Computer-aided decision-making for epilepsy and sleep diagnostics.

Abstract

The main objective of the study was first to develop two prototypes of decision support systems for epilepsy and sleep disorder diagnostics. The second goal was to examine medical decision-making with the help of these program examples and further to assess what requirements should be set to decision support systems for them to find their place in clinical work. Requirements were first defined for the systems to be developed. The requirements were, for the most part, successfully taken into consideration during the development process. Epilepsy Expert, a decision support system based on the International Classification of Epilepsies and Epileptic Syndromes (51) was then developed. An epilepsy expert was in a central role in the development work due to the nature of the classification. For sleep disorders Sleep Expert, a decision support system, was developed based on the International Classification of Sleep Disorders (52). In the developing of the system the role of experts in sleep disorders was minor as the international classification provided a good foundation for program knowledge. The knowledge of the programs was validated as follows. First, three experts were requested to provide 10 case descriptions, and then they made diagnoses of their colleagues' patients. On the basis of these diagnoses a majority agreement, 'the right diagnosis', was reached. From the same epilepsy cases the author made diagnoses with the aid of the decision tree of Epilepsy Expert. Two other physicians, who were not experts in sleep disorders, made diagnoses for sleep disorders using Sleep Expert. In the validation Epilepsy Expert proved partly incomplete, which was due in part to the weakness of the international classification. However, the section of the program whose diagnostics was based on clinical findings only was as good as the experts. In the validation of Sleep Expert the physicians who used the program did not achieve as good results as the experts. The functionality of the programs was evaluated with questionnaires. According to this limited inquiry Sleep Expert could be used in clinical work, whereas Epilepsy Expert was regarded as being weaker. As a whole Sleep Expert was better than Epilepsy Expert. The conclusions to be drawn from the study are: In the development phase factors related to users, knowledge, problem definition and the environmental adaptation of the system need to be taken into account. If international classifications are to be used as a basis for the systems' knowledge, classification should be sufficiently clear and precise in respect of individual diagnoses.(ABSTRACT TRUNCATED AT 400 WORDS)