challenges of interoperability using hl7 v3 in czech healthcare miroslav nagy, petra preckova, libor...
TRANSCRIPT
Challenges of Interoperability Challenges of Interoperability UsingUsingHL7 v3 in Czech HealthcareHL7 v3 in Czech Healthcare
Miroslav NAGYMiroslav NAGY, Petra PRECKOVA, Libor SEIDL and Jana , Petra PRECKOVA, Libor SEIDL and Jana ZVAROVAZVAROVACenter of Biomedical InformaticsCenter of Biomedical InformaticsDepartment of Medical InformaticsDepartment of Medical InformaticsInstitute of Computer Science AS CR, v.v.i.Institute of Computer Science AS CR, v.v.i.Prague, Czech RepublicPrague, Czech Republic
OutlineOutline MotivationMotivation
Processing medical reports – “synonyms” cause Processing medical reports – “synonyms” cause ambiguitiesambiguities
Interoperability initiatives in Czech healthcareInteroperability initiatives in Czech healthcare Materials and MethodsMaterials and Methods
Standardization of clinical termsStandardization of clinical terms International communication standardInternational communication standard Two EHR systems were used (commercial and Two EHR systems were used (commercial and
academic)academic) ResultsResults
Mapping of data model to classification systemsMapping of data model to classification systems Enabling semantic interoperability among EHR systems Enabling semantic interoperability among EHR systems
via SIP (Semantic Interoperability Platform)via SIP (Semantic Interoperability Platform) ConclusionConclusion
MotivationMotivation
Data exchange via free-text Data exchange via free-text reportsreports
Narrative free-text reports – prevailing form of Narrative free-text reports – prevailing form of data used for communicationdata used for communication
Reports comprise of many specific Reports comprise of many specific abbreviationsabbreviations and and synonymssynonyms, which are not , which are not consistent even in one hospitalconsistent even in one hospital
Free-text reports are always typed again and Free-text reports are always typed again and again – many again – many typostypos
Different understanding of terms can cause Different understanding of terms can cause wrong decision and thus threaten the patientwrong decision and thus threaten the patient
This practice should be stopped.This practice should be stopped.
Activities in Czech Activities in Czech healthcarehealthcare
One of early interoperability initiatives was IZIPOne of early interoperability initiatives was IZIP Main idea:Main idea:
Citizens should have a centralized copy of their Citizens should have a centralized copy of their healthcare documentation – the “Internet Health Book”healthcare documentation – the “Internet Health Book”
Main problem:Main problem: Data were stored in free text form, which hindered their Data were stored in free text form, which hindered their
usage for processing by computers (various analyses, usage for processing by computers (various analyses, decision support, etc.)decision support, etc.)
National communication standardNational communication standard DASTA bound to national nomenclature “National DASTA bound to national nomenclature “National
Code-list of Laboratory Items”Code-list of Laboratory Items” Development supported mainly by the Ministry of Development supported mainly by the Ministry of
Health of CRHealth of CR
Standardizing clinical Standardizing clinical contentcontent
We have to agree on what we are talking about.We have to agree on what we are talking about. Stating minimal data modelsStating minimal data models
At our department a Minimal Data Model of Cardiology At our department a Minimal Data Model of Cardiology was prepared - MDMCwas prepared - MDMC
Creating a mapping to coding systemsCreating a mapping to coding systems Make the model multi-lingual – just translate the Make the model multi-lingual – just translate the
coding systemcoding system Fix the meaning of the concepts in the modelFix the meaning of the concepts in the model
Despite of the amount of incompatible coding Despite of the amount of incompatible coding systems (circa 100) there is a tool (UMLS systems (circa 100) there is a tool (UMLS Knowledge Source Server) that enables to Knowledge Source Server) that enables to search a term in all coding systems at once.search a term in all coding systems at once.
Supporting interoperabilitySupporting interoperability Standardized clinical contentStandardized clinical content
Communication standardCommunication standard
Electronic health record (EHR)Electronic health record (EHR)
What have we done?What have we done? We proposed a Semantic Interoperability Platform We proposed a Semantic Interoperability Platform
(SIP) based on HL7 v3 messaging.(SIP) based on HL7 v3 messaging. We conducted an experiment were we exchanged We conducted an experiment were we exchanged
data among different Czech EHR systems using HL7 data among different Czech EHR systems using HL7 v3 messages. This experiment was a pilot v3 messages. This experiment was a pilot implementation of the SIP.implementation of the SIP.
Materials and MethodsMaterials and Methods
EHR systemsEHR systems WinMedicalc 2000WinMedicalc 2000
Medium sized commercial modular Hospital Medium sized commercial modular Hospital Information System (HIS) used among others in Information System (HIS) used among others in municipal hospital in Caslav, which participated in our municipal hospital in Caslav, which participated in our projectproject
We have chosen this HIS because it emphasizes the We have chosen this HIS because it emphasizes the advantage of structured EHR over the documentation advantage of structured EHR over the documentation based on free-texts.based on free-texts.
ADAMEKjADAMEKj Developed at our department (EuroMISE Center)Developed at our department (EuroMISE Center) Fully structured EHR system with the aim to collect Fully structured EHR system with the aim to collect
outpatient genetic and clinical data in cardiologyoutpatient genetic and clinical data in cardiology Clinical parts of domain models of both EHR Clinical parts of domain models of both EHR
systems are based on MDMCsystems are based on MDMC
Coding systemsCoding systems
In our solution we used various coding systemsIn our solution we used various coding systems ICD-10 – for terms concerning diagnosisICD-10 – for terms concerning diagnosis LOINC – laboratory resultsLOINC – laboratory results SNOMED CT – all other termsSNOMED CT – all other terms
For terms searching we used the UMLS For terms searching we used the UMLS Knowledge Source Server Knowledge Source Server (http://umlsks.nlm.nih.gov/)(http://umlsks.nlm.nih.gov/)
Communication StandardCommunication Standard Data transfers among EHR systems in our Data transfers among EHR systems in our
solution is performed by the HL7 v3 standardsolution is performed by the HL7 v3 standard It is widely accepted for the electronic interchange of It is widely accepted for the electronic interchange of
healthcare informationhealthcare information RIM, DMIM, RMIM, Storyboards, Implementation RIM, DMIM, RMIM, Storyboards, Implementation
Technology Specification (XML ITS)Technology Specification (XML ITS) The national standard DASTA has several The national standard DASTA has several
shortcomingsshortcomings Fixed to Czech healthcare onlyFixed to Czech healthcare only No tooling supportNo tooling support No support for other coding systems than NCLINo support for other coding systems than NCLI Developed in a bottom-up manner resulting in no Developed in a bottom-up manner resulting in no
development methodologydevelopment methodology
ResultsResults
Mapping the MDMC…Mapping the MDMC… MDMC consists of approx. 150 attributes, their MDMC consists of approx. 150 attributes, their
mutual relations, units and data integrity mutual relations, units and data integrity restrictionsrestrictions
85% of terms were found in at least one 85% of terms were found in at least one classification systemclassification system
Was the mapping easy? Was the mapping easy? The terms could be The terms could be classified into four groups:classified into four groups: Trouble free termsTrouble free terms Several possibilities of mapping to different synonymsSeveral possibilities of mapping to different synonyms Terms with too small or to big granularityTerms with too small or to big granularity No matching term found in any of available classification No matching term found in any of available classification
systemssystems Now we have a domain model with well defined Now we have a domain model with well defined
semantics.semantics.
Mapping the MDMC…Mapping the MDMC…
Description of encoded MDMC conceptDescription of encoded MDMC concept CodeCode Coding systemCoding system
Treatment for ischemic heart disease (situation)Treatment for ischemic heart disease (situation) 308068007308068007 SNOMED CT SNOMED CT
Left ventricular hypertrophy (disorder)Left ventricular hypertrophy (disorder) 5582700555827005 SNOMED CT SNOMED CT
Diabetes mellitus (disorder)Diabetes mellitus (disorder) 7321100973211009 SNOMED CT SNOMED CT
Swelling or edema (finding)Swelling or edema (finding) 248477007248477007 SNOMED CTSNOMED CT
Syncope (disorder)Syncope (disorder) 271594007271594007 SNOMED CTSNOMED CT
Measurement of the breath frequency in one minuteMeasurement of the breath frequency in one minute 9279-19279-1 LOINCLOINC
Measurement of intravascular diastolic pressureMeasurement of intravascular diastolic pressure 8462-48462-4 LOINCLOINC
Amount of proteins in blood sampleAmount of proteins in blood sample 2885-22885-2 LOINCLOINC
Sequelae of cerebrovascular diseaseSequelae of cerebrovascular disease I61I61 ICD-10ICD-10
Angina PectorisAngina Pectoris I20I20 ICD-10ICD-10
SIP TermsSIP Terms Local Information Model (LIM)Local Information Model (LIM)
A model derived from HL7 RIM classes – similar to D-MIMA model derived from HL7 RIM classes – similar to D-MIM Describes the domain (MDMC) of given EHR system Describes the domain (MDMC) of given EHR system
(WinMedicalc 2000, ADAMEKj)(WinMedicalc 2000, ADAMEKj) The MDMC mapping was used to fill in the “code” attribute The MDMC mapping was used to fill in the “code” attribute
of each LIM classof each LIM class LIM templateLIM template
Describes integral part of EHR (physical examination, lab. Describes integral part of EHR (physical examination, lab. results, etc.) – similar to HMDresults, etc.) – similar to HMD
Consists of classes from LIMConsists of classes from LIM XML Schema representationXML Schema representation
LIM MessageLIM Message XML document containing actual data from EHRXML document containing actual data from EHR Valid document according to the given LIM templateValid document according to the given LIM template
HL7 v3 messageHL7 v3 message Instance of an HL7 v3 message typeInstance of an HL7 v3 message type
How does it work?How does it work? LIM filler module – each EHR system has its own.LIM filler module – each EHR system has its own. HL7 v3 broker – configurable for each EHR system.HL7 v3 broker – configurable for each EHR system. SIP Schema:SIP Schema:
ConclusionConclusion
Why did we do it that way?Why did we do it that way? To lessen the stress and fear of developers interested in HL7 v3.To lessen the stress and fear of developers interested in HL7 v3. We divided the work between HL7 v3 experts and HIS experts.We divided the work between HL7 v3 experts and HIS experts. Consequently, our solution is not depending on HL7 v3, it can be Consequently, our solution is not depending on HL7 v3, it can be
changed via a new “Broker” transforming component.changed via a new “Broker” transforming component.
ConclusionConclusion Incorporation of international classification systems Incorporation of international classification systems
is the first and essential step towards is the first and essential step towards interoperability.interoperability.
Standardized clinical content would result in more Standardized clinical content would result in more reliable information stored in EHR, which will lead to reliable information stored in EHR, which will lead to less error-prone and safer care for patients.less error-prone and safer care for patients.
We made a proposal and a testing implementation We made a proposal and a testing implementation of SIP and were able to exchange data between two of SIP and were able to exchange data between two heterogeneous EHR systems.heterogeneous EHR systems.
A real challenge would be a national-wide A real challenge would be a national-wide implementation of SIP.implementation of SIP. Enable/support medical tourismEnable/support medical tourism Open the market for HIS solutions from abroadOpen the market for HIS solutions from abroad
It is necessary to translate classification systems It is necessary to translate classification systems (SNOMED CT) to Czech language(SNOMED CT) to Czech language
Thank you for your Thank you for your attention.attention.
The work was supported by the project no. The work was supported by the project no. 1M060141M06014 of the Ministry of Education, Youth and of the Ministry of Education, Youth and Sports CR.Sports CR.
Mgr. Miroslav NagyMgr. Miroslav Nagy
Center of Biomedical Center of Biomedical InformaticsInformatics
Pod Vodarenskou vezi 2Pod Vodarenskou vezi 2
Prague 8Prague 8
182 07182 07
Email: Email: [email protected]@euromise.cz
Tel. +420 266 052 081Tel. +420 266 052 081