Failure to determine the optimal number of nurses required to care for hospitalized patients impacts quality of care, hospital financial performance, and nursing turnover. Insufficient staffed nurses expose patients to serious safety events and long-term adverse outcomes. Over-staffing results in unnecessary expenditure by the hospital along with the risk of increased nursing turnover from under utilization. Determining the optimal number of nurses required at any time or day involves an intricate process of decision making by nursing leaders who work with ever changing information.

California is the first state to mandate a maximum registered nurse to patient ratio by unit to mitigate harm to patients. These nurse staffing


ratios are primarily based on patient acuity; therefore, expected number of patients along with their predicted future acuity and nurse absenteeism is required to predict future nurse staffing needs.
We hope to create a more informed staffing decisions through the creation of recurrent neural networks to predict nurse staffing needs by nursing unit using past time series data on core nurse staffing needs, census, vital signs, acuity scores, emergency department visits, and nursing absenteeism. We will develop a nurse staffing recurrent neural model and then integrate it with an existing Business Intelligence Qlik Sense patient census application to provide information on projected nursing needs within seven days. This is a case of artificial intelligence (AI) algorithm used in tandem with a BI application to provide value to consumer.



Author: Neil Garde

Coauthor(s): Dr. William Feaster Louis Ehwerhemuepha Jennifer Combs Christine Lee Duncan Yeung

Status: Project Concept