Artificial Intelligence as a "Firewall" for Medical Alerts

2019-12-03

Yesterday, a study was published in JMIR, aiming to demonstrate that artificial intelligence algorithms have the potential to reduce the number of alarms received by caregivers, alleviate alarm fatigue, and reduce the stress on doctors and nursing staff, providing better and higher quality care for patients. Alarm fatigue is a common issue in healthcare, especially in emergency situations or hospital environments. The ECRI Institute often lists "alarm fatigue" as the top patient safety hazard on its annual list, and frequent alarms and notifications have been shown to lead to physician burnout and errors in patient care. Researchers state that "hospitals may have tens of thousands of alarms every day, but 80% to 99% of them are false and do not require clinical intervention." The research team unanimously believes that one of the main root causes of alarm fatigue is the poor quality of the algorithms generating the alarms. They believe that overload can potentially create an unsafe environment for patients, and the insensitivity of healthcare teams to alarms can lead to mishandling of abnormal situations and longer response times to potentially life-threatening events. To address this, researchers developed an artificial intelligence algorithm to reduce the total number of alarms sent to providers. The algorithm uses a notification delay method to determine whether to deliver a unique notification to healthcare professionals instead of multiple alarms for the same alert situation. The algorithm can decide whether to add a false alarm probability (FAP) label to the notification and who to notify within the healthcare team. The research team applied this algorithm to a dataset that included patient monitoring data and vital signs recorded during the surgeries of 32 anesthesia patients. When comparing the number of alarms triggered by the artificial intelligence system and the number of notifications sent to nursing staff, researchers found that the reduction in notifications received by nursing staff could reach up to 99.3% of the total alarms, with an average reduction of 99.17% of the total alarms. Next, the team plans to further develop the artificial intelligence algorithm to improve its functionality, allowing it to better determine who to notify within the nursing staff based on members' expertise, experience level, availability, geographical location, and current workload status. The researchers also noted that they evaluated the algorithm in an experimental environment, and future work will focus on testing the algorithm under more realistic clinical conditions by increasing the number of patients, monitoring parameters, and types of alarms. Experts predict that with the improvement of this artificial intelligence algorithm, alarm fatigue and its associated negative consequences are expected to significantly decrease.

RECENT AI NEWS

RECENT AI TOOLS