Skip to main content
. 2020 May 6;7:177. doi: 10.3389/fmed.2020.00177

Table 1.

Statistical evaluation of the primary study endpoint (accuracy) and the secondary study endpoints (sensitivity and specificity) for the four different scenarios (training with ground truth majority decision (MD)/testing with ground truth biopsy (BIO), training with MD/testing with BIO, training with BIO/testing with MD, and training with BIO/testing with BIO).

Ground truth for training MD BIO
Ground truth for testing MD BIO MD BIO
Mean accuracy 75.03% 64.24% 64.53% 73.80%
95% CI accuracy 74.39–75.66% 62.66–65.83% 63.12–65.94% 73.10–74.51%
Mean sensitivity 76.76% 69.65% 64.31% 75.98%
95% CI sensitivity 75.36–78.15% 67.92–71.37% 62.74–65.88% 74.69–77.26%
Mean specificity 73.00% 59.05% 64.79% 71.85%
95% CI specificity 71.10–74.90% 56.56–61.54% 63.20–66.38% 71.08–72.61%