Evaluating the Effectiveness of an AI-Powered Physician Assistant in Improving Patients' and Physicians' Satisfaction in Anaesthesiology Clinics

実験的：AI-Assisted Care

The patients in this arm will receive AI-assisted care in addition to the standard care. The AI platform will take the patient's history, and then the patient will talk to the consultant as part of the routine flow.

他の：AI-Powered Physicians' Assistant

The study participant allocated to the intervention arm will interact with the AI-physician assistant application before they consult with the physician. The application will collect medical history of the patient. This will then be followed by an AI-generated clinical summary, which their physicians will receive before the consultation begins. Physicians will review this summary and ask further questions of patients if required. Any additions and changes in the patient's history will also be made. Physicians will subsequently conduct a physical examination of the patient. After this, the physician will be able to view AI- and guideline-based suggestions for the patient's assessment and pre-operative management. The recommendations can be selected, modified, or not used as per the physician's expertise. All additions within the application can be either typed manually or verbalised via an AI-assisted scribe within the application.

介入なし：Standard Care

The patients in this arm will receive standard care. The standard of care for our study is what is usually applied in Anaesthesiology outpatient clinics, with consultants seeing the patients after the resident.

Patient Satisfaction and Outpatient Experience Score (Adapted NHS Outpatient Survey - Patient Questionnaire)

Patient satisfaction will be evaluated across two domains: (1) effective utilization of waiting time and (2) receipt of patient-centered care during the outpatient visit.

Evaluated post-consultation using a structured survey adapted from the NHS Outpatient Survey: Section A (Wait Time): Wait duration (A1) and wait time utility (A2). Section B (Doctor Interaction): Patient-centered care, consultation duration, listening, and clarity (B2-B7). Section C (AI Assistant - Intervention Arm Only): Perceived listening, trust, and privacy (C1-C6). Section D (Overall Impression): Respect, dignity, and care- quality (D1-D3).

Ordinal items are assigned numerical scores to calculate a composite mean score and domain sub-scores (range: 1.0-5.0). Higher scores indicate greater satisfaction.

Daily Physician Satisfaction and Workflow Efficiency Score (End-of-Day Physician Survey)

Physician satisfaction and workflow efficiency will be evaluated using an end-of-day survey administered to physicians whose patients were recruited into the study. Evaluated using a two-part structured questionnaire:

Section 1 (Comparative Workflow): Binary comparisons across 5 domains (consultation duration, patient engagement, documentation efficiency, cognitive workload, and focus on patient care), with 3-point follow-up sub-questions for favorable intervention responses.

Section 2 (Intervention Utility): 6 intervention-specific items assessing accuracy/trust, workflow efficiency, workload relief, and overall satisfaction (5-point ordinal scales; 1 = Strongly Disagree, 5 = Strongly Agree).

Daily composite mean scores are calculated from Sections 1 and 2 (range: 1.0-5.0). Higher scores indicate greater satisfaction.

Total Physician Consultation Duration (Stopwatch Timestamps)

Evaluates total physician consultation duration (in minutes) for each patient during their outpatient clinic encounter.

Consultation time refers to the active time spent by the physician while the patient is inside the consultation room. It encompasses the time taken to inquire about symptoms, conduct physical examinations, prescribe treatments, and provide counseling.

Duration is objectively measured in minutes using manual stopwatch timestamps, initiated the moment the patient enters the consultation room and concluded when the patient departs.

Mean total consultation times will be calculated per patient and compared between the intervention and standard care arms. Lower or more optimized consultation durations reflect enhanced workflow efficiency.

Process flow evaluation outcome - Mean Queuing Time (Timestamp Tracking)

Recorded using objective timestamps across two sequential intervals:

Interval 1: Completion of counter registration to initiation of vital signs measurement.

Interval 2: Completion of vital signs measurement to patient entry into the physician consultation room.

Intervals are combined into a single aggregated total queuing time per patient visit. Lower mean queuing times indicate improved operational workflow efficiency.