XML::DT - a Perl down translation module		Table of contents		Indexes		Producing and using intelligent graphics in XML/SGML electronic publishing

Practical use of XML in Healthcare applications

Ingo Elferring

Managing Director

MedicalData Service GmbH An der Alten Ziegelei 20
Muenster Germany 48157
Phone: +49 251 141620
Fax: +49 251 1416221
Email: Elfering@medicaldataservice.de Web: www.medicaldataservice.de

Biographical notice:

Ingo Elfering is Managing Director of MedicalData Service GmbH (www.medicaldataservice.de ), a Germany-based Software Development Company. MedicalData Service was founded by Ingo Elfering and Dr. med. W.Voss 1982 as a software vendor for Practice management software for physicians. MedicalData Service developed several different systems until it decided to leave the market in 1995 and to concentrate on Internet-based HealthCare applications and Disease Management. MedicalData Service today is operating with different partners in several European Countries and in the USA. Current products of MedicalData Service include IDT (www.idt-online.de ), a secure automating message client for Healthcare and @med, a toolbox for building Internet-based Healthcare software based on the Windows DNA paradigm. Ingo Elfering is chairman of KDT working group of the QMS (www.qms- de.org ), a group of several German healthcare software vendors and medical device manufactures; KDT is a standard of interoperation between message clients and other healthcare application. He is member of HCPP working group (www.hcp-protokoll.de ); a group lead by the Bavarian Physician Association to define a set of security standards to exchange medical data over the Internet. He is involved in several initiatives regarding German Data standards like BDT and LDT. He is member of MS-HUG USA and Country contact for Germany of MS-HUG Europe (www.ms- huge.org ). Also he is a member in the MS-HUGE ActiveX for Healthcare working group. MedicalData Service is member of TeleTrust (www.teletrust.de ), an Industry Group working on Data Security and Privacy. MedicalData Service is also member of the W3C(www.w3c.org ) with special interests in XML and Query Languages for XML. MedicalData Service also observes the work of HL7 and CEN/TC 251.

ABSTRACT:

MedicalData Service has started to use XML within it's software back in April 1998. MedicalData Service presents at XML'99 three different types of how XML can be used.

XDTML: An initiative together with KBV/ZI and QMS in Germany to transform all German so-called xDT data standards into a unified XML version. The xDT standards are widely used in Germany and define with more then 500 items all medical, lab and billing data. XDTML re-uses all the work that has gone into xDT and expands it into XML, combining it with the new benefits of XML.

XML for Clinical Trials: For questionnaire based Clinical Trials MDS has developed a system to automatically transform such questionnaires into XML structures that offer great advantages over traditional data collection and storage methods.

MDS has developed a web based EMR system that can work with any kind of XML-based medical data thru effective use of a browser. EPA@med is a secure medical record system for physician networks. It uses e.g XDTML for medical record items and a special XML format for Rx-data, but it can also operate with other XML data sets.

The presentation will describe the design principles and the practical experiences gained from these projects that have all emerged into live and up-and-running real world applications.

Introduction

@med
IDT

MDS is a software development company, developing software for Healthcare. It was founded 1982 and developed software for Practice Management Systems until 1995. Now it is developing secure communication software (IDT), which allows secure message transport of healthcare data over the Internet, and can automated data exchange. It is an add-on to existing Practice Management Systems. Furthermore MedicalData Service develops software for Disease Management-style applications (@med), which is a toolbox for applications based on the Windows DNA Paradigm including server components and client extensions.

XDTML

XDT

XDTML

XDTML is an initiative of a small group of software vendors and the German national physician association. In Germany currently so-called xDT standards are widely used.

Examples are: ADT = billing/claims information BDT = medical data, indication-based disease data GDT = medical device data LDT = lab data and lab orders

XDT is a ASCII File standard based around numeric tag's and field content like

0123000234>

where the structure is length | tag | tag-content. In the example the tag 3000 is for a Patient ID, in this case 234. Each line is terminated via CR/LF. Currently over 500 items/tags are defined that describe nearly all areas of healthcare data.

XDTML was started as a straightforward approach to ease the transition process from xDT to XML. Therefore it was developed with the goal that the XML tags should be build from the existing numeric xDT tags and that the nesting structure of xDT to be defined by XDTML based on a generic rule.

The XDTML approach is build around a component structure, that defines a owner and a target for the information and then a container holding different blocks of XML each capable to expressed different information and allowing the use of XML Namespaces. This approach reflects current work in Germany for KVDT a container based xDT extension for new billing data standards.

XDTML introduces some new concepts in xDT like ID/IDREF linking of information within the documents, digital signatures within the XML document, LDAP lookup support for identifying physicians and patients.

A first version of XDTML is defined and examples are available. Data structure and data typing was defined using Schemas and several examples of XML and XSL files where produced. The work is now being consolidated and a first more official version of XDTML is expected in Summer 99. Some work still has to be done on XDTML, but it shows the direction clearly.

Example: (see also www.medicaldataservice.de\\xml for full examples, Schema etc.)

<XDT.ITEMPATIENT ITEMID="I1">
	<F3000>1</F3000>
	<F3100></F3100>
	<F3101>Musterpat</F3101>
	<F3102>Heinz</F3102>
	<F3103>1950-04-05</F3103>
	<F3104>Dr.</F3104>
	<F3105>9876543210</F3105>
	<F3107>An der alten Ziegelei 20</F3107>
	<F3110>1</F3110>
	<sF3626>0251/326432</F3626>
</XDT.ITEMPATIENT>

<XDT.ITEMBDT ITEMID="I2">
	<F3000>1</F3000>
	<F6200>1998-03-03</F6200>
	<F6330>Unters.-Befund
	<F6331>Gr. 179 cm, Gew. 88,5 kg, (BMI 28) RR 163/90//85. Das Abdomen tastet
sich weich. Leber und Milz sind nicht vergrößert. Eine pathologische
Resistenz oder Druckdolenz ist über dem Abdomen nicht feststellbar. Die
Nierenlager sind nicht klopfempfindlich und die Bruchpforten geschlossen.
</F6331></F6330>
	<F6332>Techn.-Unters.-Befund
	<F6333>SONOGRAPHISCH sind Leber, Gallenblase, Milz sowie dir gr.
Bauchgefäße gut
darstellbar, unauffällig und bei gez. Palpation nicht schmerzhaft.
Das Pankreas ist wegen Üsberlagerung nicht sicher abgrenzbar.
Deutlich Splitterung im Pyelonreflex li.
deutlich mehr als re. Parenchymsaum bds. schmal.</F6333></F6332>
	<F6334>Un.-Empf.
	<F6335>Vorab empfahl ich dem Pat., bei Beschwerden Riopan-Gel einzunehmen
und üsber
den Effekt zu berichten. Fenistil ret. empfohlen.</F6335></F6334>
	<F6210>Riopan-Gel</F6210>
	<F6210>Fenistil ret.</F6210>
	<F6220>oB</F6220>
</XDT.ITEMBDT>

XDTML is in real-world use. Different German software vendors are starting to implement support for it and recently talks with national hospital software vendors were held to integrate secondary care information systems and practice management systems based on XDTML.

MDS has developed a prescription budget-controlling tool as a web app that receives xDT data files from physicians, translates them into XDTML data on the server and adds Rx-information like billing codes, prices etc.

Findings:

XDTML is easy to understand for people coming from xDT. It allows therefore also fast implementation for vendors and is very flexible due to it's component architecture.

Clinical Trial Generator

Clinical Trial

The clinical trial generator is an application based on the @med toolbox. It is build to transform automatically WinWord Questionnaires into DHTML applications and build data storage via XML files into this. The application takes questionnaires, which allow structured data entry for patients. These questionnaires can work together with EPA@med, MedicalData Service's EMR application and with @med's XML-based medical data repository.

The data entered by the physicians can be structured to any XML definition based on XML-templates.

The Clinical Trial Applications allows therefore remote and offline data entry and the data is send to a web server via secure EMail where it can be analyzed. The server can also derive data, e.g. applying assumptions on the data set and store the modified XML data. The @med server then allows online analytics of this data via a secured website. The first Clinical Trail currently based on this technology is currently operated for SmithKline Beecham's Avandia Diabetes Product. The software is installed in 8 European countries and data about 8000 patients will be collected and transmitted to the server where automatic analyses are run and published back via a website to the team members. The project is working and real life data is used.

Findings

XML gave MDS the possibility to easily design data structures that can hold in one database data from different countries that each have additional questions and collect data in different granularity. The XML/DHTML approach was easy and fast to develop. Based on @med the full project was finalized by a 3 people team in 2 months

Web based medical record based on XML

EHR

EMRl
Medical Record

Web

MedicalData Service has developed based on it's @med toolbox a full web-technology based EMR product called EPA@med.

EPA@med uses XML as it's native storage format of medical data. This data is then stored within a Microsoft SQL 7.0 Database. Due to it's nature EPA@med can simultaneously handle different XML data standards in one database. This was proved a few weeks ago in live test where several thousand entries of XDTML, GEHR via XML and HL7 via XML for stored in one database and accessed as one patient record.

EPA@med use DOCTYPEs to differentiate the handling of several data standards and can therefore be configured based on DOCTYPEs.

Viewing and usage of data is based on applying XSL stylesheets. @med uses XSL pattern-matching for indexing and querying data in it's database.

EPA@med uses the browser as it's user interface, e.g. to display a record entry a stylesheet is used, hyperlinking as a record entry is supported, therefore allowing links into PACS or hospital systems. Fully display of records is also done via a browser window, therefore supporting stylesheets, scripts and validation.

EPA@med can also collect via secure EMail data from different sources.

EPA@med is build based on an extensive security architecture using several different technologies like SSL, DCOM, S/MIME, ActiveX Documents etc.

The product is launched and is currently installed in first pilots in Germany to help physicians to better manage emergency practices and group practices.

Findings

We found building an EMR around the native use of XML very powerful and flexible. The use of browser- technology and stylesheets etc. allowed us to build a very configurable system, e.g. data display and data entry using DHTML for different specialties. XML/XSL and DOCTYPE-based configuration allows support within one product of different data standards. We found hyperlinking as a powerful way to link into other systems like hospitals or PACS. Overcoming the several different security issues was a major task in the project, which was completed successfully and now gives us a very strong and secure web based EMR system.

Tool we use

In our development we use mainly Microsoft tools. Therefore we used Internet Explorer 5.0 Beta 2 and Final Version and IE 4's XSL/XML ActiveX Controls. We found the support in IE 5 for XML, Schema and XSL very good, even in the Beta 2 Version.

The programmers used XML DOM to access the data and found this a very good way to keep the software very flexible. Also this eased the developed of parsing XML data greatly.

We are developing several special COM objects for indexing and querying data based on XSL pattern matching.

We found Schema's to be better then pure DTDs.

We found only few programmer tools for XML/XSL and in Feb 99 there where for example no good XSL or XML Editors.


XML::DT - a Perl down translation module		Table of contents		Indexes		Producing and using intelligent graphics in XML/SGML electronic publishing