Demonstrations
Contents
From time to time we may deploy demonstrations of DS2 software and link to the demonstrations from this page. Please try out the demos, see screenshots below, or build a demonstration yourself by checking out the Downloads and Installation page. Send questions or comments to sharps@cdsframework.org.
Current Demonstrations
The are no demonstrations currently available.
Screenshots
Previous Demonstrations
The following demonstrations were available on the web previously. If you would like to request that these demonstrations be made available again, you may send email to sharps@cdsframework.org.
Inference Analyzer Demo
Demo overview:
DS2 Inference Analyzer is a web application to:
- Visualize patients’ clinical concepts and co-occurrences
- Evaluate predicates to asses predictive capabilities
- Leverage predicates to make inferences, find “missing” concepts, “hide” concepts, view impact of edits/redactions
For more information on the Inference Analyzer, please see the Inference Analyzer project page or read about it in the Technical white paper. Please note that the Inference Analyzer is only one of the component projects in DS2.
Demo configuration:
- Deployment dates: 10/7/2014 - 5/28/2016
- Server location: New York, NY
- DS2 Projects installed: DS2 Data Scripts, NaiveBayesPredicate and RandomForestPredicate, Inference Analyzer
- Training data set: CDC 2010 hospital discharges (weighted)
- Inference Analyzer data set: CDC 2010 hospital discharges (unweighted)
- Predicate Type: Naive Bayes, multi-label classifier
- Attributes: CCS with custom DS2 categories for HIV (custom categories are marked by asterisks)
Click here to access the demo site (demo link is no longer available). Want to install it yourself and customize it? See the Downloads and Installation page.
Predicate/Reducer Demo
Demo overview:
The Predicate/Reducer "cda-ws-web" web application allows you to:
Upload a CCD and convert it to vMR
View a high-level summary of clinical facts in the vMR along with Predicate/Reducer "tags" indicating HIV, Mental Health, or Substance Abuse
Specify a set of simple privacy preferences (whether to disclose HIV, Mental Health, or Substance Abuse) and redact clinical facts based on those preferences combined with the Predicate/Reducer decisions
Generate a CDA Consent document based on the simple privacy preferences
Download the vMR, tagged vMR, redacted CCD, and Consent document
For more information on the Predicate/Reducer "cda-ws-web" web application, please see the CDA Redaction Web Service and Application project page or read about it in the Technical white paper.
Demo configuration:
- Deployment dates: 8/19/2014 - 5/28/2016
- Server location: New York, NY
- DS2 Projects installed: DS2 Data Scripts, NaiveBayesPredicate, Predicate/Reducer, CDA Redaction Web Service and Application
- Training data set (for Naive Bayes classifier) : CDC 2010 hospital discharges (weighted)
- Predicate Type: DS2-developed OpenCDS deterministic rules for HIV, Mental Health, and Substance Abuse; Naive Bayes classifier called from OpenCDS via SimpleProbabilisticPredicate interface in addition to deterministic rules for HIV
- Attributes: CCS with custom DS2 categories for HIV
Tips for use:
- Feel free to upload any CCD - but be aware that OpenCDS does not accept clinical documents with code elements that do not contain code or originalText attributes, for example:
<routeCode codeSystem="2.16.840.1.113883.3.26.1.1" codeSystemName="NCI Thesaurus" displayName="Intramuscular" />
<code codeSystem="2.16.840.1.113883.6.88" nullFlavor="UNK">
<code codeSystem="2.16.840.1.113883.6.96" codeSystemName="SNOMED CT" nullFlavor="OTH">
- In these cases, a code attribute, for example, code="0" must be added to the element in the CCD before uploading, or else the demonstration application will return an error.
- Note that the red Predicate/Reducer "tags" (ILHIE_HIV, ILHIE_MentalHealth, ILHIE_SubstanceAbuse) will appear in the summary display regardless of the preferences selected at the top of the screen. Preferences drive the redaction, not the tagging.
- Note that many fake sample CCDs from around the Internet (including some of the samples that ended up in cda-web-service in the "samples" directory) were generated by hand and have entries that were cut-and-paste multiple times, repeating the same "id" element over and over again in the CCD. This is invalid, and the Predicate/Reducer does not work properly with these documents (tags will be applied to all clinical facts with the same id, etc.). As long as you use a real, valid CCD, you should not have this problem.
- Known issue: If a single clinical fact is marked with multiple reducer tags (i.e., a fact that is both HIV and Mental Health), the summary display at the top currently fails to show the tag.
- Known limitation: CCD redaction does not currently redact from <text> sections of the CCD (it would be a trivial modification to redact all text)
- Try one of the sample CCDs attached to this page – click on one of the links below to save the sample CCD to your computer; then go the the demo site, below, and upload the sample CCD to the Predicate-Reducer:
Click here to access the demo site (demo link is no longer available). Want to install it yourself and customize it? See the Downloads and Installation page.
Inference Analyzer Demo (multi-year data set)
Demo overview:
DS2 Inference Analyzer is a web application to:
- Visualize patients’ clinical concepts and co-occurrences
- Evaluate predicates to asses predictive capabilities
- Leverage predicates to make inferences, find “missing” concepts, “hide” concepts, view impact of edits/redactions
For more information on the Inference Analyzer, please see the Inference Analyzer project page or read about it in the Technical white paper. Please note that the Inference Analyzer is only one of the component projects in DS2.
Demo configuration:
- Deployment dates: 5/20/2014 - 10/5/2014
- Server location: Dallas, TX
- DS2 Projects installed: DS2 Data Scripts, NaiveBayesPredicate and RandomForestPredicate, Inference Analyzer
- Training data set: CDC 2006, 2007, 2008 hospital discharges (weighted)
- Inference Analyzer data set: CDC 2009 hospital discharges (unweighted)
- Predicate Type: Naive Bayes, multi-label classifier
- Attributes: CCS with custom DS2 categories for HIV (custom categories are marked by asterisks)
Click here to access the demo site (demo link is no longer available). Want to install it yourself and customize it? See the Downloads and Installation page.