Skip to main content
. 2022 Jun 14;6(6):e34366. doi: 10.2196/34366

Table 2.

The average score for bias metrics in the random forest–based mental health assessment algorithm (average of 100 iterations).

Bias metrics Observed score, mean (SD) Ideal score
Delta accuracy (%) 15.85 (0.22) 0
Delta true positive rate (%) −0.88 (8.39) 0
Delta false positive rate (%) 33.43 (13.50) 0
Statistical parity difference (%) 26.1 (4.16) 0
Disparate impact 0.682 (0.049) 1.0