Background
I tracked the surgical procedures of a real-life novice surgeon in their first year of performing prostatectomy cases and captured their total procedure duration for each case. I then fit a log-linear regression model to understand whether a learning curve can identified with high confidence; Also included is the Surgeon’s rate at which they complete cases (average procedures completed per day) over the same period.
Plot Insights
As this surgeon gains experience, they take overall less time to finish the surgery, and their observed procedure durations have less variability1, becoming more and more predictible:
- January 2020: Surgeon has limited prostetectomy case experience, and Surgeon’s procedures take between 100 amd 160 minutes to complete.
- January 2021: Surgeon has completed around 80 prior cases and Surgeon’s procedures take between 83 and 95 minutes to complete.
Additionally as surgeon progresses through their first year of procedures, they are able to complete cases at an increasing rate:
- January 2020: Surgeon has limited/no prostetectomy case experience, and surgeon performs cases at an average rate of 1 case every 6 days
- January 2021: Surgeon has completed around 80 prior cases, and surgeon performs cases at an average rate of 1 case every 2 days
Footnotes
This is consistent with a surgery like prostetectomy where case complexity remains relatively consistent across patients.↩︎