Table 1 Multilevel Model Results of Passive Sensing Variables Predicting Daily GPT-Derived Activation.

From: Measuring activation during behavioral activation therapy: a proof-of-concept study using smartphone sensors and LLM-derived ratings in adolescents with anhedonia

Model

Predictor

N

Estimate (SE)

statistic

p value

95% CI

ActScore

(Intercept)

323

0.167 (4.172)

0.040

0.969

[−9.411, 9.745]

ActScore person-centered

323

0.178 (0.078)

2.27

0.024

[0.024, 0.332]

ActScore mean

323

0.780 (1.16)

0.672

0.519

[−1.880, 3.440]

homDist

(Intercept)

319

2.91 (0.064)

45.39

<0.001

[2.76, 3.05]

homDist person-centered

319

0.000 (0.000)

0.335

0.738

[−0.000, 0.000]

homDist mean

319

0.000 (0.000)

1.95

0.075

[−0.000, 0.000]

radiusMobility

(Intercept)

319

3.021 (0.136)

22.24

<0.001

[2.722, 3.321]

radiusMobility person-centered

319

0.001 (0.001)

0.924

0.356

[−0.001, 0.002]

radiusMobility mean

319

−0.002 (0.005)

−0.334

0.744

[−0.013, 0.009]

percentHome

(Intercept)

319

4.388 (0.793)

5.53

<0.001

[2.651, 6.125]

percentHome person-centered

319

−0.007 (0.003)

−2.53

0.012

[−0.012, −0.002]

percentHome mean

319

−0.018 (0.010)

−1.78

0.101

[−0.040, 0.004]

DayPlaces

(Intercept)

319

2.673 (0.319)

8.37

<0.001

[2.001, 3.345]

DayPlaces person-centered

319

0.045 (0.019)

2.31

0.021

[0.007, 0.083]

DayPlaces mean

319

0.077 (0.078)

0.978

0.340

[−0.087, 0.240]

  1. ActScore – Daily phone accelerometer activity score. homDist – Maximum distance (in km) from home traveled that day. radiusMobility – Radius (in km) covering all places visited in a day. percentHome – Percentage of time spent at home each day. DayPlaces – Number of distinct locations visited each day. Variables labeled person-centered reflect day-to-day fluctuations within individuals. Variables labeled mean reflect each participant’s average across the study period.
Back to article page