Cognitive Computing: Integrating Machine Learning Algorithms into Clinical Workflows
St. Joseph’s Healthcare Hamilton (SJHH) is currently in the process of operationalizing the Deterioration Index, a Cognitive Computing model that acts as an extension of the Epic electronic health record to help detect early signs of patient deterioration in order to determine the likelihood that a patient will experience an escalation in care (i.e., transfer to ICU, Code Blue, or CCRT intervention) or mortality event during their hospital stay in real time.
Traditionally, significant resources have been devoted to identifying and treating patients who are clinically deteriorating. Existing tools and early warning systems used to determine patient acuity, such as EWS/CREWS, have been relatively simplistic and required double documentation by the clinical team. This meant that any interventions tended be reactionary; patients would already be deteriorating by the time they were identified as such. With the introduction of the Deterioration Index at SJHH, clinicians are able to take a proactive approach with a model that improves upon the traditional early warning systems and dynamically responds to information entered into the patient’s chart through any Epic workflow, providing the most up-to-date risk assessment based on the information available in the patient’s chart.
SJHH is currently using a phased implementation strategy to operationalize the model, first activating Deterioration Index alerts for the Critical Care Response Team (CCRT) to optimize clinical workflows, minimize the presence of noise in the data model, and review the current state of related nursing assessments and training, prior to operational expansion across inpatient units.
