Table 3 Contributions of ML-predicted versus clinically rated BvA and alogia to cognitive/social functioning, beyond demographics (entered in step 1).

From: Using machine learning of computerized vocal expression to measure blunted vocal affect and alogia

 

DV: Cognitive functioning

DV: Social functioning

 

R2

F

B (se)

R2

F

B (se)

Symptom of interest: blunted vocal affect (BvA)

Unique contribution of Clin Rat BvA

Step 2: Predicted measure Only

0.08

4.13*

−0.53 (0.23)

0.11

4.62*

−0.66 (0.31)*

Step 3: Clin Rat measure only

0.07

0.47

−0.35 (0.33)

0.00

0.13

−0.05 (0.15)

Unique contribution of ML BvA

Step 2: Clin Rat measure only

0.07

4.65*

−0.18 (0.08)*

0.07

3.15+

−0.18 (0.10)+

Step 3: Predicted measure only

0.02

1.14

−0.35 (0.33)

0.04

1.58

−0.55 (0.44)

Symptom of interest: Alogia

Unique contribution of Clin Rat Alogia

Step 2: Predicted measure only

0.00

0.02

−0.03 (0.018)

0.10*

4.88*

−0.46 (0.21)*

Step 3: Clin Rat measure only

0.02

0.90

−0.17 (0.18)

0.01

0.55

−0.18 (0.24)

Unique contribution of ML alogia

Step 2: Clin Rat measure Only

0.01

0.74

−0.13 (0.14)

0.09*

4.11*

−0.36 (0.18)

Step 3: Predicted measure Only

0.01

0.18

0.09 (0.22)

0.03

1.32

−0.32 (0.28)

  1. Note: Step 1 demographics R2 = 0.19; Step 1 demographics R2 = 0.01.
  2. ML machine learning, BvA blunted vocal affect, Clin Rat clinical rating, se standard error.
  3. *p < 0.05.