Modelling Language for Domain-Specific Decisions in Healthcare

Abstract

Healthcare is a major expenditure in today’s economy. In order to save costs while maintaining or even improving quality, optimal practitioner support is required. In this thesis, it is investigated, how actors that are dealing with the transferal management processes and decisions, can be assisted with the help a domain-specific modelling language. The transferal management domain can be assigned to the administrative pathway in healthcare and is as an example dealing with the interface between acute hospital and rehabilitation clinic.

In an extensive literature review, I have found out, that business process management and process modelling, as a part of it, do enjoy a high importance in companies. In the last few years, standards for modelling processes such as BPMN are more and more adopted and accepted. However, these standards are usually not domain-specific but general-purpose. The healthcare sector however, due to its characteristics, seems to require a domain-specific modelling language, which provides domain-specific constructs and is easy to use and to understand. This has also been agreed by several researchers, who developed a domain-specific modelling language for healthcare to provide better support. These languages were always targeting the clinical pathway and thus no solution for the more administrative activities and decisions in transferal management was identified.

The University of Applied Sciences St. Gallen is conducting a research project to provide better process and decision support for transfer management. The results of this project (mainly the transferal management reference process and the description of a specific transferal management case) are used to define requirements for the domain-specific modelling language. The analysis of the provided documentation shows, that a combination of several existing languages is needed to reflect domain-specific peculiarities. Namely, BPMN, DMN, CMMN, Control Element Model, Organization Model and Documents and Knowledge Model are used. These languages are not only combined but domain-specific concepts are added while unneeded concepts are removed. To properly define a new domain-specific modelling language, three parts are available as highlighted in literature review: abstract syntax, concrete syntax and semantics. All these are compiled for the new language in this work.

The development of the modelling language is taking place in ADOxx Development Toolkit, which is a metamodelling platform accepted by the research community. Each language has been extensively adapted to reflect domain-specific concepts such as dozens of new modelling elements or about 300 new attributes. References between different elements ensure integration and an easy navigation between model parts. With the integration of DMN into the new language, decisions can be modelled down to each individual rule. To enable standardization and exchange in the domain, important healthcare and rehabilitation standards such as DefReha©, ICF and ICD are considered.

To demonstrate, that the reference process and the provided use case cannot only be readily modelled in the new language, but at the same time are improved and fulfil the requirements, respective models are provided and evaluated. The models created in ADOxx Modelling Toolkit especially proof, that due to the integration of several languages and standards, most of the necessary information is now concentrated at one point. This was also revealed in the evaluation with a focus group of domain and modelling experts. They further emphasized that the new language achieves to simplify the modelling process and the actual models. Another mentioned main advantage is especially the integration of DMN, CMMN and important standards of the domain.

In the end of the work I have identified possibilities for future research. Due to the limited amount of time which was available to develop the language, the respective library can be further improved with additional and more advanced functionalities. This implies to more and more introduce automation and executability of the language. While with the consideration of the most relevant standards of the domain, already an important step is done, there is further possibility to fully cover all relevant standards. Also an applicability testing of the developed language in other scenarios and environments would be required. This would include to allow the modelling of other transferal management scenarios besides only the transfer from acute hospital to rehabilitation clinic.