Medical guidelines that lay down state-of-the-art rules for doctors are an important tool in modern medical practice. But though the number of guideline documents has proliferated in recent years, the quality, clarity and overall usefulness of the texts could be improved, resulting in even greater benefit to patients and medical practitioners.
A team of European researchers working on the IST-funded Protocure II project set out to do just that, developing a method to make guidelines not only more accurate and useful to doctors, but also far easier for national healthcare authorities to generate and update. Most strikingly, they did it by looking at guidelines not as simple texts, but as modular software programs that can be written in a programming language.
“In an earlier project, Protocure I in 2002, we analysed medical guidelines and determined that they can be written in a formal way using a programming language, then verified by computer to eliminate inconsistencies, repetitions and other flaws. In Protocure II we created the tools to translate the guidelines into a programming language, verify them and keep them up to date,” explains project coordinator Mar Marcos of the Universitat Jaume I in Spain.
Turning guideline creation into a software process
In essence, the researchers in Protocure I likened the guideline to a computer program. In Protocure II, which ended in June 2006, they took that analogy a step further by likening guideline creation to software engineering.
A key outcome of the project is a process for translating raw text guidelines into machine-readable and machine-verifiable languages based on XML. This process is strongly supported by an interactive guideline mark-up tool named Delta.
"Delta allows us to track the whole transformation process, which can be crucial for incorporating changes made to the original guideline texts," Marcos says. The project partners also make use of KIV, a software verification system that enables them to check the transformed guideline against a set of generic rules that are common to all guidelines.
The Protocure II system was tested successfully on a guideline for treating breast cancer, which had been developed by one project partner, the Dutch Institute for Healthcare Improvement (CBO).
By making guideline development more efficient and less prone to errors, the researchers’ efforts promise considerable benefits for the national healthcare authorities and institutes that draft guideline documents. With Protocure II’s method their texts, which can sometimes run to 200 pages, should have fewer inconsistencies, making them of greater value to doctors and healthcare coverage planners.
Toward “living guidelines”
With the system making guideline contents easier to update, specialised healthcare organisations will be able to update the guidelines faster to keep pace with new technologies, new drugs and the results of clinical trials. According to the team, Protocure II could underpin the development of true “living guidelines”, which change as fast as medicine advances.
Using a programming language as a basis for such guidelines also enables them to be used as a relatively simple decision-support system for doctors, further increasing their value to the medical community.
But despite the evident potential of the Protocure II tools, Marcos admits that deployment of the system would represent a major change from existing practice for the health institutes that draft guidelines. Nonetheless, the work of the project generated considerable interest at the October 2006 eChallenges conference in Barcelona.
A further obstacle at present is that the system has to be implemented and the transformation of guidelines carried out by IT specialists. However, Marcos says that one of the possibilities the Protocure II partners are considering is a way to support the implementation of guidelines, possibly within a future European research project.
Department of Computer Engineering and Science
Universitat Jaume I
Campus de Riu Sec
Tel: +34 964 728 288
Fax: +34 964 728486
Source: Based on information from Protocure II