The Intelligent ICU Pilot Study: Using Artificial Intelligence Technology for Autonomous Patient Monitoring

摘要: Currently, many critical care indices are repetitively assessed and recorded by overburdened nurses, e.g. physical function or facial pain expressions of nonverbal patients. In addition, essential information on patients their environment not captured at all, in a non-granular manner, sleep disturbance factors such as bright light, loud background noise, excessive visitations. this pilot study, we examined the feasibility using pervasive sensing technology artificial intelligence for autonomous granular monitoring critically ill Intensive Care Unit (ICU). As an exemplar prevalent condition, also characterized delirious non-delirious environment. We used wearable sensors, light sound high-resolution camera to collected data analyzed deep learning statistical analysis. Our system performed face detection, recognition, action unit head pose expression posture actigraphy analysis, pressure level visitation frequency detection. were able detect patient's (Mean average precision (mAP)=0.94), recognize (mAP=0.80), postures (F1=0.94). found that all expressions, 11 activity features, during day, night, levels, levels night significantly different between (p-value<0.05). summary, showed is feasible can be characterizing conditions related factors.