The electronic medical record (EMR) is a central issue of health care computing. Pursuing the knowledge-based focus of interest, this paper reports the results of 3GEMR (3. Generation Electronic Medical Record management system), a subproject of CUBIQ [¹].

As a thesaurus the SNOMED International [²] was chosen. As a structure of medical data the problem oriented medical record was selected. The knowledge handling methodology was a PROLOG extension worked out for project CUBIQ.

The result of the application development is a prototype, which is able to handle the whole SNOMED as a knowledge-base. The prototype conducts a structured interview which collects patient data including medical history and complaints of patient. These data originate a hierarchic set of “problem lines”. The physician can enter his findings and results of tests. The working diagnosis evokes a plan. The data of a patient are presented by a graphic user interface, that shows three layers of information. The prototype is filled up with rules describing the conceptual network of history data, complaints, findings related to chest pain as a group diagnosis

Some problems aroused while development of 3GEMR was going on. While the original SNOMED III source database was converted to PROLOG understandable format, we have discovered several problematic areas. There are problems with the hierarchic structure within and among modules. Even the increasing number of dimensions, now called modules did not solve the problem of multiple (sometimes hidden) hierarchies within one module. A next problem was the limited capability of SNOMED to represent time related data. While the problems were numerous, SNOMED was able to represent most of the patient related data.

The next level was the proposed data structure of medical record. The set of meta-rules describing the problem-oriented record had to be modified and extended during the development. Representing medical history and duration of complaints emphasised the problem of time related data presentation and uncertainty. The problem oriented basic structure based on Weed's approach turned to be applicable in building a patient data-base.

The programming environment on the technical level was able to provide a usable environment for running the 3GEMR management system.

The original approach to build an electronic medical record management system instead of scratch from already proved tools and resources turned to be useful. Experience shows that advantages and disadvantages of using such tools and resources are balanced.