Improving Time Efficiency in Discharge Documentation Using an AI Copilot Agent: A Prospective Audit in General Adult Psychiatry
Catherine WrightAims:
To assess the time difference in creating CDDs and EDDs manually versus using Copilot in general adult psychiatry.
To assess the quality of AI generated CDDs and EDDs.
Methods:
Resident doctors will be asked to time themselves while writing a CDD and EDDmanually for a selected patient. They will then use the Copilot agent to generate a CDD and EDD for the same patient and record the time taken. The time difference will be analysed to determine the efficiency gains. Qualitative feedback will also be collected regarding the usability and accuracy of the Copilot-generated documents.
Rationale: The audit was prompted by the auditor’s personal experience with dyslexia and the challenges of managing discharge documentation in a high-pressure clinical environment. The Copilot agent offers a potential solution to reduce documentation time and improve patient flow.
Service areas/teams included: The audit was carried out in Royal Cornhill hospital and included data from general adult psychiatry.
Sample Size: 5 patient cases.
Metrics Recorded:
Time taken with Copilot vs. without Copilot, Number of TrakCare pages referenced, Number and nature of mistakes, Qualitative comments on errors, Minutes saved and percentage time saved,
Results:
For CDDs:
Average time saved: 13 minutes 13 seconds per case. Average percentage time saved: 75.3% Largest amount of time saved: 17 minutes 34 seconds (≈74.5% reduction). Largest percentage of time saved: 76.6%.