Table 2 Citalopram model statistics for both studies’ final test sets: statistics table describing the success of the citalopram model in predicting response and no-response for citalopram-treated participants, in STAR*D and PGRN-AMPS final test datasets.

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

Statistic

STAR*D

PGRN-AMPS

Balanced accuracy

60.5%

61.3%

p value (balanced accuracy)

<0.001

<0.01

Sensitivity

67%

75.5%

Specificity

54%

47.1%

Accuracy

59.8%

68.2%

PPV

59.3%

58.8%

NPV

62%

65.8%

  1. n = 251 (STAR*D), n = 132 (PGRN-AMPS).
Back to article page