Reliability of an Automated Decision Support System for Behavioral Treatment Planning

Automated decision support systems can provide inexperienced individuals with expert advice. But while a computer will always arrive at the same conclusion from a given set of data, human users bring individual variability to decision making. This study, therefore, examined the reliability of the decision support component of an expert system, the Mental Retardation-Expert (MR-E), which assists clinicians who treat problem behaviors of individuals with mental retardation or developmental disabilities. Four clinicians independently used MR-E to obtain consultations on 31 separate cases abstracted from the published literature. They reached a high level of agreement on functional hypotheses regardless of clinical experience, familiarity with MR-E, use of MR-E's basic or advanced modes, or consecutive number of consultations obtained.