Knowledge discovery and mining for nurse activity and patient data

This research examines the workload of nurses to identify important factors for developing a nurse assignment simulation. Data on patients and the locations of nurses were collected from Baylor Regional Medical Center. To preserve the privacy of nurses, patients and the medical center, an encryption code was developed as part of pre-processing the data. Factor variables of interest included categorical variables like month, shift, nurse type, diagnosis and location. The response variable was time spent per location per visit at different locations, such as a patient room or the reception desk. Various data mining methods were employed to extract important knowledge, specifically, a regression approach for categorical factors and tree-based methods. Results of these methodologies are discussed for their merits in their application to nurse assignment simulation and optimization.