Table 1 Algorithm test statistics.

From: Optimizing prediction of response to antidepressant medications using machine learning and integrated genetic, clinical, and demographic data

Statistic

Citalopram

Venlafaxine

Sertraline

Mean

Balanced accuracy

60.5%

74.3%

75.5%

70.1% (6.8)

p value (balanced accuracy)

<0.001

<0.01

<0.01

Sensitivity

67%

70%

69.2%

68.7% (1.3)

Specificity

54%

78.6%

81.8%

71.4% (12.4)

Accuracy

59.8%

75%

75%

69.9% (7.2)

PPV

59.3%

76.6%

79.2%

71.7% (8.8)

NPV

62%

72.4%

72.7%

69% (4.9)

  1. Statistics table describing the success of the algorithm in predicting response and no-response per medication in the final STAR*D test set. n = 251 (citalopram), n = 24 (venlafaxine), n = 24 (sertraline). Standard deviations are given within parentheses.