This systematic review explored how artificial intelligence (AI) based on machine learning algorithms and natural language processing is used to address and report patient safety outcomes. The review suggests that AI-enabled decision support systems can improve error detection, patient stratification, and drug management, but that additional evidence is needed to understand how well AI can predict safety outcomes.  

The authors reviewed studies using qualitative and quantitative methods to describe nursing workarounds related to the electronic health record (EHR) in direct care activities. Workarounds generally fit into three categories – omission of process steps, steps performed out of sequence, and unauthorized process steps. Probable causes for workarounds were identified, including organizational- (e.g., knowledge deficits, non-formulary orders), environmental-, patient- (e.g., barcode/ID not accessible), task- (e.g., insufficient time), and usability-related factors (e.g., multiple screens to complete an action). Despite nurses being the largest workforce using EHRs, there is limited research focused on the needs of nurses in EHR design.

The authors surveyed over 4,000 care aids from 93 urban nursing homes in Western Canada to assess the association of work environment with missed and rushed essential care tasks. During their most recent shift, over half of care aids (57.4%) reported missing at least one essential care task and two-thirds (65.4%) reported rushing at least one essential care task. Work environments with better work culture and more effective leadership were associated with fewer missed or rushed care tasks.