Clinical Databases with openEHR (tutorial)

Pablo Pazos Gutierrez^a, Koray Atalag^b,... (please add your name here)

^a openEHR en español, Asociación Chilena de Informática en Salud, CaboLabs , ^bUniversity of Auckland and openEHR New Zealand...., ^c (please add your affiliation here)

Abstract and Objective

Databases for Clinical Information Systems are difficult to design and implement, especially when the design involves the implementation of a formal specification or standard. The openEHR specifications offer a very expressve an generic model for clinical data structures, allowing semantic interoperability and compatibility with other standards like HL7 CDA, FHIR, and ASTM CCR. Software Developers find a great complexity in designing openEHR compliant databases since the specifications do not include any guidelines in that area. This tutorial will expose different requirements, design principles, technologies and techniques that can be used to tackle that problem

Keywords:

EHR; Clinical Database Design; openEHR; Clinical Information Systems; EHR.

Specific educational goals

• The main goal is to shorten the gap between the openEHR specifications and database design and implementation for software developers.
• Secondary goals include: share experiences and lessons learned, enabling discussion about Clinical Database implementation (a topic that is not discussed or taught often enough), and spreading the use of standards at the database level on Clinical Information System development.

Expected outcomes

After the tutorial, we expect attendees to have a better understanding of the openEHR specifications and scope, to know the main requirements of clinical data structures and database technologies that meet those requierments, to learn techniques and good practices that can be used to create different kinds of openEHR-compliant clinical data repositories.

Expected attendees

Software Developers, Database Managers, Software Architects Technical Leaders, Decision Support Specialists, and Business Intelligence Experts, participating in healthcare informatics projects or planning to.

Tutorial description

This tutorial will expose many of the challenges developers have when designing and implementing Clinical Databases for Clinical Information Systems in general, and openEHR Databases in particular. The main motivation of the tutorial is to help reducing the gap between the openEHR formal specifications and a concrete database implementation.

Attendees will receive introductory information to give them context about common requirements of Clinical Databases, including clinical information structures for free text, semi-structured and fully structured clinical records, storing atomic data and clinical documents, querying and using data for healthcare (documentation, order management, Clinical Decision Support, healthcare plan follow-up for chronic diseases and other conditions, etc.), and secondary uses like clinical research, education, public health, quality assurance and complying with medico-legal requirements, among other areas.

Then we'll review the key concepts of the openEHR specs that affect database design (EHR, Folder, Composition, Entry, ItemStructure, DataValue) and the openEHR dual-model (information + archetypes), showing database implementation methodologies and techniques, and how to design the database to provide services for storing complete health records and query those records in the “openEHR way”, by creating queries that use clinical concept definitions instead of relying on the database technology (SQL, XQuery, JSON-based queries, OO queries, etc.), making queries a generic, reusable and shareable artifact between implementations. All this in order to implement the requirements aforementioned.

The main content of the tutorial will include a review of different technologies (relational, document-oriented, key/value) and techniques to actually design and implement openEHR-based databases, including lessons learned from implementation experiences and good practices that may help to tackle implementation problems.

We'll explain how to implement some functional requirements related to health record versioning (to support modifications, amendments and corrections at the database level), clinical data validation and indexing to improve query performance.

Will end with the conclusions about the different approaches to Clinical openEHR Database design and implementation.

During the tutorial, audience participation and discussion will be encouraged.

1. General Organization of the Tutorial Proposal

The tutorial general topics will include:

• Clinical Database requirements that are derived from EHR/EMR requirements and Clinical Documentation standards requirements (ISO 18308, CEN/ISO 13606, openEHR, HL7 CDA, ASTM CCR)
• Review of the core concepts of the openEHR specification, focusing on the ones that will affect the Clinical Database Design (e.g. EHR, Folder, Composition, Section, Entry, DataStructure, ItemStructure and DataValue).
• Implementation methodologies, techniques, good practices and experiences for different types of databases (relational, document-oriented, key/value).
• Considerations about clinical data validation, storage, versioning, indexing and querying.

2. The workshop structure and arguments

The speaker will lead help the attendees to understand key points of openEHR database design, based on previous experiences working on clinical database design and implementation and a deep knowledge of the openEHR specifications. This tutorial will expose the general topics aforesaid following this organization:

• Requirements
  • EHR purposes from ISO 18308 that guide to different uses of data in healthcare:
    • Transactional
    • Aggregated data
    • Charting
    • Reporting
    • Analysis
    • Research
    • Education
    • Medico-legal proof
  • Clinical records organization and main entities (based on EHRA from ISO 18308)
    • Concepts modeled by different standards like HL7, openEHR and CEN/ISO 13606.
    • Clinical information hierarchy
• openEHR Information Model core concepts, Archetypes and Operational Templates
  • Structure, content definition, ways of implementation at a database level.
  • Discussion about difficult areas of implementing openEHR-based clnical databases, opinions from developers.
    • The need of openEHR metadata in the database alongside with data.
• Implementation technologies
  • Different database models to fufill different kinds of requirement: relational, document-oriented, key/value.
  • Relational model techniques
    • Object-Relational Mapping
    • Schema auto-generation from archetypes
    • High level data indexing
    • Pros and cons of the relational model for clinical databases
  • General concepts about document-oriented approach, storing and querying for openEHR clinical documents.
    • XML and JSON databases.
    • openEHR in XML and JSON.
    • Querying mechanisms.
    • Pros and cons of the document-oriented approach for clinical databases.
  • General concepts about key/value approach, storing and querying for openEHR clinical data.
    • openEHR Archetype/Template paths and instance paths.
    • Pros and cons of the key/value approach.
• Clinical databases and information management summary:
  • Data validation
  • Data storage
  • Data versioning (an openEHR requirement, mentioned in ISO 18308, and supported by HL7 CDA documents)
  • Data indexing
  • Data querying
• Practical experience and conclusion.
  • Brief discussion about requirements, approaches and techniques.

Workshop speakers

• Pablo Pazos; openEHR en español, ACHISA, CaboLabs; Uruguay
• Koray Atalag, openEHR New Zealand, University of Auckland, New Zealand
• ...

Resources

....