Table 4 Performance of predicting non-remission status of MDD.

From: Multimodal digital assessment of depression with actigraphy and app in Hong Kong Chinese

 

No. of subjects (no. of MDD subjects)

No. of features

F1-score

Sensitivity

Specificity

PPV

NPV

Single modality of digital features

       

Subjective happiness levela

81 (16)

1

0.49

0.75

0.68

0.36

0.92

Actigraphyb

74 (13)

17

0.55

0.92

0.69

0.39

0.98

Facial expressionb

75 (16)

5

0.41

0.63

0.61

0.30

0.86

Voiceb

69 (16)

4

0.50

1.00

0.40

0.33

1.00

NLPb

70 (16)

2

0.37

1.00

0

0.23

/

All digital modalitiesb

60 (13)

29

0.64

0.69

0.87

0.60

0.91

All digital modalities + HADS-Db

56 (11)

30

0.70

0.64

0.96

0.78

0.91

HADS-Da

76 (14)

1

0.58

0.64

0.87

0.53

0.92

  1. NLP natural language processing, HADS-D Hospital Anxiety and Depression Scale—Depression subscale.
  2. aCutpoint optimization for continuous one-dimensional feature.
  3. bArtificial neural networks.