Table 3 Human evaluation of clinical documentation with and without PEACH

From: Clinical and economic impact of a large language model in perioperative medicine: a randomized crossover trial

Metric

Without PEACH (n = 28)

With PEACH (n = 28)

p valuea

Preference Rate

10/28 (35.7%)

16/28 (57.1%)

0.180b

Issues List Present

12 / 28 (43.9%)

20/28 (71.4%)

0.050

Minor Errors

3/28 (10.7%)

2/28 (7.1%)

1.000

Major Errors

1/28 (3.6%)

0/28 (0.0%)

1.000

  1. aFisher’s Exact test.
  2. bCohen’s Keppa κ = 0.71.
Back to article page