User Interface and openEHR data

The openEHR Community and Foundation are interested in the standardisation of data entry for the health record. This will enhance the comprehensive framework for data standardisation that openEHR already offers, and should make implementation far more straightforward. There are a number of approaches that are underway internationally. The important opportunity is now to coordinate efforts to build the required user interface widgets and programs to work with openEHR out of the box.

Introduction

The form of data entry that is particularly important to support from an openEHR perspective is structured data entry. While natural language processing of long transcriptions from audio offers some prospects for indexing it is unlikely to lead to safe automatic processing in the near future. Natural language processing, does however, offer a good deal more when the input is somewhat structured (it is easier to process text into structure if the scope of what might be recorded is limited). Language processing can also be done in real time to overcome ambiguity through interaction.

Aspects of the user interface

Different aspects of the user interface need to be developed to provide ease of use and implementation. These are the display of data, the input widgets that are specific to different openEHR data types and natural language processing within a structured environment. Each of these are considered separately.

Display of data

Ocean Informatics has developed generic display scripts per reference model class as XSL snippets. This allows a standardised display of information held in openEHR format and transformed to XML if there is no particular display mandated in the local system. Overriding display can mandated on a per archetype basis. The generic display XSL fragments have been provided as candidate openEHR resources. Some examples of display XSL fragments for common archetypes are provided by way of comparison.

Input using widgets

The input widgets that are required of three types:

• Data type widgets e.g. DV_DATE_TIME, DV_BOOLEAN
• Reference model controls e.g. ELEMENT (which incorporates data types, names, nullflavor), CLUSTER, EVENT etc.
• Archetype controls which incorporate the lower level widgets for commonly used archetypes e.g. Weight, blood pressure, Symptom etc.

Input with natural language processing

Environments which have openEHR 'built-in' and enable transformation of natural language to structured information within specific template contexts including:

• Word-processor
• Hand-writing recognition

Other approaches

Information about other approaches to openEHR data including:

• Web forms and openEHR
• Php and openEHR

Resources

There are a number of initiatives around the world aiming to build standard widgets for screens in general and others for health in particular.

Health Specific Efforts

• Specification
  • The UK Common User Interface Project: A Microsoft and UK NHS initiative with substantial documentation.
• Examples
  • Tony Shannon's example of a data entry aid which provides an example of an aid for structured data input. openEHR's archetypes and templates provide a means of recording the meaning of such statements in a computable form.

General Efforts

• Java
  • YUI - Yahoo User Interface: An open source effort to standardise generic GUI widgets