Cloud Computing & Big Data
Author: Julia Heaton
Coauthor(s): Works Cited: Lee, E. K., Mejia, A. F., Senior, T., & Jose, J. (2010). Improving Patient Safety through Medical Alert Management: An Automated Decision Tool to Reduce Alert Fatigue. Retrieved July 10, 2017, from https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3041356/ Alert Fatigue. (n.d.). Retrieved July 10, 2017, from https://psnet.ahrq.gov/primers/primer/28/alert-fatigue
Status: Project Concept
Physician Alert Fatigue
Background: Doctors have become bombarded with medical alerts. Electronic medical records and developing technologies that constantly monitor a patient’s condition have only contributed to the problem. A 2014 study showed that one hospital with “66 adult intensive care unit beds generated more than 2 million alerts in one month, translating to 187 warnings per patient per day” (Alert Fatigue). This unintended consequence of the computerization of medicine has caused doctors to ignore a vast majority of notifications, even what might be life-threatening alerts. Due to the vast volume of alerts recent research has shown “49 to 96% of alerts are overridden by EMR users” (Lee). Some hospitals are already experimenting with reducing the number of moderate severity alerts which has shown improvement in the reaction time of doctors to more pressing alerts.
Goal: Facilitate patient-doctor communication, organize a doctor’s workflow, and limit distraction by reducing the volume of medical alerts and organizing the system in which doctors are notified.
Method: Artificial intelligence will rank the severity of alerts, increase alert specificity, and organize alerts into manageable categories. The computer system will learn what types of notifications doctors ignore and compare those ignored notifications with the accuracy of patient outcome. After the computer undergoes supervised learning, it will automatically put moderate unnecessary alerts on a “do not disturb” setting or remove them completely. Alerts will be customized to the patient’s condition. For example a patient with diabetes who normally has low insulin levels, the EMR system will not constantly notify the doctor of low insulin levels in comparison to a healthy person without diabetes. However, the doctor will be given an alert if the patient with diabetes falls below the customized setting of insulin levels that is abnormal. Alerts will be sorted to the exact person who needs to be notified. An orthopedist does not need to know the patient needs a vaccine, only the primary doctor needs to be notified. Alerts will be categorized into a “To Do List” with the most important alerts filtered to the top along with a time frame in which action must be taken whether that be a week, 24 hours, or immediately. Only severe alerts will be disruptive. Alerts will also be filtered with matching color systems into the following sub categories of dose alerts, drug-to-drug interaction alerts, and drug allergy alerts.
Discussion: Although electronic medical records have improved charting techniques, the advance of technology in medicine has dramatically increased the magnitude of alerts that doctors receive on a daily basis. Using artificial intelligence to balance the volume and organize of medical alerts, doctors will not be as distracted by constant notifications and are more aware of immediate attention alerts. Ultimately artificial intelligence used to organize alerts can improve patient care by granting more time and accuracy for a doctor to respond to the needs of a patient.