Improving Clinical Efficiency Using Artificial Intelligence Scribe for Documentation (ICE-AID Study): A Retrospective-Prospective Cohort Study at a Quaternary Children's Hospital and Health Service.

内容摘要

To evaluate the impact of an artificial intelligence medical scribe (AIMS) on clinical documentation efficiency, document quality, clinician-patient interaction, clinician well-being, consumer perspectives, overtime, and transcription costs. Retrospective-prospective study of AIMS project cohort including medical, allied health and nursing practitioners at a quaternary children's hospital and health service (HHS). Outcome measures included time-to-finalise outpatient correspondence, clinicians' Reflections survey (1&#x2009;month), document quality survey using modified Physician Documentation Quality Instrument-9 tool's domains (3&#x2009;months), System Usability Scale (SUS) (3&#x2009;months). Patient/consumer experience surveys, overtime and cost analyses. Total 131 participants including 89 medical, 36 allied health and 6 nurse practitioners used AIMS showing significant reduction in median time-to-finalise outpatient letters from 7.9&#x2009;days (Interquartile range [IQR] 19&#x2009;days) in pre-AIMS era to 14&#x2009;min (IQR 6&#x2009;days) (p&#x2009;<&#x2009;0.001). In the Reflections survey participants reported improved: patient engagement (62/71, 87%), well-being (80%, 57/71) and work-life balance (66%, 47/71). The document quality survey showed acceptable document quality with median score 4, in 9 of the 10 survey domains (scale 1-to-5, n&#x2009;=&#x2009;77). Mean SUS score was 74 (n&#x2009;=&#x2009;77) suggesting above average usability. Study showed transcription cost savings of 16&#x2009;903 AUD, but no definitive impact on overtime claims. Patient/consumer feedback (n&#x2009;=&#x2009;333) was highly positive with over 90% reporting improved clinician-patient interaction and welcoming future use of AIMS. This study showed significantly improved documentation efficiency, acceptable document quality, and reduced transcription costs with significant benefits to clinician well-being and clinician-patient interaction with the use of an ambient AI medical scribe. Australian New Zealand Clinical Trials Registry: ACTRN12625000432415.