A Neural Network as a Potential Means for Aiding Nursing Diagnosis

This paper describes and discusses the application of neural networks, a new computer technology, to nursing diagnosis. An assessment tool is constructed based on the North American Nursing Diagnosis Association (NANDA) diagnoses as listed by Gordon. Assessment questions are answered by yes/no or numerical scores. Answer sheets are used to produce diagnostic “patterns” that are then taught to ART-2, an advanced type of neural network. After training, the neural network can recognize and classify diagnostic patterns. A pilot test using the bowel elimination diagnostic categories is summarized. The diagnosis of test cases made by the ART-2 generally agreed with those made independently by the researcher and another nurse.