| Abstract: | SUMMARY Social workers and other mental health workers lack clinical decision support tools to predict which clients are at greatest risk of psychiatric rehospitalization. Artificial neural networks (ANNs), are computer decision support tools that make prediction and classification decisions based on accumulated experience and information contained in successfully solved cases (correct decisions). This study evaluates the use of ANNs in predicting rehospitalization of severely mentally ill outpatients. Eight Bayesian ANN models achieved correct prediction rates ranging from 75% to 93% for two prediction conditions. These results support the utility of Bayesian ANN models in the development of clinical decision support tools. |
