S18: Terminologies and Meaningful Use: the Role of NLM-S. Srinivasan (Part 1)
Content
UMLS Terminology Services (Knowledge Sources Server)
Suresh Srinivasan, M.S. National Library of Medicine National Library of Medicine
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Vocabulary standards for Meaningful use
• SNOMED CT or ICD‐9‐CM for up‐to‐date problem SNOMED CT or ICD 9 CM for up to date problem list of diagnoses • RxNorm drug terminologies for Stage I objectives; RxNorm drug terminologies for Stage I objectives; bridge to full RxNorm adoption • LOINC for lab tests (should be able to receive LOINC for lab tests (should be able to receive messages with LOINC codes and populate record) • UMLS (source of multiple terminologies in unified UMLS (source of multiple terminologies in unified format, cross‐mapping for interoperability)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Terminology products from the NLM Terminology products from the NLM
• Unified Medical Language System (UMLS)
– Bi‐annual releases – Download or DVD (by request)
• RxNorm
– Monthly releases, weekly updates – Synchronized with the UMLS twice a year
• SNOMED CT (U.S. National Release Center)
– International core release – U.S. Extension (planned) U.S. Extension (planned)
• CORE problem list subset of SNOMED CT • Medical Subject Headings (MeSH)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Terminology projects simplified workflow Terminology projects – simplified workflow
Preprocessing, data inflow
On a schedule set by policy
Data Repository
Editing, Q/A
Production
Produced terminology artifacts
Subject Matter Experts (SME)
Browse, Web, subset, publish, bundle, etc Synchronization
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
UMLS Knowledge Sources Server (KSS) UMLS Knowledge Sources Server (KSS)
• First Web version in 1995 First Web version in 1995 • Online access to the UMLS Metathesaurus, Semantic Network, SPECIALIST lexicon and Semantic Network SPECIALIST lexicon and Information Sources Map • A A command line tool for scripted access d li lf i d (implemented in C using TCP sockets) – simple, string‐based API i l i b d API
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
KSS (continued) KSS (continued)
• Port to Java in 2003 Port to Java in 2003 • Richer API to support more functionality and access to all UMLS data elements access to all UMLS data elements • Web, RMI and XML/socket APIs • More object‐oriented • Lexical view of the data
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
KSS (continued) KSS (continued)
• 2006 ‐ Some interactive technologies 2006 Some interactive technologies introduced, e.g., Ajax hierarchies and auto‐ suggest • Portlets to customize user interface (uPortal) • W bS i Web Services API API • More atomic view • Authentication using CAS (single sign‐on)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
MetamorphoSys
• Java application to install customize and Java application to install, customize and browse subsets of the UMLS • Generates UMLS output in variety of formats Generates UMLS output in variety of formats • Download or on DVD – mmsys.zip • Default subsets, e.g., level 0 sources only • Masks the complexity of our distribution p y format (.nlm)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Where we are going with KSS Where we are going with KSS
• Transition from research to production group Transition from research to production group within NLM (2009) • Simultaneously being re‐engineered to address Simultaneously being re engineered to address issues with security, memory, scalability, performance and accessibility performance and accessibility • Building infrastructure and capacity to support related applications reliably related applications reliably • Modules for browser, authentication, downloads, account creation, annual report submission account creation annual report submission
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
UMLS Terminology Services (portal) UMLS Terminology Services (portal)
• Authenticated browse access to UMLS, individual ut e t cated b o se access to U S, d dua terminologies (SNOMED CT) • Download of built artifacts (UMLS, SNOMED CT, RxNorm, CORE problem list subset, etc.) • Web services API (legacy and newer) • Programmatic authentication for UMLS/SNOMED licensees (portal users) using API • Request submission for SNOMED CT • Single portal to access all these resources
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Upgrading underlying technologies Upgrading underlying technologies
• Data center (cooling, UPS), Storage (dual controllers, RAID, fast recovery, expandable to ~14PB, 10Gb Ethernet) Virtualization/cloud, Linux, BigIP, 3DNS, offsite • Virtualization/cloud, Linux, BigIP, 3DNS, offsite • Oracle RAC – scalability and availability • Spring application framework, Maven for builds • Hibernate for ORM, Lucene + LVG • Google Web Toolkit (GWT), CAS (single sign‐on) • Stateless server for better scalability Stateless server for better scalability • Maintainability, security, patching • Official CA signed certificates g
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Some ideas for the future Some ideas for the future
• Matching your local terminology to UMLS – atc g you oca te o ogy to U S flexible techniques (similar to LSVT) g • Additional default‐subsets, e.g., clinical use, meaningful use mandated terminologies • Default subsets pre‐generated for download • Customization (user profiles) • New versions of API with richer functionality – true terminology service; RESTful • Internal for NLM – better reporting, statistics
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
More ideas.. More ideas..
• Connections from IHTSDO workbench for Co ect o s o S O o be c o modelers to link to UMLS content • Local installations of the browser and related systems • Better integration with MetamorphoSys • Accessibility improvements (section 508) and mobile versions • Integrate with UMLS archive (1990‐2010) • Improved documentation, training materials
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Your ideas welcome! Your ideas welcome!
Contact information: [email protected] (1‐888‐FIND‐NLM) [email protected] (301‐435‐3174)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Principles
• • • • • Represent data structures as objects with their natural connections with each other. Represent true “Metathesaurus” data structures. Represent true “Source” data structures. p Achieve more explicit semantics (SourceTermType, SourceAttributeName, etc) – specifically to support ontology formalisms. Achieve harmonization across sources while allowing sources to retain Achieve harmonization across sources while allowing sources to retain their own explicit definitions of things.
– Shared TTY, RELA, ATN
• • • •
Full semantic model with “extensibility” mechanisms y Represent all known data structures from all sources in UMLS Clearly delineate “first class” properties from “connected” attributes Well defined fields/properties of each object. Well defined fields/properties of each object
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Important Questions Important Questions
• • What are the fundamental needs of the common object model? Is the common model expected to accommodate all of the needs of all of the UMLS applications, or do we plan on having a "core" model that has the intersection of needed functionality and is expanded upon by each individual application, according to its needs? Do we expect that this "core" model will be exposed to the public? p p p Is this common domain model a conceptual object/service model? Is the common domain model also a model for specific data representations? As part of the development of a conceptual model, will we make implementation decisions, like using interfaces vs. classes? How much code do we intend to write? Do we intend this like using interfaces vs classes? How much code do we intend to write? Do we intend this code to be the “real” code for the future model? Does the common domain model include the notion of shared or common "services"? What is criteria for deciding what is a "first order" attribute of a domain model object vs. a generically connected Attribute? What auxiliary data is needed in addition to the core domain objects? Are there objects to represent this data, or just files/tables? For example, consider atoms_ui table that tracks AUI assignments over time.
• • •
• • •
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
“Kinds” of Objects Kinds of Objects
• • • • • Content Metadata History i Statistics Auxiliary
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Content
• Stable primary key UI – maintained across time – p y y corresponding to identity fields. • Full AIM Semantics • Metathesaurus Data
– Concept – Lexical Info Lexical Info – String Info
• Source Data
– Atom, Attribute, Relation, Tree Position
• Mappings, Subsets, Content Views, Value Sets • Atom Cluster, SourceAtomCluster
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Meta Content Diagram Meta Content Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Source Content Diagram Source Content Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Atom Cluster Inheritance Diagram Atom Cluster Inheritance Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Clusters Venn Diagram (1) Clusters Venn Diagram (1)
LUI1 LUI2
SUI1
SUI3
SUI4 split across CUI1,CUI2 LUI3 split across CUI1,CUI2 LUI1,LUI2 merged in CUI1 LUI1 LUI2 merged in CUI1 LUI2,LUI3 merged in CUI2
SUI2
SUI4 CUI1 LUI3 CUI2
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Clusters Venn Diagram (2) Clusters Venn Diagram (2)
SDUI1 LUI1 LUI2 CODE1 LUI3
LUI7 CODE2 CODE3 LUI6 LUI5 CODE4 L2 CODE5 CODE6 LUI4
CODE1
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
RRF Limitations (1) RRF Limitations (1)
• No explicit model of “concept” or corresponding objects for SCUI, SDUI, CODE, LUI, SUI CODE LUI SUI
– Leads to inflated representation of DA, MR, ST
• • • • • • • •
Conflation of “obsolete” and “suppressible” Lack of support for “AIM” semantics L k f t f “AIM” ti Definitions can only be connected to atoms Some statistical data misrepresented as content (e.g. MED<year> attributes in MRSAT) attributes in MRSAT) Unable to represent attribute referential integrity as a database foreign key (conflation of different “kinds” of attributes) Relations have confusing directionality (should be more explicit about Relations have confusing directionality (should be more explicit about “to” and “from”) SIB relations are inconsistent Identifiers in MRHIST are not resovled to UMLS UIs. Identifiers in MRHIST are not resovled to UMLS UIs
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
RRF Limitations (2) RRF Limitations (2)
• Hierarchies are only atom‐based (e.g. no STYPE in y ( g MRHIER) • Confusing/awkward representation of Subset information. Conflation of Subset with Concept. information Conflation of Subset with Concept • Confusing/awkward representation of content view member attributes view member attributes • Conflation of Content View and Concept. • Conflation of MapSet and Concept. p p • Metadata model is very limited and awkward to extend.
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Open Sources/Tools Open Sources/Tools
• Unifying tooling infrastructure across UMLS Unifying tooling infrastructure across UMLS and terminology projects • Modern tools and technologies Modern tools and technologies
– SVN, Maven –S i L Spring, Lucene – GWT, CAS – Hibernate, Lime Survey
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
26
Modules
• SVN
– File based configuration management tool – Integrates well with Eclipse (development tool) Integrates well with Eclipse (development tool)
• Maven
– Project Management Tool – Automatic build/deploy/test projects – Automatic project dependency management and update (via repositories)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
27
Modules (contd.) Modules (contd.)
• Spring: application framework (IOC, AOP) Sp g: app cat o a e o ( OC, O )
– – – – Application is highly configurable Declarative programming Reduces code base A number of pre‐built modules: Security, DOA, TM, MVC, Messaging… MVC Messaging
• Lucene: search engine
– Searches are serviced by Lucene (not Database) Searches are serviced by Lucene (not Database) – High‐performance, full featured and cross platform text search engine
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
28
Modules (contd.) Modules (contd.)
• GWT: Google Web Toolkit
– – – – – – – – – – Simple and powerful Web based GUI tools Built in support for AJAX Allows development in JAVA p Converts JAVA code to JAVA Script to run on client machine Multi‐browser compliant 508 Compliant (with modifications) p ( ) A single sign on system Provides security via pluggable backend security modules Provides security via pluggable backend security modules Easy to integrate applications Only the user and one and only one CAS module knows the passwords in an enterprise passwords in an enterprise
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
29
• CAS: Central Authentication System
Modules (contd.) Modules (contd.)
• Hibernate: ORM libraries be ate: O b a es
– Provides object relational mapping – Removes SQL coding – Declarative mapping of objects, associations, and constraints to the underlying database – Supports la y loading and various fetch strategies Supports lazy loading and various fetch strategies – Efficient with internal caching across calls – Supports query caches Supports query caches – QA by referential integrity
• Blocks bad data entering the system
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
30
Deployment Diagram Deployment Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
31
Work with IHTSDO workbench Work with IHTSDO workbench
SNOMED CT request submission workflow SNOMED CT request submission workflow UMLS content from the UTB/KSS Pubmed and literature searching b d d li hi Flexible searching with UMLS lexical tools Maintenance environment for UMLS Semantic Network • Collaborate on ICD10 mapping • • • • •
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Suresh Srinivasan, M.S. National Library of Medicine National Library of Medicine
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Vocabulary standards for Meaningful use
• SNOMED CT or ICD‐9‐CM for up‐to‐date problem SNOMED CT or ICD 9 CM for up to date problem list of diagnoses • RxNorm drug terminologies for Stage I objectives; RxNorm drug terminologies for Stage I objectives; bridge to full RxNorm adoption • LOINC for lab tests (should be able to receive LOINC for lab tests (should be able to receive messages with LOINC codes and populate record) • UMLS (source of multiple terminologies in unified UMLS (source of multiple terminologies in unified format, cross‐mapping for interoperability)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Terminology products from the NLM Terminology products from the NLM
• Unified Medical Language System (UMLS)
– Bi‐annual releases – Download or DVD (by request)
• RxNorm
– Monthly releases, weekly updates – Synchronized with the UMLS twice a year
• SNOMED CT (U.S. National Release Center)
– International core release – U.S. Extension (planned) U.S. Extension (planned)
• CORE problem list subset of SNOMED CT • Medical Subject Headings (MeSH)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Terminology projects simplified workflow Terminology projects – simplified workflow
Preprocessing, data inflow
On a schedule set by policy
Data Repository
Editing, Q/A
Production
Produced terminology artifacts
Subject Matter Experts (SME)
Browse, Web, subset, publish, bundle, etc Synchronization
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
UMLS Knowledge Sources Server (KSS) UMLS Knowledge Sources Server (KSS)
• First Web version in 1995 First Web version in 1995 • Online access to the UMLS Metathesaurus, Semantic Network, SPECIALIST lexicon and Semantic Network SPECIALIST lexicon and Information Sources Map • A A command line tool for scripted access d li lf i d (implemented in C using TCP sockets) – simple, string‐based API i l i b d API
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
KSS (continued) KSS (continued)
• Port to Java in 2003 Port to Java in 2003 • Richer API to support more functionality and access to all UMLS data elements access to all UMLS data elements • Web, RMI and XML/socket APIs • More object‐oriented • Lexical view of the data
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
KSS (continued) KSS (continued)
• 2006 ‐ Some interactive technologies 2006 Some interactive technologies introduced, e.g., Ajax hierarchies and auto‐ suggest • Portlets to customize user interface (uPortal) • W bS i Web Services API API • More atomic view • Authentication using CAS (single sign‐on)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
MetamorphoSys
• Java application to install customize and Java application to install, customize and browse subsets of the UMLS • Generates UMLS output in variety of formats Generates UMLS output in variety of formats • Download or on DVD – mmsys.zip • Default subsets, e.g., level 0 sources only • Masks the complexity of our distribution p y format (.nlm)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Where we are going with KSS Where we are going with KSS
• Transition from research to production group Transition from research to production group within NLM (2009) • Simultaneously being re‐engineered to address Simultaneously being re engineered to address issues with security, memory, scalability, performance and accessibility performance and accessibility • Building infrastructure and capacity to support related applications reliably related applications reliably • Modules for browser, authentication, downloads, account creation, annual report submission account creation annual report submission
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
UMLS Terminology Services (portal) UMLS Terminology Services (portal)
• Authenticated browse access to UMLS, individual ut e t cated b o se access to U S, d dua terminologies (SNOMED CT) • Download of built artifacts (UMLS, SNOMED CT, RxNorm, CORE problem list subset, etc.) • Web services API (legacy and newer) • Programmatic authentication for UMLS/SNOMED licensees (portal users) using API • Request submission for SNOMED CT • Single portal to access all these resources
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Upgrading underlying technologies Upgrading underlying technologies
• Data center (cooling, UPS), Storage (dual controllers, RAID, fast recovery, expandable to ~14PB, 10Gb Ethernet) Virtualization/cloud, Linux, BigIP, 3DNS, offsite • Virtualization/cloud, Linux, BigIP, 3DNS, offsite • Oracle RAC – scalability and availability • Spring application framework, Maven for builds • Hibernate for ORM, Lucene + LVG • Google Web Toolkit (GWT), CAS (single sign‐on) • Stateless server for better scalability Stateless server for better scalability • Maintainability, security, patching • Official CA signed certificates g
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Some ideas for the future Some ideas for the future
• Matching your local terminology to UMLS – atc g you oca te o ogy to U S flexible techniques (similar to LSVT) g • Additional default‐subsets, e.g., clinical use, meaningful use mandated terminologies • Default subsets pre‐generated for download • Customization (user profiles) • New versions of API with richer functionality – true terminology service; RESTful • Internal for NLM – better reporting, statistics
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
More ideas.. More ideas..
• Connections from IHTSDO workbench for Co ect o s o S O o be c o modelers to link to UMLS content • Local installations of the browser and related systems • Better integration with MetamorphoSys • Accessibility improvements (section 508) and mobile versions • Integrate with UMLS archive (1990‐2010) • Improved documentation, training materials
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Your ideas welcome! Your ideas welcome!
Contact information: [email protected] (1‐888‐FIND‐NLM) [email protected] (301‐435‐3174)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Principles
• • • • • Represent data structures as objects with their natural connections with each other. Represent true “Metathesaurus” data structures. Represent true “Source” data structures. p Achieve more explicit semantics (SourceTermType, SourceAttributeName, etc) – specifically to support ontology formalisms. Achieve harmonization across sources while allowing sources to retain Achieve harmonization across sources while allowing sources to retain their own explicit definitions of things.
– Shared TTY, RELA, ATN
• • • •
Full semantic model with “extensibility” mechanisms y Represent all known data structures from all sources in UMLS Clearly delineate “first class” properties from “connected” attributes Well defined fields/properties of each object. Well defined fields/properties of each object
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Important Questions Important Questions
• • What are the fundamental needs of the common object model? Is the common model expected to accommodate all of the needs of all of the UMLS applications, or do we plan on having a "core" model that has the intersection of needed functionality and is expanded upon by each individual application, according to its needs? Do we expect that this "core" model will be exposed to the public? p p p Is this common domain model a conceptual object/service model? Is the common domain model also a model for specific data representations? As part of the development of a conceptual model, will we make implementation decisions, like using interfaces vs. classes? How much code do we intend to write? Do we intend this like using interfaces vs classes? How much code do we intend to write? Do we intend this code to be the “real” code for the future model? Does the common domain model include the notion of shared or common "services"? What is criteria for deciding what is a "first order" attribute of a domain model object vs. a generically connected Attribute? What auxiliary data is needed in addition to the core domain objects? Are there objects to represent this data, or just files/tables? For example, consider atoms_ui table that tracks AUI assignments over time.
• • •
• • •
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
“Kinds” of Objects Kinds of Objects
• • • • • Content Metadata History i Statistics Auxiliary
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Content
• Stable primary key UI – maintained across time – p y y corresponding to identity fields. • Full AIM Semantics • Metathesaurus Data
– Concept – Lexical Info Lexical Info – String Info
• Source Data
– Atom, Attribute, Relation, Tree Position
• Mappings, Subsets, Content Views, Value Sets • Atom Cluster, SourceAtomCluster
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Meta Content Diagram Meta Content Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Source Content Diagram Source Content Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Atom Cluster Inheritance Diagram Atom Cluster Inheritance Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Clusters Venn Diagram (1) Clusters Venn Diagram (1)
LUI1 LUI2
SUI1
SUI3
SUI4 split across CUI1,CUI2 LUI3 split across CUI1,CUI2 LUI1,LUI2 merged in CUI1 LUI1 LUI2 merged in CUI1 LUI2,LUI3 merged in CUI2
SUI2
SUI4 CUI1 LUI3 CUI2
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Clusters Venn Diagram (2) Clusters Venn Diagram (2)
SDUI1 LUI1 LUI2 CODE1 LUI3
LUI7 CODE2 CODE3 LUI6 LUI5 CODE4 L2 CODE5 CODE6 LUI4
CODE1
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
RRF Limitations (1) RRF Limitations (1)
• No explicit model of “concept” or corresponding objects for SCUI, SDUI, CODE, LUI, SUI CODE LUI SUI
– Leads to inflated representation of DA, MR, ST
• • • • • • • •
Conflation of “obsolete” and “suppressible” Lack of support for “AIM” semantics L k f t f “AIM” ti Definitions can only be connected to atoms Some statistical data misrepresented as content (e.g. MED<year> attributes in MRSAT) attributes in MRSAT) Unable to represent attribute referential integrity as a database foreign key (conflation of different “kinds” of attributes) Relations have confusing directionality (should be more explicit about Relations have confusing directionality (should be more explicit about “to” and “from”) SIB relations are inconsistent Identifiers in MRHIST are not resovled to UMLS UIs. Identifiers in MRHIST are not resovled to UMLS UIs
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
RRF Limitations (2) RRF Limitations (2)
• Hierarchies are only atom‐based (e.g. no STYPE in y ( g MRHIER) • Confusing/awkward representation of Subset information. Conflation of Subset with Concept. information Conflation of Subset with Concept • Confusing/awkward representation of content view member attributes view member attributes • Conflation of Content View and Concept. • Conflation of MapSet and Concept. p p • Metadata model is very limited and awkward to extend.
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
Open Sources/Tools Open Sources/Tools
• Unifying tooling infrastructure across UMLS Unifying tooling infrastructure across UMLS and terminology projects • Modern tools and technologies Modern tools and technologies
– SVN, Maven –S i L Spring, Lucene – GWT, CAS – Hibernate, Lime Survey
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
26
Modules
• SVN
– File based configuration management tool – Integrates well with Eclipse (development tool) Integrates well with Eclipse (development tool)
• Maven
– Project Management Tool – Automatic build/deploy/test projects – Automatic project dependency management and update (via repositories)
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
27
Modules (contd.) Modules (contd.)
• Spring: application framework (IOC, AOP) Sp g: app cat o a e o ( OC, O )
– – – – Application is highly configurable Declarative programming Reduces code base A number of pre‐built modules: Security, DOA, TM, MVC, Messaging… MVC Messaging
• Lucene: search engine
– Searches are serviced by Lucene (not Database) Searches are serviced by Lucene (not Database) – High‐performance, full featured and cross platform text search engine
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
28
Modules (contd.) Modules (contd.)
• GWT: Google Web Toolkit
– – – – – – – – – – Simple and powerful Web based GUI tools Built in support for AJAX Allows development in JAVA p Converts JAVA code to JAVA Script to run on client machine Multi‐browser compliant 508 Compliant (with modifications) p ( ) A single sign on system Provides security via pluggable backend security modules Provides security via pluggable backend security modules Easy to integrate applications Only the user and one and only one CAS module knows the passwords in an enterprise passwords in an enterprise
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
29
• CAS: Central Authentication System
Modules (contd.) Modules (contd.)
• Hibernate: ORM libraries be ate: O b a es
– Provides object relational mapping – Removes SQL coding – Declarative mapping of objects, associations, and constraints to the underlying database – Supports la y loading and various fetch strategies Supports lazy loading and various fetch strategies – Efficient with internal caching across calls – Supports query caches Supports query caches – QA by referential integrity
• Blocks bad data entering the system
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
30
Deployment Diagram Deployment Diagram
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
31
Work with IHTSDO workbench Work with IHTSDO workbench
SNOMED CT request submission workflow SNOMED CT request submission workflow UMLS content from the UTB/KSS Pubmed and literature searching b d d li hi Flexible searching with UMLS lexical tools Maintenance environment for UMLS Semantic Network • Collaborate on ICD10 mapping • • • • •
U. S. N A T I O N A L L I B R A R Y O F M E D I C I N E
