A Knowledge-based Method for Verifying the Reliability of Clinical DSSs

Recently, the usage of innovative decision support systems (DSSs) for monitoring the subject's health status in the daily living is becoming a common practice to provide real aid in chronic patients' management. Rule-based implementations of such DSSs simulate the decision-making process described in clinical guidelines by also allowing new/existing rules to be directly and dynamically added/edited by the physicians. However, this task is error-prone and can generate different kinds of anomalies that compromise the effectiveness and correctness of the final DSSs. In order to face such an issue, this paper presents a novel algorithm for verifying the reliability of clinical guidelines, encoded in the form of complex rules with also statistical or trend patterns in their conditions, so as to determine two categories of potential structural anomalies, i.e. inconsistency or redundancy. Moreover, in order to provide a deep insight about the verification outcome, the method has been enriched with a taxonomy for classifying the detected anomalies. The algorithm has been evaluated on different rule bases in terms of time-processing, proving its efficiency. This research work has been developed within the EU IST CHRONIOUS Project, devoted to define a generic platform for remotely monitoring the health status of chronic patients.