Saving lives with data: Identifying patients at risk of decline
Many hospitals combine early warning systems with rapid response teams (RRT) to detect patient decline and respond with elevated care. Predictive models can minimize RRT events by identifying at-risk patients, but modeling is difficult because events are rare and features are varied. Emily Spahn explores the creation of one such patient-risk model and shares lessons learned along the way.
|Talk Title||Saving lives with data: Identifying patients at risk of decline|
|Speakers||. . (ProKarma)|
|Conference||Strata + Hadoop World|
|Conf Tag||Big Data Expo|
|Location||San Jose, California|
|Date||March 14-16, 2017|
Many hospitals rely on rapid response teams (RRTs) to intervene when patients experience a rapid decline. When hospital staff observe a patient in decline, they raise an RRT event, summoning the team to the patient’s bedside. RRT events are rare, occurring in less than 0.5% of all hospital patients, but response teams are a substantial resource, deployed to achieve the best possible patient outcomes. However, optimizing their utilization is a constant challenge. Teams, include medical professionals versed in critical care, use the time between routine rounds and events to monitor patients, looking for early indications of possible decline. Sometimes it’s clear who warrants additional attention, but often shift changes, patient movements, and subtle changes in patient conditions complicate the process. Emily Spahn explores the creation of a patient-risk model that uses machine-learning techniques to produce a patient risk score, in order to assist the RRT in focusing their efforts. This proof-of-concept project uses electronic health records to model the likelihood of a patient experiencing near future health declines indicative of an RRT event. The model produces a simple patient risk score on a scale of 0 to 10. By highlighting patients at risk, the hope is that early intervention can avoid an RRT event altogether. Emily discusses the technical challenges of this modeling effort—which uses tools from the Hadoop and Python ecosystems to study these rare events—and shares the approaches taken to bring consultants, technology providers, and hospital administration and staff onto same page, guiding the project toward success.