Machine Learning Optimization of ICU Care: A Cost-Effectiveness Model

Machine learning and artificial intelligence have been more readily utilized with increasing prevalence of Electronic Medical Records in health care. As the costs of healthcare delivery continue to rise, new solutions for increasing efficiency are of great public interest. The intensive care unit accounts for a disproportionate amount of resource utilization per patient and so this is a worthwhile arena to focus attention. Although clinical prediction tools and machine learning algorithms can benchmark ICU performance, legacy models do not give information on best ways to optimize modifiable risk factors for important ICU outcomes. This model explores the cost effectiveness of a machine learning model which benchmarks modifiable risk factors for ICU length of stay. This machine learning system might inform the ICU of problem areas for which to target best practices and optimize ICU length of stay.