An Ontological Clinical Decision Support System Based on Clinical Guidelines for Diabetes Patients in Sri Lanka

Health professionals should follow the clinical guidelines to decrease healthcare costs to avoid unnecessary testing and to minimize the variations among healthcare providers. In addition, this will minimize the mistakes in diagnosis and treatment processes. To this end, it is possible to use Clinical Decision Support Systems that implement the clinical guidelines. Clinical guidelines published by international associations are not suitable for developing countries such as Sri Lanka, due to the economic background, lack of resources, and unavailability of some laboratory tests. Hence, a set of clinical guidelines has been formulated based on the various published international professional organizations from a Sri Lankan context. Furthermore, these guidelines are usually presented in non-computer-interpretable narrative text or non-executable flow chart formats. In order to fill this gap, this research study finds a suitable approach to represent/organize the clinical guidelines in a Sri Lankan context that is suitable to be used in a clinical decision support system. To this end, we introduced a novel approach which is an ontological model based on the clinical guidelines. As it is revealed that there are 4 million diabetes patients in Sri Lanka, which is approximately twenty percent of the total population, we used diabetes-related guidelines in this research. Firstly, conceptual models were designed to map the acquired diabetes-related clinical guidelines using Business Process Model and Notation 2.0. Two models were designed in mapping the diagnosis process of Type 1 and Type 2 Diabetes, and Gestational diabetes. Furthermore, several conceptual models were designed to map the treatment plans in guidelines by using flowcharting. These designs were validated by domain experts by using questionnaires. Gruninger and Fox's method was used to design and evaluate the ontology based on the designed conceptual models. Domain experts' feedback and several real-life diabetic scenarios were used to validate and evaluate the developed ontology. The evaluation results show that all suggested answers based on the proposed ontological model are accurate and well addressed with respect to the real-world scenarios. A clinical decision support system was implemented based on the ontological knowledge base using the Jena Framework, and this system can be used to access the diabetic information and knowledge in the Sri Lankan context. However, this contribution is not limited to diabetes or a local context, and can be applied to any disease or any context.