Table 4 The performance on binary class stress level prediction on the StudentLife dataset using chronological five-fold cross-validation.

From: Dynamic clustering via branched deep learning enhances personalization of stress prediction from mobile sensor data

Model

Precision

(@Recall\(\approx\)0.9)

PR AUC

ROC AUC

LSTM

0.798 ± 0.004

0.832 ± 0.009

0.579 ± 0.013

CALM-Net

0.835 ± 0.007

0.923 ± 0.002

0.779 ± 0.005

Branched CALM-Net

0.836 ± 0.006

0.925 ± 0.002

0.782 ± 0.004

  1. [bold] values indicate top performers in each category.
Back to article page