An AIoT enabled system for optimizing data retrieval in the intensive care unit evaluated in a randomized crossover pilot trial

Healthcare providers (HCPs) in the intensive care unit (ICU) frequently face information overload, which can result in cognitive fatigue and decision-making errors. This study compares the efficiency and accuracy of data collection between an artificial intelligence of things (AIoT)-enabled ICU command center (CC) and a hospital information system (HIS). A randomized crossover pilot trial was conducted with ICU-trained HCPs, who collected data from the most critically ill ICU patients, selected based on their Acute Physiologic Assessment and Chronic Health Evaluation-II (APACHE II) score, using either the CC-HIS or HIS-CC sequence. Data collection time, accuracy, and subgroup differences were evaluated. The effect of increased data volume on time and accuracy was also assessed. The study enrolled 21 HCPs to collect data from five ICU patients. After excluding incomplete or insufficient data, 184 data sets were analyzed. The results showed that the CC significantly reduced data collection time by 41.8% (p < 0.0001), with a mean reduction of 6.75 min per patient (8.5 vs. 15.2 min, p < 0.0001). Furthermore, the CC improved data accuracy by 2.07% (93.8% vs. 95.9%, p = 0.0002). These findings indicate that the AIoT-enabled CC improves both the efficiency and accuracy in data collection in the ICU.