to ~350k messages/day
15:57:03 [ericP]
... ~300K TB messages/year
15:58:09 [mscottm]
q+ on summary of SemWeb advantages
15:58:34 [ericP]
Cecil: could scale to influenza
15:59:00 [ericP]
... at worst case (4 month window), 50-75M, so ~ 200K message/day
15:59:17 [ericP]
... in a surveillance, you're also looking at folks who don't have it
16:00:13 [ericP]
... feeds from 800 VA hospitals, + laps a quest and labcore, ...
16:00:23 [ericP]
... congress says we need response in 2 mins
16:00:50 [ericP]
... had to put everything in memory
16:01:03 [ericP]
... biosense lost funding
16:01:18 [mscottm]
16:01:51 [ericP]
mscottm: summary of SemWeb advantages is very different from our usual tech demos in HCLS
16:02:05 [ericP]
... what are your SemWeb wins?
16:02:16 [ericP]
... what could be improved?
16:03:00 [ericP]
charlie: would like formal continuation
16:03:13 [ericP]
... to help us find focal points in HCLS
16:03:43 [ericP]
Cecil: SemWeb is a flexible way to extract knowledge
16:03:59 [ericP]
... we were given a TB messaging system and a deadline
16:04:25 [ericP]
... 7 days before deadline, CDC said we'd like to upgrade a 1.2 of our implementation guideline
16:04:39 [ericP]
... had around 35 new rules and 100 terminology changes
16:05:05 [ericP]
... because everything CDC gave us was in the OWL. expected to do it in 4 days
16:05:36 [ericP]
... made it on 4 days with no additional charge to CDC
16:06:00 [ericP]
... big commercial motivation is the flexibility at responding to rapidly changing knowledge
16:06:23 [ericP]
... at NCI, i wanted to build an EMR system
16:06:37 [ericP]
... NCO SHARP projects kind of get to this
16:06:49 [ericP]
... win 1: rapid software engineering
16:07:00 [ericP]
... win 2: rule validation
16:07:13 [ericP]
... win 3: can infer things that a human has problems inspecting
16:07:47 [mscottm]
Nice to hear that experience in the field confirms my main sales pitch about advantage of SemWeb tech for software: easier maintenance and change, agile development, effectively lower cost.
16:07:51 [ericP]
... .. (large systems (e.g. BRIDG's UML) hard to swap into a brain)
16:08:27 [Zakim]
16:10:05 [Zakim]
16:11:43 [iker]
+1 April 4th
16:13:30 [Zakim]
16:13:40 [Zakim]
16:14:01 [Zakim]
16:14:13 [michel]
zakim, who is here?
16:14:13 [Zakim]
On the phone I see Tony.a, charlie, egombocz, bosse, Joanne_Luciano, Cecil, mscottm, StuartTurner, iker, Bob_Powers, ericP, [IPcaller]
16:14:15 [Zakim]
On IRC I see michel, mscottm, RRSAgent, Stuart, qiip, iker, Cecil, mr_sticky, troy, egombocz, Zakim, amrapali, Bosse, Guoqian, bobP, MacTed, ericP
16:14:23 [michel]
rrsagent, draft minutes
16:14:24 [RRSAgent]
I have made the request to generate michel
16:14:29 [michel]
rrsagent, make log world-visible
16:16:04 [Zakim]
16:16:23 [Zakim]
16:16:35 [Zakim]
16:16:36 [Zakim]
16:16:36 [Zakim]
16:16:37 [Zakim]
16:16:37 [Zakim]
16:16:39 [Zakim]
16:16:40 [Zakim]
16:16:41 [Zakim]
16:16:42 [Zakim]
16:16:43 [Zakim]
SW_HCLS()11:00AM has ended
16:16:45 [Zakim]
Attendees were Bob_Powers, Tony, EricP, charlie, egombocz, bosse, +1.518.276.aaaa, Joanne_Luciano, +1.415.537.aabb, mscottm, +1.301.443.aacc, Cecil, +1.507.269.aadd, mayo,
16:16:48 [Zakim]
... StuartTurner, iker, [IPcaller]
16:19:06 [egonw]
egonw has joined #HCLS