Improving emergency department patient-doctor conversation through an artificial intelligence symptom taking tool: an action-oriented design pilot study.

IMPORTANCE Communication between patients and healthcare professionals is frequently challenging in the crowded emergency department (ED), with few opportunities to develop rapport or empathy. Digital tools for patients and physicians have been proposed as helpful but their utility is not established. OBJECTIVE To evaluate a patient-facing digital symptom and history taking, as well as handover tool in the waiting room. DESIGN A two-phase, questionnaire-based quality improvement study. Phase I observations guided iterative improvement, which was then further evaluated in Phase II. SETTING ED of a German tertiary referral and major trauma hospital providing interdisciplinary treatment for an average of 120 patients daily. PARTICIPANTS All patients who were willing/able to provide consent, excluding patients: (i) with severe injury/illness requiring immediate treatment; (ii) with traumatic injury; (iii) incapable of completing a health assessment; or, (iv) under 18 years old. Of 1699 patients presenting to the ED, 815 were eligible based on triage level. With available recruitment staff, 135 were approached, of whom 81 were included in the study. INTERVENTION/OBSERVATION Patients entered information into the tool, which generated a handover report to be accessed via a clinician dashboard. All users completed evaluation questionnaires. Clinicians were trained to observationally assess the tool as a prototype, without relying upon it for clinical care. MAIN OUTCOMES AND MEASURES Patient and clinician Likert scale ratings of tool performance. RESULTS Respondents were strongly positive in endorsing the tool9s usefulness in facilitating conversation (75% of patients, 73% physicians, 100% nurses). Nurses judged the tool as potentially time saving, whilst physicians assessed it as time saving only in some ED medical specialisms (e.g. Surgery). Patients understood the tool questions and reported high usability. The proportion of patients, physicians and nurses who would recommend the tool was 78%, 53% and 76%. CONCLUSIONS AND RELEVANCE The system has clear potential to improve patient-HCP interaction and make efficiency savings in the ED. Future research and development will extend the range of patients for which the history collection has clinical utility.