Last Updated: 04/29/2011
NEXT EMAIL
From: Peter Beninato Sent: Friday, April 29, 2011 11:42 AM To: Mendis, Michael E.; i2b2 AUG Members
Subject: RE: i2b2 - v1.4 - default.htm - check role
Hi,
Thanks for the info.
Is there any way to make it contingent on role?
I was unable to get the approach outlined by Brian Wilson to work either. Javascript/web is not my strength.
I can always just comment out the link, and set up an alternative default_analysis.htm page that includes the analysis link.
It seems though that it might be a useful feature to have an approach to turn on/off links based on role. For instance, admin too, besides analysis.
Thanks for any replies.
Peter
From: Mike Mendis Sent: Friday, April 29, 2011 6:21 AM To: Peter Beninato; i2b2 AUG Members
Subject: Re: i2b2 - v1.4 - default.htm - check role
Peter,
To disable analysis tools: EDIT the i2b2_config_data.js file in the root of your webclient and change the AllowAnalysis from true to false.
{ name: "appdev", domain: "i2b2dev", urlCellPM: "http://site/i2b2/rest/PMService/", allowAnalysis: true,
debug: true }
JavaScript errors: You will need to reference the i2b2 framework such as <script type="text/javascript" src="js-i2b2/i2b2_loader.js"></script>
And will also need to initialize the code, if you look at the default.htm around this area: /******************* INITIALIZE *******************/
mike
On 4/28/11 2:25 PM, Peter Beninato wrote:
Hi,
Trying to figure how to check a user before presenting a link to them in default.htm. I'm trying to hide the Analysis Tools link.
Do I need to "import" the PM js in order to make the test below?
I see that there are calls to other parts of PM, up in the javascript section of the default.htm: i2b2.PM.doLoginDialog();
So I'm not sure why the call below doesn't work.
I get, "i2b2.PM is undefined" as error
<Script Language="JavaScript"> if (i2b2.PM.model.userRoles.length == 2 && i2b2.PM.model.userRoles[0] == "DATA_OBFSC") { // eventually document.write ahref tag document.write('testme'); } </Script>
Thanks.
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Peter Beninato Sent: Thursday, April 28, 2011 12:15 PM To: Arvinder Choudhary; i2b2 AUG Members; Mendis, Michael E.; Murphy, Shawn N.
Subject: RE: Optimal Block size for i2b2
Can you be sure the performance difference is only related to block size?
Might there be something else impacting I/0. Are they both on RAID 10?
From: Arvinder Choudhary Sent: Thursday, April 28, 2011 9:07 AM To: i2b2 AUG Members; Mike Mendis; Shawn N. Murphy
Subject: Optimal Block size for i2b2
Mike/Shawn and i2b2 Team:
We are want to know what is the Optimal block size of Oracle database for the i2b2 Application. We are currently have 2 databases with 8K and 32K block size. In 8K block size database i2b2 Application runs atleast 50 times faster (for the huge queries). I looked the explain plan for the same query on both the databases and it looks better on 32K block size database.
Regards
Arvinder Choudhary
BE MS PMP
Medical informatics Project Director
University of Kansas Medical Center
3901 Rainbow Blvd, KS 66160
NEXT EMAIL
From: Audrey Hureau Sent: Thursday, April 28, 2011 7:56 AM To: i2b2 AUG Members
Subject: SHRINE
Hi everyone,
I am interested in SHRINE so I would like to know if there is a SHRINE package that I could download. Thank you all.
Audrey Hureau
NEXT EMAIL
From: Murphy, Shawn N. Sent: Wednesday, April 27, 2011 6:40 PM To: Russ Waitman; i2b2 AUG Members Cc: jhuan; josh.denny
Subject: RE: ECG and UHC data sources in your i2b2 instance
Hi Russ,
Yes, I would expect indentified (pre) UHC data to be perfect for i2b2.
There is a tutorial on how to build a cell around the PFT model at: https://www.i2b2.org/software/tutorial.html
Thanks,
Shawn.
From: Russ Waitman Sent: Wednesday, April 27, 2011 9:00 AM To: i2b2 AUG Members; Murphy, Shawn N. Cc: jhuan; josh.denny
Subject: RE: ECG and UHC data sources in your i2b2 instance
Shawn, Thanks for the feedback.
For the UHC data, we'd be catching it before it's sent to UHC and de-identified but we'd grab the same variables. If the concepts are good, it's a valid starting point for incorporating financially based data into i2b2. In the long run, many i2b2 sites are probably UHC sites. If they are used to UHC benchmarking and comfortable with that data, having the same elements in i2b2 provides familiarity.
But our i2b2 installs would also have rich clinical data that is linked. It may also help with comparison across i2b2 sites for efforts like SHRINE to know that the financial components are based on the UHC approach. I didn't know if someone's already been down this road. I am not that knowledgeable about the work involved in preparing UHC datasets but the thought was raised by our hospital.
Regarding ECGs, we'll look at the PFT cell. I'd need to check with Josh Denny at Vanderbilt to see how hard it would be to incorporate that logic into i2b2 in a similar manner. Practically, I've got to figure out if I could get my hands on the cardiologists' reports here. We're at the start of a collaboration with researchers at our other campus where the topic of QT prolongation arose. I was wondering if people had incorporated any type of ECG data (either QTC from the machine or interpretations) into their i2b2s (beyond diagnosis).
Russ
Murphy, Shawn N. 4/27/2011 6:35 AM
Hi Russ, I know that an issue with the UHC data is that it is usually provided back in a "de-identified" way making it difficult to reliably match with the internal clinical data.
For the ECG data, the workflow you are thinking of associating with processing the interpretations sounds similar to the PFT cell workflow, although your engine is considerably more sophisticated of course. If you do make a cell around your interpretation engine, that would be a great cell to have available to the community.
Thanks, Shawn.
From: Russ Waitman Sent: Tuesday, April 26, 2011 4:47 PM To: i2b2 AUG Members
Subject: ECG and UHC data sources in your i2b2 instance
Hi, We are looking at supplementing our i2b2 instance with more information and wondered if anyone had experience incorporating the following data. The electronic medical record here is Epic but it isn't used for billing.
Cardiology ECGs.
Our hospital uses Tracemaster for ECGs quantitative information and cardiology interpretation. Some of this data is interfaced back to Epic but our initial sense is it's not the data we would like (QRS duration as opposed to QTC).
Are other sites with i2b2 storing things like corrected QT intervals and cardiologists interpretations? – If so, are you storing the quantitative value from the machine and/or the cardiologist interpretation?
In a previous efforts, we compared the two: Denny JC, Miller RA, Waitman LR, Arrieta MA, Peterson JF. Identifying QT prolongation from ECG impressions using a general-purpose Natural Language Processor. Int J Med Inform. 2009 Apr;78 Suppl 1:S34-42. Epub 2008 Oct 19.
Hospital Billing data and UHC benchmarking
We've loaded the diagnoses (ICD9) from Epic which are clinician entered. These diagnoses mainly derive from the problem lists or as reasons for ordering. This data goes back to Nov 2007. Most of this information is inpatient, cancer center, and emergency room. The clinics have started adopting Epic for their outpatient EMR in earnest last Fall.
We've incorporated the outpatient billing data (ICD and CPTs) from a separate system (GE IDX). This data goes back to 2000.
We would like to bring in for comparison and going back in time the hospital billing diagnoses, insurance status, and MS-DRGs. The hospital billing data is maintained in Seimens SMS. While we could get a feed or dump from SMS, one approach would be to build off of data that is already extracted and provided to the
University Healthsystem Consortium (UHC). I know other data is also reported to UHC surrounding pharmacy utilization.
Has anyone looked at incorporating data extracted for UHC benchmarking into their i2b2 instance? It could be attractive in that it's part of an existing data cleaning process and people doing analysis in UHC could compare apples with i2b2 which may incorporate additional clinically derived data.
Russ Waitman Associate Professor Director of Medical Informatics Department of Biostatistics University of Kansas Medical Center http://informatics.kumc.edu
NEXT EMAIL
From: Russ Waitman Sent: Wednesday, April 27, 2011 9:00 AM To: i2b2 AUG Members; Murphy, Shawn N. Cc: jhuan; josh.denny
Subject: RE: ECG and UHC data sources in your i2b2 instance
Shawn, Thanks for the feedback.
For the UHC data, we'd be catching it before it's sent to UHC and de-identified but we'd grab the same variables. If the concepts are good, it's a valid starting point for incorporating financially based data into i2b2. In the long run, many i2b2 sites are probably UHC sites. If they are used to UHC benchmarking and comfortable with that data, having the same elements in i2b2 provides familiarity.
But our i2b2 installs would also have rich clinical data that is linked. It may also help with comparison across i2b2 sites for efforts like SHRINE to know that the financial components are based on the UHC approach. I didn't know if someone's already been down this road. I am not that knowledgeable about the work involved in preparing UHC datasets but the thought was raised by our hospital.
Regarding ECGs, we'll look at the PFT cell. I'd need to check with Josh Denny at Vanderbilt to see how hard it would be to incorporate that logic into i2b2 in a similar manner. Practically, I've got to figure out if I could get my hands on the cardiologists' reports here. We're at the start of a collaboration with researchers at our other campus where the topic of QT prolongation arose. I was wondering if people had incorporated any type of ECG data (either QTC from the machine or interpretations) into their i2b2s (beyond diagnosis).
Russ
Murphy, Shawn N, 4/27/2011 6:35 AM
Hi Russ,
I know that an issue with the UHC data is that it is usually provided back in a "de-identified" way making it difficult to reliably match with the internal clinical data.
For the ECG data, the workflow you are thinking of associating with processing the interpretations sounds similar to the PFT cell workflow, although your engine is considerably more sophisticated of course. If you do make a cell around your interpretation engine, that would be a great cell to have available to the community.
Thanks, Shawn.
From: Russ Waitman Sent: Tuesday, April 26, 2011 4:47 PM To: i2b2 AUG Members
Subject: ECG and UHC data sources in your i2b2 instance
Hi, We are looking at supplementing our i2b2 instance with more information and wondered if anyone had experience incorporating the following data. The electronic medical record here is Epic but it isn't used for billing.
Cardiology ECGs.
Our hospital uses Tracemaster for ECGs quantitative information and cardiology interpretation. Some of this data is interfaced back to Epic but our initial sense is it's not the data we would like (QRS duration as opposed to QTC).
Are other sites with i2b2 storing things like corrected QT intervals and cardiologists interpretations?
If so, are you storing the quantitative value from the machine and/or the cardiologist interpretation?
In a previous efforts, we compared the two:
Denny JC, Miller RA, Waitman LR, Arrieta MA, Peterson JF. Identifying QT prolongation from ECG impressions using a general-purpose Natural Language Processor. Int J Med Inform. 2009 Apr;78 Suppl 1:S34-42. Epub 2008 Oct 19.
Hospital Billing data and UHC benchmarking
We've loaded the diagnoses (ICD9) from Epic which are clinician entered. These diagnoses mainly derive from the problem lists or as reasons for ordering. This data goes back to Nov 2007. Most of this information is inpatient, cancer center, and emergency room. The clinics have started adopting Epic for their outpatient EMR in earnest last Fall.
We've incorporated the outpatient billing data (ICD and CPTs) from a separate system (GE IDX). This data goes back to 2000.
We would like to bring in for comparison and going back in time the hospital billing diagnoses, insurance status, and MS-DRGs. The hospital billing data is maintained in Seimens SMS. While we could get a feed or dump from SMS, one approach would be to build off of data that is already extracted and provided to the University Healthsystem Consortium (UHC). I know other data is also reported to UHC surrounding pharmacy utilization.
Has anyone looked at incorporating data extracted for UHC benchmarking into their i2b2 instance? It could be attractive in that it's part of an existing data cleaning process and people doing analysis in UHC could compare apples with i2b2 which may incorporate additional clinically derived data.
Russ Waitman Associate Professor Director of Medical Informatics Department of Biostatistics University of Kansas Medical Center
NEXT EMAIL
From: Jack London Sent: Tuesday, April 26, 2011 10:24 AM To: i2b2 AUG Members members
Subject: vital_status_cd codes
The distributed demo data appears to have undefined codes in the PATIENT_DIMENSION table for VISTAL_STATUS_CD, such as "F" "B" and "T"
The CRC documentation (v1.5.2) lists only "N" "Y" "M" and "X" as valid codes. Is this correct?
Jack
Jack London, Ph.D. Research Professor Cancer Biology Thomas Jefferson University Director, Informatics Shared Resource Kimmel Cancer Center 808 BLSB, 233 S. 10th St. Philadelphia, PA 19107
NEXT EMAIL
From: Phillips, Lori C. Sent: Tuesday, April 26, 2011 9:37 AM To: Audrey Hureau; AUG Members
Subject: RE: some issues
Audrey,
Do you see any ontology in the navigate view at all?
My guess is that the table_access table has not been configured for your metadata.
Place an entry in this table that specifies the root node(s) of your metadata.
c_table_cd can be whatever you want it to be; it just needs to be unique.
Also, verify that the information in the table ont_db_lookup is correct.
As for your second problem, Raj indicated that you need to have the files
Stax-api-1.0.1.jar and jaxb-api.jar in the JBOSS_HOME/lib/endorsed folder.
Please verify that they appear.
Lori
From: Audrey Hureau Sent: Tuesday, April 26, 2011 8:16 AM To: i2b2 AUG Members
Subject: some issues
Hi everyone,
I installed i2b2 1.5 on my server and I have some issues. I loaded my own metadata in the I2B2 table but when I log into the workbench client, I do not see these concepts in the ontology (navigate view). Is there something that I omitted to do?
Furthermore, when I try to update the ontology in the Hive using the Synchronization in the Edit Terms view, there is an error that appears during the uploading data file to FRC process. You can find enclosed the error that appears. It may be a jboss problem.
Do you have any ideas to fix my problems?
Thank you all.
Audrey Hureau
NEXT EMAIL
From: Churchill, Susanne E. Sent: Wednesday, April 20, 2011 3:02 PM To: i2b2 AUG Members
Subject: Date clarification – AUG June Meeting
Please note that the dates of the AUG meeting (June 28-29) are a Tuesday/Wednesday combo. The SHRINE meeting is June 29-30, a Wednesday/Thursday combo.
From: Churchill, Susanne E. Sent: Wednesday, April 20, 2011 1:59 PM To: i2b2 AUG Members
Subject: FW: AUG June Meeting
Hi Folks,
The i2b2 team is looking forward to seeing you ALL at our first dedicated AUG Annual Meeting here in Boston, June 28-29, 2011. Attached please find a high level Agenda and a logistics advisory for accommodations. Please note the following:
1. We will offer a half day NLP Workshop on the morning of June 28. Agenda under development.
2. For planning purposes please note that there is an invitational meeting on SHRINE immediately following the close of the AUG meeting on June 29 and extending through noon on June 30. Invitations are going out now and should include all the i2b2 sites with interest in data sharing.
3. There will be no registration fee as we are seeking sponsorships to cover the costs of the meeting.
4. Note that we have allocated a significant time block for the poster session - as one of the main drivers in organizing this meeting was to allow networking and thereby identify synergies, solutions, collaborations, etc. We strongly encourage you to strut your stuff at whatever level you would like to do so. We may be able to arrange for staged demos if there is sufficient interest. We will group the posters by subject area and intend to use this forum to cover some of the areas you expressed interest in in our earlier poll for ideas (e.g., temporal relations, population health, integrating novel data, use cases, new plug ins under development, custom extensions). The posters will stay up for the duration so it may be possible to organize a concurrent break out session if there is a compelling topic of interest to a critical mass who have already toured the posters.
5. We will host a networking buffet dinner on the 28th at the Conference Center and will assume if you are attending the meeting you will also join us for dinner. Lunch will be provided on the first day (June 28).
Would you please:
1. Register to attend via a simple return "yes" to this message; please note if you will not attend the dinner.
2. Register with me your intention to bring a poster no later than June 17. Posters can be up to 4' x 4'.
3. Let me know if you'd like to offer a demo during the poster slot and I'll work with you to accommodate.
4. Note that the NLP Agenda is not yet completed but will follow shortly with separate request for registration.
Thanks! Questions to me, please.
Susanne
Susanne E. Churchill, Ph.D. Executive Director i2b2/Nat'l Center for Biomedical Computing 77 Avenue Louis Pasteur, Room 255 Boston, MA 02115
NEXT EMAIL
From: Mendis, Michael E. Sent: Tuesday, April 19, 2011 3:29 PM To: Tom Mish; i2b2 AUG Members
Subject: Re: Details on Project Request plugin
Glad that the Project Request is working out for you. You can find that file on the www.i2b2.org/software page in the Optional Components section.
Here is a direct link: https://www.i2b2.org/software/download.html?d=214
Besides the intall doc that you have, that is all the current documentation on this plugin.
On 4/19/11 3:24 PM, Tom Mish wrote:
All,
I'm looking for more details on the Project Request plugin. I've found the installation guide which appears to be a handy-dandy users guide to how to use the interface. It all works as advertised. Looking around, I found in an old email archive that I should look for some sql scripts in i2b2DerivedDataMarts-src-14.zip that will help me with this task. I can't find these anywhere.
Any help would be appreciated...
Thomas Mish, MS
Systems/Clinical Informatics Biomedical Computing Group School of Medicine and Public Health, UW-Madison Rm 256E, 610 Walnut St.
Madison, WI 5726
NEXT EMAIL
From: Peter Beninato Sent: Tuesday, April 19, 2011 12:13 PM To: i2b2 AUG Members
Subject: i2b2 - v1.6 RC3 - admin – Users
Hi,
Working setting up the AGG_SERVICE_ACCOUNT and wanted to share some feedback:
1) "Unidentified" User under projects
Not sure where this is coming, there is a blank user under each project when you display for Roles it is identified as "undefined"
2) Drop-down lists and alphabetization
Not sure how user friendly this version of the Admin app is. It used to be you could add Roles pretty
easily now this is a separate. Could users be presented in a drop-down when trying to add to projects. And when listing under projects are righthand side, could they be in alphabetic order
3) AGG_SERVICE_ACCOUNT
It seems this user needs aggregation privledge/role. Does this imply that it needs to be added to all projects?
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Kuttan, Rajesh Sent: Monday, April 18, 2011 8:25 AM To: Murphy, Shawn N.; Peter Beninato; i2b2 AUG Members
Subject: RE: i2b2 - v1.6 RC3 – tns
Hi Peter,
The TNS error when running a CRC query could be related to the database configuration. You can find which datasouce config is giving this error from the server log.
The changes to the CRC JMS config(crc-queryresponse-q-service.xml) is to delete the unused response messages, which would speedup the query and this change can also be applied to the older CRC versions.
Thanks Raj
From: Murphy, Shawn N. Sent: Sat 4/16/2011 10:20 AM To: Peter Beninato; i2b2 AUG Members Cc: Kuttan, Rajesh
Subject: RE: i2b2 - v1.6 RC3 - tns
Hi Peter,
A major fix occurred in 1.6 RC3 in the CRC Cell to JMS that at times was preventing queries that had completed from returning a notification of such to the client. Raj can comment.
Thanks, Shawn.
From: Peter Beninato Sent: Thursday, April 14, 2011 6:42 PM To: i2b2 AUG Members
Subject: i2b2 - v1.6 RC3 - tns
Hi,
I'm running a local install on Windows 7.
Installed once, trying to use schemas that were unique to 1.6 as I had a local 1.4 running.
I noticed that the JMS tables were not created in i2b2pm16.
It seemed to me that even though I edited config files for that schema, my guess was that it was still pointing to i2b2pm as I saw some sort of record update date.
Perhaps that schema name is hard-coded somewhere in application. fyi.
Oddly, I could logon and see ontology, but when I ran query it didn't seem to come back.
So. I abandoned dual install intention, and pointed all schemas to 1.4 schemas ( they don't have "16" at end like i2b2demodata and not i2b2demodata16).
I cleared schemas and installed v1.6
I could logon and went to run query. didn't come back.
I noticed in the jboss log an error pertaining to TNS ( sorry I didn't scrape it out) like
ERROR can't resolve TNS: localhost:XE
Ok. it is the end of day. I go home. ZZZZ
Day 2
Well you know the webclient folder is named webclient_v1-6, and the documentation says "webclient", so I rename (WARNING this is RED HERRING).
Hey I can run query, rename back to "webclient_v1-6". I can still run query.
Good and bad.
Not sure what problem was.
It could be a network problem as my tnsnames is stored out on a share. Or???
Has anyone else experienced flaky TNS errors with v1.6 RC3?
Peter Beninato - OCTRI DW Developer
Oregon Health & Science University
NEXT EMAIL
From: Murphy, Shawn N. Sent: Saturday, April 16, 2011 10:17 AM To: Peter Beninato; i2b2 AUG Members
Subject: RE: i2b2 - RC ---> Production ready?
Hi Peter and AUG members,
Zak, Susanne, Mike and I had a long discussion about this [deploying a release candidate (RC) to a production environment] at our weekly i2b2 meeting yesterday. Or official position on this is that the RC's are for informing the community of new features and functionality but should NOT be considered production ready. After version 1.6 is released we will be opening up the RC's to parties outside of the AUG, and it will be accompanied by an "alpha" type license. A very important part of the agreement will be to NOT consider RC's as production ready, especially considering liability issues.
There is nothing to prevent you from taking the code and performing your own QA, and at your own risk to make these part of your production environment. However, although we are obviously very interested in bugs and issues that arise in the RC, there is no commitment to fixing the code in the RC and likely any fixes will be present only in the next RC.
The RC"s expressly do NOT have upgrade scripts. Please do not expect them in the RC's. The features that are not yet officially released in the RC's (although they may be there in the code) should NOT be considered functional. It is doubtful the unreleased features have been QA'd at all, and they may easily change in the next RC.
Finally, please note that we smooth out compatibility with older releases from your feedback. The client software should be able to adapt to the previous release of Core i2b2 software (although obviously will not be able to show the new functionality, it should not just crash). If you find this is not the case, please let us know.
Thanks, Shawn.
From: Peter Beninato Sent: Thursday, April 14, 2011 6:56 PM To: i2b2 AUG Members
Subject: i2b2 - RC ---> Production ready?
Hi,
Notwithstanding my previous post about tns issues.
What is the consensus about the wisdom of deploying a release candidate to a production environment?
Are release candidates QA'd for features that are included?
For instance, v1.6 RC3 has the ability to link by encounter_num, was that functionality fully-tested?
Are a full set of unit and acceptance tests run on release candidates?
Does the test suite run on release candidates include regression tests to ensure that existing functionality is not broken?
Is it folly to install a release candidate to production?
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Murphy, Shawn N. Sent: Saturday, April 16, 2011 10:00 AM To: Peter Beninato; i2b2 AUG Members
Subject: RE: i2b2 - v1.6 RC3 - service account
Hi Peter,
Yes, we needed to "upgrade" the permissions of the service account to "aggregated data user" in order for the Ontology cell to obtain the unobfuscated patient counts from the CRC to populate the total_num column of the Ontology cell. The change in name reflects this change. The old service account can continue to be used, but this new account will be needed for the patient counts to be obtained properly.
Thanks, Shawn.
From: Peter Beninato Sent: Thursday, April 14, 2011 7:04 PM To: i2b2 AUG Members
Subject: i2b2 - v1.6 RC3 - service account
Hi,
Is the AGG_SERVICE_ACCOUNT new in version 1.6?
What is it's pupose?
Could the application not use the schema owner logon which has the credentials configured in the application?
Now there are two service account the OBSfucation account and the AGG account.
Do these users really need to be in the userbase/admin application?
For security, the passwords need to be changed from the default passwords.
Peter Beninato - OCTRI DW Developer
Oregon Health & Science University
NEXT EMAIL
From: Þorvarður Jón Löve Sent: Friday, April 15, 2011 10:43 AM To: i2b2 AUG Members
Subject: i2b2 implementation plan?
Dear i2b2 AUG members,
I am working at implementing i2b2 as a solution to identify sets of patients for research at a University Hospital in Reykjavik Iceland. We will give PI´s access to i2b2, using data available in an administrative clinical data warehouse that runs on Oracle.
When implementing i2b2 at your institutions, did you create a formal implementation plan? Would you be willing to share this, to give us an idea of the steps and time needed in your case?
Best,
Thorvardur Love, MD MMSc Rheumatologist, LSH Iceland
NEXT EMAIL
From: Murphy, Shawn N. Sent: Wednesday, April 13, 2011 8:34 AM To: Pul AVILLACK; i2b2 AUG Members
Subject: RE: i2b2 – Requests on primary and/or secondary diagnosis
Hi Paul.
The most elegant way may be the upcoming modifiers code which would allow this directly in version 1.6. The other two options would be to code it into the metadata (like you suggest) or set it up as a value that accompanies your ICD10 codes.
There is some discussion on modifiers on the i2b2 community Wiki in the developers forum under "data model" topic.
Thanks,
Shawn.
From: Paul AVILLACH Sent: Tuesday, April 12, 2011 3:33 PM To: i2b2 AUG Members
Subject: i2b2 - Requests on primary and/or secondary diagnosis
Hi! We have ICD10 classification for diagnosis (but it would be the same problem with ICD9) and we were wondering what would be the smartest way to make available the selection of only the primary and/or secondary diagnosis?
How did you make this choice available. Changing the metadata?
Thanks!
Best Regards
Paul Avillach
MD, PhD Student - Assistant Hospitalo-Universitaire - Informatique médicale
DIH-HEGP - APHP - 20 rue Leblanc - Paris 15
http://www.hegp.fr
INSERM UMRS 872, eq. 20 - SPIM -Université Paris Descartes
http://ics.upmc.fr/
http://www.spim.jussieu.fr/
NEXT EMAIL
From: Mendis, Michael E. Sent: Tuesday, April 12, 2011 12:15 PM To: Daniel Sullivan Cc: Jack London; Murphy, Shawn N.; i2b2 AUG Members members
Subject: Re: Inconsistent version values in Documentation and Settings
Daniel,
Thanks for catching that, we will make sure it is updated for the next rc. But these values are used when generating the return XML to the client.
In this case it will not cause a issue, but in some causes it might. The client uses the version return to make sure the component being run is compatible with the server. Such as in the new modifiers, it will only run on 1.6 i2b2 instances. So if the server sends a 1.5, than the client will inform the user that this feature is unavailable.
mike
On 4/12/11 12:08 PM, Daniel Sullivan wrote:
I am going through the build process for 1.6 and I cam across the following config setting for the CRC Loader namespace (please look at the property node for application version):
<bean id="appType" class="edu.harvard.i2b2.crc.loader.datavo.i2b2message.ApplicationType">
<property name="applicationName" value="edu.harvard.i2b2.crc.loader"/>
<property name="applicationVersion" value="1.5"/>
</bean>
So far (primarily in the documentation) I have come across similar blocks or sections where the documentation is pointing towards 1.5 but the code apparently is in 1.6 - this is the first chunk of code where I noticed a version inconsistency. The documentation does not say to change this, but should I? (and when I say 'this' I am referring to the block of property xml where the applicationversion value = 1.5).
Any thoughts? Does this setting really do anything?
NEXT EMAIL
From: Mendis, Michael E. Sent: Tuesday, April 12, 2011 11:50 AM To: Jack London; Murphy, Shawn N. Cc: i2b2 AUG Members members
Subject: Re: Web Client Queries using v1.6 modifier_cd
No, the next RC will have this.
mike
On 4/12/11 11:32 AM, Jack London wrote:
Shawn,
Is the v1.6 RC2 or RC3 web client query interface capable of using the modifier_cd?
Jack
Jack London, Ph.D. Research Professor Cancer Biology Thomas Jefferson University Director, Informatics Shared Resource Kimmel Cancer Center 808 BLSB, 233 S. 10th St. Philadelphia, PA 19107
NEXT EMAIL
From: Murphy, Shawn N. Sent: Monday, April 11, 2011 1:04 PM To: Mendis, Michael E.; Robert Schuff; i2b2 AUG Members
Subject: RE: i2b2 - v1.6 (RC3)- Data – upgrade
Especially with regards to upgrade paths this is true, as we generally have not thought through (and thus no scripts) on how the whole migration is going to occur. This is reserved for the full releases.
Thanks,
Shawn.
From: Mendis, Michael E. Sent: Monday, April 11, 2011 1:00 PM To: Robert Schuff; i2b2 AUG Members
Subject: Re: i2b2 - v1.6 (RC3) - Data - upgrade
Hi Rob,
It is true that the release candidates are fully QA'd, which is why it took longer to get RC3 out, it failed a few times in QA. But the components and features are close to production ready, the underlying XML and/or database might change slightly between release candidates and prior to the official version. We do not recommend using RC on production machines, but they are good to learn about the new features and where certain components are headed.
Mike
On 4/11/11 12:43 PM, Robert Schuff wrote:
Mike, I was under the distinct impression from speaking with Shawn that these release candidates are fully QA'd and are production quality releases rather than the test/development scenario you mention below. My understanding is that the Release Candidates are a mechanism for incrementally incorporating features into production-ready releases on the path for the full version release.
Can you all please clarify?
Thanks
Rob
Robert Schuff - Director, Clinical Research Informatics
Oregon Clinical & Translational Research Institute Oregon Health & Science University
http://www.octri.org/
From: Mendis, Michael E. Date: Mon, 11 Apr 2011 06:35:40 -0700 To: Peter Beninato, i2b2 AUG Members
Subject: Re: i2b2 - v1.6 (RC3) - Data - upgrade
We normally provide upgrade scripts and instructions from one version to another. Like 1.4 to 1.5 and from 1.5 to 1.6. For RC, we don't provide any upgrade scripts until the final is completed. The RC are intended to show the new features and run in a development/test scenario, once we release the final 1.6 it will contain a upgrade script from 1.5.
Mike
On 4/7/11 5:12 PM, Peter Beninato wrote:
Hi,
We are running v1.4 and starting to do upgrade to v1.6.
Trying to determine what schema changes there might be between the two versions?
Hoping there are none, but
In the i2b2 Installation Guide – Data, at step 2.9 Next Steps, it mentions a Hive Upgrade Guide.
The Hive and Data Upgrade guides seem to be missing from the zip file?
(I downloaded the v1.5 docs and seem them there)
Are there any updates from version 1.5 to version 1.6?
Maybe I can implement the v1.5 changes, and all will be well?
Please advise.
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Dan Connolly Sent: Monday, April 11, 2011 12:57 PM To: Arnold G Cc: i2b2 AUG Members
Subject: RE: RA requirement for the Application/Database server
We run Oracle on SuSE linux. Monitoring shows that RAM utilization never goes above 20% (of 8GB). I wonder if that's because Oracle doesn't see any way to use more memory to run our queries or if we somehow told Oracle not to use any more RAM.
Does anyone have suggestions regarding SGA_TARGET and the like?
http://download.oracle.com/docs/cd/B28359_01/server.111/b28274/memory.htm
And in the spirit of Arnold's experience report:
OS: SuSE linux 64bit
db: Oracle 10g (10.2.0.4); local 15000RPM disks in RAID 5, I think
hardware: HP dual processor cpus (6 cores each, I think; I forget the details) 8GB RAM -i2b2/jboss runs on a separate linux VM
memory usage: I'm trying to figure this out. I see sga_target=2449473536 in one of the relevant config files.
- of facts: ~0.5B (~3 years of data on ~300K patients, with demographics on ~1M patients)
- of users/usage: a handful
- response time: used to be in that 5-15 second range, but now we're tearing our hair out trying to figure out why similar queries are taking ~400 seconds.
Dan Connolly
KUMC medical informatics
On Fri, 2011-03-04 at 15:03 -0800, Arnold, Garth wrote:
Hello - I can't speak directly to needs for Oracle, but I find it useful to know the experience of others when sizing systems. In the hope that it's useful, here's my organization's i2b2 configuration and performance:
OS: Windows 2008 64-bit
db: SQL Server 2008 64-bit; disk use spread across 6 drives; drive space from iSCSI SAN
'hardware': virtual server, Intel Xeon-class CPUs; 4 vCPUs, 6GB RAM
memory usage: the i2b2 stack takes 300-400MB, SQL Server is configured to take 4GB; our DBAs suggest more RAM would be beneficial - of facts: ~300 million (10 years of data on ~500,000 members)
- of users/usage: light enough that we don't worry much about concurrent queries
response time: generally in the 5-15 second range
Garth
Garth Arnold | MANAGER, TECHNICAL INFRASTRUCTURE
Group Health Research Institute
www.ghc.org
NEXT EMAIL
From: Peter Beninato Sent: Friday, April 08, 2011 1:11 PM To: Dongsheng Zhu; i2b2 AUG Members
Subject: RE: i2b2 set value query slow for some concepts
Umm, sometimes indexes actually make things run more slowly.
One rule which of thumb, is if over 4% of data is returned, better to do full table scan. But of course you can really add hints to i2b2 queries.
Maybe you want to drop indexes.
I/O could be your bottle neck.
What Database vendor?
From: Dongsheng Zhu Sent: Friday, April 08, 2011 9:46 AM To: i2b2 AUG Members ; Peter Beninato
Subject: RE: i2b2 set value query slow for some concepts
Thanks for your input, Peter.
I was doing just one panel (one concept involved) set value query. Only a few concepts with the c_totalnum populated in our system this moment. However, some queries with concepts whose c_totalnum not populated runs very fast. I did try to populate the c_totalnum for couple of concepts with slow response, but did not do the trick. The exec plan shown the proper indexes were used.
We might go ahead to try adding some more indexes or looking into the high I/O wait issue.
Dongsheng
Peter Beninato 4/8/2011 11:02 AM
Hi,
Do you set the field, c_totalnum, in the i2b2 metadata table with the count of observations?
Apparently, the app uses that to determine which panel to go after and has an impact on query performance.
I would also suggest you go into the qt_master_query chop out the SELECT sql and run it against your database. Examine the execution plan. See what index is being used. You might want to add some indexes as you described, but I would do some testing first. For instance, I found an index on start_dt actually destroyed performance with I used any of the "set date" functionaliy, so I dropped it.
Fyi,
Peter
From: Dongsheng Zhu Sent: Thursday, April 07, 2011 12:12 PM To: i2b2 AUG Members
Subject: i2b2 set value query slow for some concepts
I2B2 community,
We are experiencing slow query response for some concepts with the set value functionality (300 to 600sec) for numeric value type. We have a total over 500million facts. It seems the response time is directly related to the total observation count for that concept. We also noticed high user I/O wait (over 96%). Anyone experienced similar situation? have anyone tried to add more indexes to the observation fact table (e.g., concept_code + nval_num) ? By the way, we are using 1.4 I2B2.
Thanks for your input
dongsheng zhu
Department of Biostatistics
KUMC
Kansas city, KS
NEXT EMAIL
From: Daniel Sullivan Sent: Friday, April 08, 2011 1:08 PM To: Dongsheng Zhu Cc: i2b2 AUG Members ; Peter Beninato
Subject: Failure to create index during data base build steps
This probably only applies if you are using MSSQL (but I would check index length constraints for ORACLE as well):
In the file: crc_create_datamart_sqlserver.sql An index is defined with 17 fields (all int) versus the previous version (16 fields) [going from 1.5 to 1.6] - they added a field to the index called [instance_num]. ANT (as currently scripted by the i2b2 build process) the messaging / failure model is partial. So, in this case, a "partial" success could be the tables get created but no index is created on ObservationFact. This would be caused by the build failing, so I doubt this is your problem, but I wanted to point this out.
I am doing some massaging of the .SQL files to remove ".dbo" or ".dbo." or "dbo." from the files (all three of these occur in the MSSQL build files).
On Fri, 8 Apr 2011, Dongsheng Zhu wrote:
Thanks for your input, Peter.
I was doing just one panel (one concept involved) set value query. Only a few concepts with the c_totalnum populated in our system this moment. However, some queries with concepts whose c_totalnum not populated runs very fast. I did try to populate the c_totalnum for couple of concepts with slow response, but did not do the trick. The exec plan shown the proper indexes were used. We might go ahead to try adding some more indexes or looking into the high I/O wait issue.
Dongsheng
Peter Beninato 4/8/2011 11:02 AM
Hi,
Do you set the field, c_totalnum, in the i2b2 metadata table with the count of observations?
Apparently, the app uses that to determine which panel to go after and has an impact on query performance.
I would also suggest you go into the qt_master_query chop out the SELECT sql and run it against your database. Examine the execution plan. See what index is being used. You might want to add some indexes as you described, but I would do some testing first. For instance, I found an index on start_dt actually destroyed performance with I used any of the "set date" functionaliy, so I dropped it.
Fyi,
Peter
From: Dongsheng Zhu Sent: Thursday, April 07, 2011 12:12 PM To: i2b2 AUG Members
Subject: i2b2 set value query slow for some concepts
I2B2 community,
We are experiencing slow query response for some concepts with the set value functionality (300 to 600sec) for numeric value type. We have a total over 500million facts. It seems the response time is directly related to the total observation count for that concept. We also noticed high user I/O wait > (over 96%). Anyone experienced similar situation? have anyone tried to add more indexes to the observation fact table (e.g., concept_code + nval_num) ?
By the way, we are using 1.4 I2B2.
Thanks for your input
dongsheng zhu
Department of Biostatistics KUMC Kansas city, KS
NEXT EMAIL
From: Peter Beninato Sent: Thursday, April 07, 2011 5:28 PM To: Key, Dustin; i2b2 AUG Members
Subject: RE: Querying age and enrollment
Hi Dustin,
In regards to #1 below
See attached txt file for routines that create i2b2 ontology metadata entry for "Visit Age" and also function for creating observation_facts.
Peter
From: Key, Dustin Sent: Thursday, April 07, 2011 10:48 AM To: i2b2 AUG Members
Subject: Querying age and enrollment
Hello! I have two questions related to the population of i2b2 tables. Would someone help me walk through some of the details and/or talk about similar challenges encountered/ruled-out potential solutions that have been encountered?
My first question is related to how to populate observation_fact (and also metadata?) so that leafs which represent a date or an age can be queried in a more interactive manner. For instance, let's say I have a leaf that is called "Age at Diagnosis." Would someone walk me through what I would need to populate so that when the user dropped this leaf into the query, a pop-up or similar option appears with the age range lets the user choose which ages will be reflected in the query result?
Second, I'm interested in whether anyone has created an enrollment subject area, or something similar. For instance, I'm interested in something that addresses the enrollment of patients in a health plan with a starting and ending date. Has anyone had experience with a similar ontology? Ideally we would like to let a user specify that they are interested in patients who were enrolled between two points in time.
Currently, we've implemented a method that handles enrollment by presenting one leaf node per year. In turn, we create a related facts for a patient, one for each enrollment year (e.g. ENR:2009) if said patient were enrolled during the majority of that year. I think we could make our current implementation less granular, and go to---say, quarters, but is there a better way?
In our observation_fact table, we are loading only the start_date field (along with encounter_num, patient_num, concept_cd.) Are there other fields we could load in order to gain more functionality related to my questions above?
Thanks!
Dustin
Group Health Research Institute
NEXT EMAIL
From: Peter Beninato Sent: Thursday, April 07, 2011 1:01 PM To: Murphy, Shawn N.; i2b2 AUG Members Cc: Churchill, Susanne E.
Subject: RE: i2b2 - Release Candidate - web page
Hi,
The "reset password" functionality worked today. I logged on and downloaded.
Thanks.
From: Murphy, Shawn N. Sent: Thursday, April 07, 2011 9:59 AM To: Peter Beninato; i2b2 AUG Members Cc: Churchill, Susanne E.
Subject: RE: i2b2 - Release Candidate - web page
Hi Peter, Yes, RC3 should be on there today, but not sure about your account. Please give Susanne Churchill a buzz.
Thanks,
Shawn.
From: Peter Beninato Sent: Wednesday, April 06, 2011 4:32 PM To: i2b2 AUG Members
Subject: i2b2 - Release Candidate - web page
Hi,
Rumor had it that 1.6 RC3 was going to be released.
Went to the RC page, but am having trouble logging in.
Also, when I go to reset password, I get "Sorry, an error occurred please try again."
Can someone investigate? Do I still have an account for release candidates?
Thanks.
Peter Beninato - OCTRI DW Developer Oregon Health & Science University
NEXT EMAIL
From: Russ Waitman Sent: Tuesday, April 05, 2011 12:31 PM To: isaac Kohane; i2b2 AUG Members members Cc: d mcnair; Randy Spoeri; John Keighley; Jonathan Mahnken; Judith Warren; Jo
Wick; Matthew Mayo
Subject: Re: Post about Cerner
Zac, That is a very timely post. I am giving a talk at Cerner on Friday as part of a KC regional statistics conference (ASA) on our use of i2b2 at KUMC and flowsheet data.
http://www.amstat.org/education/idad/
I'll see if I can learn more about Cerner's use of i2b2. I don't have a Cerner account but Judy Warren should (who works with them on educational simulation, SEEDS project).
I am ccing Randy Spoeri (Cerner Senior Director of Health Analytics) and Doug McNair (Cerner SVP of Research-Life Sciences) who may also be able to comment.
Russ Waitman
Associate Professor Director of Medical Informatics Department of Biostatistics University of Kansas Medical Center http://informatics.kumc.edu
Isaac Kohane 4/5/2011 10:43 AM
I just posted this: http://i2b2-zak.blogspot.com/2011/04/cerner-adopts-i2b2-model.html
If anyone has any insight or comments, feel free to share to this list.
Zak
NEXT EMAIL
From: Phillip Reeder Sent: Monday, April 04, 2011 7:07 PM To: Travis Nagler Cc: i2b2 AUG Members
Subject: Re: New 1.5 installation question
If I'm remembering my versions correctly, I think I was able to get it running with jboss 5.2.6, but i2b2 wouldn't work with 5.2.7. I'd suggest downgrading that to start with.
Does anyone have a matrix of the highest versions of the various components that are known to work?
Phillip
On Apr 4, 2011, at 5:03 PM, Travis Nagler wrote:
I am going through the installation instructions on the Hive site for a v1.5.2 install, starting with the Data Repository installation, but downloading and installing newer/current versions of the required software (JBoss 6.0.0_Final, Ant 1.8.2, Java 1.6.0_24 and Axis2 1.5.4) than what's listed in the instructions.
Other than the versions, I followed the instructions exactly for Java, JBoss, Ant, and Axis2 that were given. But now that I am to installing the CRC portion, the first thing it says to do is stop JBoss by running "./ $JBOSS_HOME/bin/shutdown.sh".
I do that but get an error:
[blah@blahBlah bin]$ ./shutdown.sh
Exception in thread "main" java.lang.NoClassDefFoundError: org/jboss/Shutdown
Caused by: java.lang.ClassNotFoundException: org.jboss.Shutdown
at java.net.URLClassLoader$1.run(URLClassLoader.java:202)
at java.security.AccessController.doPrivileged(Native Method)
at java.net.URLClassLoader.findClass(URLClassLoader.java:190)
at java.lang.ClassLoader.loadClass(ClassLoader.java:307)
at sun.misc.Launcher$AppClassLoader.loadClass(Launcher.java:301)
at java.lang.ClassLoader.loadClass(ClassLoader.java:248)
Could not find the main class: org.jboss.Shutdown. Program will exit.
[blah@blahBlah bin]$
Can anyone help me out, please? Or do I need to just install it with the older software versions, some of which are almost 3 years old?
You could email me directly if you need any more information which might help.
Thanks,
Travis
Travis H. Nagler
UC Davis Health System 2921 Stockton Blvd., Suite 1470 Sacramento, CA, 95817 http://www.ucdmc.ucdavis.edu
NEXT EMAIL
From: Audrey Hureau Sent: Monday, April 04, 2011 4:44 AM To: i2b2 AUG Members
Subject: Problem searching new features
Hello everyone,
I have a problem about add new features to the i2b2 workbench. When I click on Help - Add new features..., the wokbench begins to search for new features but it leaves over 6% and then an error message appears saying :
"Network connection problems encountered during search."
Then, I click on details for further information and I read :
"Unable to access https://www.i2b2.org/updateSite.
Error parsing site stream. [Premature end of file.]
Premature end of file.
Error parsing site stream. [Premature end of file.]
Premature end of file."
Then, I click on "ok" and a second message appears : "No features found on the selected site(s). Choose a different site or site category."
Does anyone know how I can fix this? Where can I choose a different site? Thank you all for your help.
Audrey.
