Original ArticlesArtificial Intelligence Forecasting Census and Supporting Early DecisionsGriner, Todd E. DNP, RN, NEA-BC; Thompson, Michael MS; High, Heidi MBA, RN; Buckles, Jenny MSN, RN-BC Author Information Cedars-Sinai, Los Angeles, California. Correspondence: Todd E. Griner, DNP, RN, NEA-BC, Critical Care Services, Cedars-Sinai, 6S02 Saperstein Critical Care Tower, 8700 Beverly Blvd, Los Angeles, CA 90019 ([email protected]). The authors declare no conflict of interest. Nursing Administration Quarterly: October/December 2020 - Volume 44 - Issue 4 - p 316-328 doi: 10.1097/NAQ.0000000000000436 Buy Metrics Abstract Matching resources to demand is a daily challenge for hospital leadership. In interdisciplinary collaboration, nurse leaders and data scientists collaborated to develop advanced machine learning to support early proactive decisions to improve ability to accommodate demand. When hundreds or even thousands of forecasts are made, it becomes important to let machines do the hard work of mathematical pattern recognition, while efficiently using human feedback to address performance and accuracy problems. Nurse leaders and data scientists collaborated to create a usable, low-error predictive model to let machines do the hard work of pattern recognition and model evaluation, while efficiently using nurse leader domain expert feedback to address performance and accuracy problems. During the evaluation period, the overall census mean absolute percentage error was 3.7%. ALEx's predictions have become part of the team's operational norm, helping them anticipate and prepare for census fluctuations. This experience suggests that operational leaders empowered with effective predictive analytics can take decisive proactive staffing and capacity management choices. Predictive analytic information can also result in team learning and ensure safety and operational excellence is supported in all aspects of the organization. © 2020 Wolters Kluwer Health, Inc. All rights reserved.