AI clinical simulation vs. standardized patients

Two published university rate cards put fully loaded standardized patient time at roughly $123 to $177 an hour (UCLA and the University of Utah rate cards, as published at August 2026). For a 60-resident program that works out to roughly $27,000 to $68,000 a year, depending on how many encounters each resident gets. Read that as the SP-time component rather than the whole bill: faculty time, space, case development, and local geography sit on top of it and vary too much between programs to total honestly.

Because there is no defensible national figure to publish. Programs pay different actor wages and use different amounts of training, faculty time, space, and administrative support, so a per-encounter range that hides those components is not comparable between institutions. The published hourly rate above is the SP-time component and stops there deliberately. Bosse and colleagues took the same approach in a study of fifth-year medical students, counting man-hours instead of inventing a dollar figure: their standardized patient group required 172 man-hours against 112 for peer role-play, a 53.6% difference in that one program (BMC Medical Education, 2015). That study did not test AI patients, and its numbers should not be generalized past its own design.

At repetition, not at assessment. A standardized patient encounter brings human judgment and physical presence, which is what an OSCE, a summative assessment, or a complex scenario needs. What the format cannot absorb is a learner who needs the same conversation fifteen times, because each attempt costs an actor booking, a room, and often a faculty observer. The useful question for a program is not a fabricated cost range against a software price. It is whether learners get enough practice before they reach the limited encounters where human time matters most.

No. AI clinical simulation is designed to extend standardized patient programs, not replace them. Programs continue to use SPs for live, high-stakes assessment, while AI simulation provides on-demand deliberate practice between those encounters.

Standardized patient encounters are bounded by the number of trained SPs, available rooms, and faculty observers. AI clinical simulation can run concurrent encounters on demand, so a cohort can practice without booking a separate actor and room for each learner.

Each learner receives the same published case and rubric version. Every score cites transcript evidence, so faculty can inspect the rating. ClinicalSim does not claim that its ratings are more accurate or fairer than faculty judgment.