March 2013 03-23-05 The Road to Semantic Interoperability The Path Not (Often) Taken Ken Rubin Chief Architect, Federal Healthcare Portfolio HP Enterprise Services ken.rubin@hp.com Disclaimers The information that follows is derived from either public information or personal experience. This information is a goodfaith representation, and every effort has been made to assure its accuracy, currency, and vendor/product neutrality. Nonetheless, these slides do not necessarily reflect the official position of HP, HL7, the US Government, the Veterans Health Administration, or any organizational affiliation. The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 2 Understanding the Business Challenge The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 3 A Little Background about Healthcare • Healthcare is casually referred to as the “Trillion-dollar Cottage Industry”, spanning geography, organizational boundaries, languages, cultures • Healthcare is collaborative • Health is an incredibly complex domain (more on this later!) • Health lags significantly behind other market sectors in % investment in IT, though the gap is narrowing • The Health IT (HIT) Landscape is largely dominated by a select number of well established vendors • HIT challenges are global; no country has this solved (though some are closer than others) The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 4 So why the push for interoperability? • When was the last time you went to see a doctor? • Were you asked for your allergies? • Were you asked about your medical history? • Were you asked about your current medications? • How confident are you that you remembered every medication? • What about your loved ones? An aging parent? • Have you ever brought a prescription slip to a pharmacy? • Do you know if any of your over-the-counter medications interact with your prescriptions? • When was your last Hep-A booster? Tetanus? The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 5 So, What is Interoperability? Quiz time. Are you interoperable if… – …you receive data that can be manually keyed or scanned into you system? – …batch extracts of data can be received and used on a periodic basis? – …you are able to write an adapter to interchange with a business partner’s IT system? – …you use accepted industry standard wire protocols? – …you can receive and ingest an XML document on-demand (real-time, or near real-time) from a business partner into your system? – …you use Web Services? The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 6 What is semantic interoperability? Quiz time. Are you interoperable if… – … you lack processes to validate accuracy of data entry? – …data is represented differently across systems within your organization – … you and your business partners use different codes? – … you and your business partners use the same codes, but use them differently? – …you use standardized codes to represent data, but allow localized extensions? The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 7 So, let’s contrast “IT” with “HIT” • A typical info system … • A typical healthcare IT system… – Supports the need for persistence – Impacts life-or-death decisions, literally. – Is designed to meet performance requirements – Has extreme high-availability requirements – Supports concurrency, scalability – Has Near-real-time performance expectation – Is designed by a DBA in conjunction with a project team – Must be capable of integrating content from external sources – Are closely coupled with the application it supports – Has a usable system life of 2-10 years The Road to Semantic Interoperability – Extreme sensitivity to privacy and security considerations – Must maintain data access for the lifetime of the patient (or longer!) March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 8 Real Requirements for an Electronic Health Record… • Capable of integrating data from our business partners and patients themselves • Data from business partners must maintain consistency in its meaning • Manage approximately 3000 unique data elements in 14 functional domains (laboratory, pharmacy, vitals, demographics, encounters, radiology/nuclear medicine, etc.) • Provide medical alerts for approximately 500k drug-drug and drug-allergy interactions and contraindications • Maintain data integrity for 75-years post mortem of patient (up to 150 years) retaining durable meanings The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 9 Tackling This Challenge… Key Solution Elements The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 10 Role of an Information Model • Establishes a basis for harmonizing and standardizing semantics • Clarifies data typing • Determines bindings to relevant terminologies • Assures consistent information representation • To depict structure and semantic relationships supporting (among other things) • Provides guidance for logical database design • Provides basis for message payload (e.g., interface parameters) The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 11 Understanding Terminology 101 • Terminology (or ontology) is a structured representation of data • Terminology is needed to allow for data comparability and consistency • Formal terminologies are based on concept codes that themselves have no inherent meaning • Key to this principle is to distinguish the concept itself from the label • Just because you have a common label (known as “surface form”) doesn’t mean you have a shared understanding • Let me show you what I mean… The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 12 Understanding Terminology 201 Same concept, different “surface form” Biblioteca Medellin, photo taken from Wikipedia •Library of Congress image from visitingdc.com website •Both photos believed to be in the public domain The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 13 Understanding Terminology 301 Same “surface form”, different concepts Salmon photo courtesy of Carly & Art, via Flicker, Creative Commons License •Potato Skins photo courtesy of Scorpions and Centaurs, Creative Commons License The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 14 Understanding Terminology 401 Same surface form, different concepts First photo courtesy of avlxyz, via Flicker, Creative Commons License Second photo taken from The Coffee Club (Australia) Website The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 15 So Let’s Explore “Computable” Data… Not all data representations are created equal • Content stored as strings without an underlying terminology cannot be used for [clinical] reasoning, alerts, interactions, epidemiology • Adherence to data constraints (e.g. 1000/500?) • A simple example: how many genders are there? • A VHA example: getting to Yes • The effort and importance of knowledge engineering and terminology cannot be overstated The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 16 A Simple example: “Yes-No” Codes (slide adapted from VHA) • The Standard Terminology Model for a collaborative across the US Federal Health programme reviewed “Yes-No” Code sets • Identified were 3396 instances of “Yes-No” use in DoD, VA and IHS • There were 30 unique ways to say it • E.g. Yes = 1 No = 2 • E.g. Yes = 0 No = 1 etc etc The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 17 LOINC for Hgb A1c • HEM 4547-6 HEMOGLOBIN A1/HEMOGLOBIN.TOTAL SFR PT BLD QN HAEMOGLOBIN A1 19980618 NAM • HEM 4548-4 HEMOGLOBIN A1C/HEMOGLOBIN.TOTAL SFR PT BLD QN HGB;HAEMOGLOBIN;GLYCATED;GLYCOSYLATED 20000322 MIN A02540 16500 QU60433 • HEM 17855-8 HEMOGLOBIN A1C/HEMOGLOBIN.TOTAL SFR PT BLD QN CALCULATED HGB;HAEMOGLOBIN;GLYCATED;GLYCOSYLATED20000322 MIN • HEM 4549-2 HEMOGLOBIN A1C/HEMOGLOBIN.TOTAL SFR PT BLD QN ELECTROPHORESIS HGB;HAEMOGLOBIN;GLYCATED;GLYCOSYLATED 19980618 NAM A02540 16500 QU60433 • HEM 17856-6 HEMOGLOBIN A1C/HEMOGLOBIN.TOTAL SFR PT BLD QN HPLC HGB;HAEMOGLOBIN;GLYCATED;GLYCOSYLATED 20000322 MIN • HEMOGLOBIN.GLYCATED PT BLD QN HEMOGLOBIN GLYCOSYLATED;GLYCOHEMOGLOBIN; GLYCOHAEMOGLOBIN;HAEMOGLOBIN. GLYCATED;HAEMOGLOBIN GLYCOSYLATED 19980618 DEL 16500 MSH94D0 06454 The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 18 NDC: 46193073801 NDC: 00686027720 NDC: 00555046506 NDC: 48695117305 NDC: 00555046505 NDC: 00047007032 NDC: 00555046502 NDC: 00047007024 NDC: 00054475833 NDC: 00223255002 NDC: 00054475831 NDC: 00223255001 NDC: 00054475825 NDC: 00364075690 NDC: 54441019750 NDC: 00364075602 NDC: 54441019725 NDC: 00364075601 NDC: 54441019715 NDC: 52953000304 NDC: 54441019711 NDC: 00378018210 NDC: 54441019710 NDC: 00378018201 NDC: 00182175810 NDC: 51432097106 NDC: 00182175801 NDC: 00677104110 NDC: 00228232796 NDC: 00677104105 NDC: 00228232710 NDC: 00677104101 NDC: 00046042199 NDC: 54569055650 NDC: 00046042198 NDC: 00102333502 NDC: 00046042195 NDC: 46193073810 NDC: 00603548921 NDC: 46193073805 NDC: 53258015313 NDC: 52544030505 NDC: 54269010901 NDC: 52544030501 NDC: 51813007299 NDC: 53633032116 NDC: 51813007290 NDC: 53633032110 NDC: 51813007260 NDC: 12071044010 NDC: 53492301303 NDC: 54441004350 NDC: 53492301302 NDC: 54441004325 NDC: 53492301301 NDC: 54441004310 NDC: 00591555404 NDC: 54441004305 NDC: 00591555401 NDC: 54441004301 NDC: 52493063960 The Road to Semantic Interoperability NDC: 49884010610 NDC: 00839711420 NDC: 49884010605 NDC: 51285032190 NDC: 51285032160 NDC: 52985003606 NDC: 52985003601 NDC: 51285032112 NDC: 51285032109 NDC: 51285032105 NDC: 51285032102 NDC: 00555036505 NDC: 00555036502 NDC: 00308627099 NDC: 00308627060 NDC: 00308627030 NDC: 54697006301 NDC: 00302573210 NDC: 11146091299 NDC: 00719179413 NDC: 00719179410 NDC: 00904041180 NDC: 00904041160 NDC: 00894633104 NDC: 00894633103 NDC: 00894633102 NDC: 00894633101 NDC: 10647042101 NDC: 50111046707 NDC: 50111046703 NDC: 50111046701 NDC: 52555001010 NDC: 52555001001 March 2013 NDC: 00093060010 NDC: 00093060001 NDC: 50053310901 NDC: 00615256113 NDC: 47679070204 NDC: 40039006001 NDC: 47679070201 NDC: 00025090152 NDC: 46703009410 NDC: 00025090131 NDC: 46703009401 NDC: 47202255103 NDC: 52584018410 NDC: 47202255101 NDC: 00363690810 NDC: 12027008902 NDC: 53489045101 NDC: 12027008901 NDC: 00157052610 NDC: 53487014510 NDC: 00157052601 NDC: 00781134413 NDC: 00054875825 NDC: 00781134410 NDC: 00005310931 NDC: 00781134401 NDC: 00005310923 NDC: 53978003410 NDC: 00071007024 NDC: 00117134405 NDC: 00046042191 NDC: 00117134401 NDC: 00046042181 NDC: 51316009004 NDC: 00046042180 NDC: 11146094210 NDC: 00046042162 NDC: 52544030551 NDC: 00046042161 NDC: 52544030510 NDC: 00046042160 NDC: 00839711416 NDC: 00814644630 NDC: 00536430910 NDC: 54697006305 NDC: 00536430905 NDC: 54697006304 NDC: 00536430901 NDC: 54697006303 NDC: 35470050801 NDC: 54697006302 NDC: 00143150225 NDC: 49884010601 NDC: 51608042104 NDC: 00686018201 NDC: 51608042102 NDC: 51079027760 NDC: 10465042109 NDC: 00349845190 NDC: 00721002301 NDC: 00349845110 NDC: 54421011001 Content in the presentation may be re-used so long as page 19 attribution is provided. All rights reserved. NDC: 00349845101 NDC: 19458042007 NDC: 50053310902 NDC: 19458042001 NDC: 00179038202 NDC: 00179038201 NDC: 54738046703 NDC: 54738046701 NDC: 00182181289 NDC: 00603548932 NDC: 46193073801 NDC: 00686027720 NDC: 00555046506 NDC: 48695117305 NDC: 00555046505 NDC: 00047007032 NDC: 00555046502 NDC: 00047007024 NDC: 00054475833 NDC: 00223255002 NDC: 00054475831 NDC: 00223255001 NDC: 00054475825 NDC: 00364075690 NDC: 54441019750 NDC: 00364075602 NDC: 54441019725 NDC: 00364075601 NDC: 54441019715 NDC: 52953000304 NDC: 54441019711 NDC: 00378018210 NDC: 54441019710 NDC: 00378018201 NDC: 00182175810 NDC: 51432097106 NDC: 00182175801 NDC: 00677104110 NDC: 00228232796 NDC: 00677104105 NDC: 00228232710 NDC: 00677104101 NDC: 00046042199 NDC: 54569055650 NDC: 00046042198 NDC: 00102333502 NDC: 00046042195 NDC: 46193073810 NDC: 00603548921 NDC: 46193073805 NDC: 53258015313 NDC: 52544030505 NDC: 54269010901 NDC: 52544030501 NDC: 51813007299 NDC: 53633032116 NDC: 51813007290 NDC: 53633032110 NDC: 51813007260 NDC: 12071044010 NDC: 53492301303 NDC: 54441004350 NDC: 53492301302 NDC: 54441004325 NDC: 53492301301 NDC: 54441004310 NDC: 00591555404 NDC: 54441004305 NDC: 00591555401 NDC: 54441004301 NDC: 52493063960 The Road to Semantic Interoperability NDC: 49884010610 NDC: 00839711420 NDC: 49884010605 NDC: 51285032190 NDC: 51285032160 NDC: 52985003606 NDC: 52985003601 NDC: 51285032112 NDC: 51285032109 NDC: 51285032105 NDC: 51285032102 NDC: 00555036505 NDC: 00555036502 NDC: 00308627099 NDC: 00308627060 NDC: 00308627030 NDC: 54697006301 NDC: 00302573210 NDC: 11146091299 NDC: 00719179413 NDC: 00719179410 NDC: 00904041180 NDC: 00904041160 NDC: 00894633104 NDC: 00894633103 NDC: 00894633102 NDC: 00894633101 NDC: 10647042101 NDC: 50111046707 NDC: 50111046703 NDC: 50111046701 NDC: 52555001010 NDC: 52555001001 March 2013 NDC: 00093060010 NDC: 00093060001 Propranolol 10Mg Tab NDC: 50053310901 NDC: 00615256113 NDC: 47679070204 NDC: 40039006001 NDC: 47679070201 NDC: 00025090152 NDC: 46703009410 NDC: 00025090131 NDC: 46703009401 NDC: 47202255103 NDC: 52584018410 NDC: 47202255101 NDC: 00363690810 NDC: 12027008902 NDC: 53489045101 NDC: 12027008901 NDC: 00157052610 NDC: 53487014510 NDC: 00157052601 NDC: 00781134413 NDC: 00054875825 NDC: 00781134410 NDC: 00005310931 NDC: 00781134401 NDC: 00005310923 NDC: 53978003410 NDC: 00071007024 NDC: 00117134405 NDC: 00046042191 NDC: 00117134401 NDC: 00046042181 NDC: 51316009004 NDC: 00046042180 NDC: 11146094210 NDC: 00046042162 NDC: 52544030551 NDC: 00046042161 NDC: 52544030510 NDC: 00046042160 NDC: 00839711416 NDC: 00814644630 NDC: 00536430910 NDC: 54697006305 NDC: 00536430905 NDC: 54697006304 NDC: 00536430901 NDC: 54697006303 NDC: 35470050801 NDC: 54697006302 NDC: 00143150225 NDC: 49884010601 NDC: 51608042104 NDC: 00686018201 NDC: 51608042102 NDC: 51079027760 NDC: 10465042109 NDC: 00349845190 NDC: 00721002301 NDC: 00349845110 NDC: 54421011001 Content in the presentation may be re-used so long as page 20 attribution is provided. All rights reserved. NDC: 00349845101 NDC: 19458042007 NDC: 50053310902 NDC: 19458042001 Designing for Interoperability Ability to Interoperate High Low The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 21 Localization Concept: The Virtual Boundary • External Standards •Legacy Applications Organizational Compliance - Organizational Semantics (Info Model) - Integration compliance (SOA services) - Governance The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 22 Semantic Support in HIT Standards… V2 Msgs V3 Msgs CDA Services Data Descriptions are Explicit • Behavior is Explicit • • Has medico-legal Meaning • • Provides clinical context • Allows customization/extensibility • • • • Allows for ontologically-based data Allows for non-ontologically-based data Provides for self-discovered semantics The Road to Semantic Interoperability March 2013 • Content in the presentation may be re-used so long as attribution is provided. All rights reserved. * Dependent upon implementation or us The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 24 Enterprise Architecture The Pragmatists View • Done correctly, EA establishes the target to which the organization is heading • EA models typically address business, information, and technology views • An EA program establishes a “source of truth” for standards • More importantly, the EA specifies not just stds & technologies are to be used, but how • EA is only effective when supporting and engaging stakeholders, and where the program has governance authority The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 25 EASE ON DOWN THE ROAD… The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 3/20/2013 26 First Stop: Version Management Why include it? How to approach it: • EVERYTHING is going to change, that includes: • Apply a formal, rigorous approach to version management (perhaps the MOST IMPORTANT element in managing continuous change) • Every time something is deployed that is changed, version it. Not just software, but also information assets • Understand and get religious about dependency management • Be very cautions and judicious about any “physical” deletes – The data and the structures containing it – The expectations of the data – The terminology(ies) – The interfaces – The applications The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 01/01/2011 27 Second Stop: Legacy Enablement Why include it? How to approach it: • Investment in current systems and infrastructure is too significant to ignore or to start ‘green field’ • Use EA to determine system role, especially where its data is the “authoritative source” • Neither business case nor organizational ‘will’ (not to mention budget) exists to replace all of your existing IT • New infrastructure will need to more effectively interact with legacy The Road to Semantic Interoperability • Review information exchange needs within and outside of the organization • Map current messaging and interfaces into logical information constructs • “Service-enable” the legacy by adding request/response interfaces carrying structured payload (e.g., SOA) March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 01/01/2011 28 Third Stop: Foster Peaceful Co-Existence of Structured and Unstructured Data Why include it? Key Solution Considerations • Health today significantly depends on unstructured data • Consider structured information with the end objectives in mind (e.g., clinical quality, process improvement, epidemiology) • Huge push worldwide to move to structured information (SNOMED, ICD10/11) • Structured information offers many benefits (e.g., analytics, comparability, decision-support), but • TCO of structured information must consider organizational adoption (staff training, quality/oversight processes, etc.) • User experience is of paramount importance • Structured information does not come ‘free’ The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 01/01/2011 29 Fourth Stop: Metadata Management Why include it? Key Solution Considerations • As information interchange increases, data content will increasingly come from multiple sources • Note that metadata describes what is in the “container”, allowing the recipient to comprehend what they received • Today, metadata management is done implicitly based upon source (e.g. “This record is from the xxxx General Hospital”) • Use models based on ontologically as the basis for your metadata • We will never see ubiquity of data representation and use (e.g., variance is here to stay) The Road to Semantic Interoperability • Version management applies to metadata too • Allow for discovery via use of [public] registries March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 01/01/2011 30 Final Stop: Context Management Why include it? Key Solution Considerations • Information structures are an implicit component of healthcare and used in business terms today (e.g., “Blood Pressure Reading”, “Discharge Summary”, etc.) • Thoughtful units of composition • As [clinical] information flow grows, receipt of standalone data items is insufficient • Applied metadata management • Ability for constructs to retain medico-legal meaning • Seek out existing sources for representing contextual elements (structure-oriented standards) • Current transport model either predicates that the sender “knows” what the recipient will need, or that the recipient will “know” what to ask for The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 01/01/2011 31 RELEVANT WORK The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. 3/20/2013 32 Asset Inventory Asset What it is for Common Terminology Service (CTS II) Establish a standards-based way of interacting with and managing complex coding systems and vocabularies. Defines interfaces for the storage, management, and maintenance of terminology API for Knowledgebases (API4KB) Integrates various reasoning systems in a loosely coupled, hybrid environment, to allow complex sets of services to be developed and combined. Defines APIs for Parsing,. Reasoning, Persistent Storage, Queries, Rules and query of metadata to understand the capabilities of the service offering behind the API Ontology Definition Metamodel (ODM) Bridges traditional UML modeling and software engineering popular ontology languages, (RDF, RDF /S), the Web Ontology Language (OWL), Topic Maps (TM), and Common Logic (CL). Supports interchange and management of vocabularies and ontologies with MOF/XMI tools and repositories, as well profiles and graphical representation for developing vocabularies and ontologies in UML tools Retrieve Locate Update Service (RLUS) To manage location and retrieval of healthcare content Defines an abstract service interface for create, read, update, delete functions hDATA RESTful Transport Specification REST binding for data retrieval using SOA (RLUS for REST) Provides a REST binding for create, read, update, delete functions Clinical Decision Support Service (DSS) To analyze patient data / assess knowledge rules. Establish interface for passing in patient data and returning relevant clinical protocol hDATA Record Format Specification A hierarchical format with metadata tagging for organizing / representing [clinical] data Provides a superstructure into which [clinical] data can be slotted and transported. The Road to Semantic Interoperability What it does March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 33 Acknowledgements Several of these slides have been shamelessly borrowed from the Healthcare Services Specification Project, a joint standards collaboration between OMG and Health Level Seven (HL7) Their slideware licensing allows reuse with attribution. This is attribution The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 34 Thank you! Ken Rubin Chief Architect, Federal Healthcare ken.rubin@hp.com The Road to Semantic Interoperability March 2013 Content in the presentation may be re-used so long as attribution is provided. All rights reserved. page 35