Usability Evaluation of a Central Monitoring System with AI-Based Cardiac Arrest Prediction in the ICU

<b>Background/Objectives</b>: The incidence of cardiac arrest among critically ill patients has been increasing, with many patients experiencing clinical exacerbation prior to the event. Early detection and rapid treatment are essential to reduce the risks associated with cardiac arrest; however, difficulties such as limited ICU resources and inadequate monitoring of vital signs reduce the effectiveness of treatment. Various cardiac arrest prediction systems have been developed to overcome these issues. This study performed a summative evaluation of a Central Monitoring System with AI-based Cardiac Arrest Prediction. <b>Methods</b>: A summative usability evaluation was conducted in a simulated ICU environment with 22 ICU nurses experienced in using patient monitoring devices. Participants completed tasks based on the device workflow and then filled out the System Usability Scale (SUS) and satisfaction surveys, with task performance and survey responses analyzed to assess usability. <b>Results</b>: The usability test achieved a task success rate of 90%, with critical tasks achieving success rates ranging from 73% to 100%. The SUS score was 67.3 ("OK"), and the satisfaction survey showed an average score of 4.5, indicating generally positive user perception. <b>Conclusions</b>: Participants generally rated the system as acceptable, although some tasks showed lower success rates due to design issues such as poor button visibility. Further studies in clinical settings are needed to evaluate the system's effectiveness, user experience, and contribution to the timely detection of cardiac arrest.