Table 3 Weighted F1-scores (F1) for each classifier to distinguish PwMS from HC and the MS type (none, RRMS and PMS)

From: Modeling multiple sclerosis using mobile and wearable sensor data

 

PwMS vs Healthy (Binary)

MS Type (3-class)

Group

LR

FCNN

RF

XGB

LR

FCNN

RF

XGB

Behavioral

0.78 (0.10)

0.75 (0.12)

0.72 (0.12)

0.72 (0.11)

0.62 (0.22)

0.58 (0.18)

0.55 (0.22)

0.54 (0.20)

Physiological

0.73 (0.12)

0.70 (0.12)

0.68 (0.14)

0.68 (0.11)

0.51 (0.15)

0.49 (0.15)

0.47 (0.15)

0.48 (0.15)

Sleep routine

0.37 (0.21)

0.35 (0.20)

0.45 (0.11)

0.46 (0.14)

0.29 (0.12)

0.29 (0.12)

0.30 (0.12)

0.33 (0.12)

Performance

0.76 (0.14)

0.76 (0.15)

0.73 (0.12)

0.70 (0.13)

0.39 (0.13)

0.44 (0.14)

0.44 (0.15)

0.42 (0.15)

Beh+Phys

0.73 (0.12)

0.70 (0.13)

0.69 (0.14)

0.68 (0.12)

0.50 (0.16)

0.48 (0.16)

0.47 (0.15)

0.48 (0.16)

Beh+Sle

0.79 (0.11)

0.74 (0.12)

0.72 (0.12)

0.72 (0.11)

0.62 (0.22)

0.59 (0.19)

0.56 (0.22)

0.54 (0.21)

Beh+Per

0.77 (0.15)

0.77 (0.14)

0.73 (0.12)

0.70 (0.13)

0.40 (0.13)

0.44 (0.14)

0.43 (0.15)

0.41 (0.14)

Phy+Sle

0.73 (0.12)

0.69 (0.13)

0.69 (0.13)

0.69 (0.11)

0.50 (0.16)

0.48 (0.16)

0.45 (0.16)

0.49 (0.16)

Phy+Per

0.74 (0.11)

0.70 (0.12)

0.70 (0.13)

0.67 (0.13)

0.51 (0.16)

0.48 (0.16)

0.44 (0.15)

0.48 (0.16)

Per+Sle

0.76 (0.15)

0.76 (0.15)

0.73 (0.13)

0.70 (0.13)

0.39 (0.13)

0.44 (0.12)

0.44 (0.13)

0.43 (0.14)

Beh+Per+Phy

0.74 (0.12)

0.70 (0.13)

0.69 (0.13)

0.68 (0.11)

0.51 (0.16)

0.47 (0.16)

0.43 (0.15)

0.46 (0.17)

Per+Phy+Sle

0.74 (0.13)

0.70 (0.13)

0.68 (0.13)

0.69 (0.11)

0.50 (0.15)

0.46 (0.15)

0.45 (0.14)

0.46 (0.16)

Beh+Phy+Sle

0.73 (0.12)

0.70 (0.12)

0.68 (0.12)

0.68 (0.11)

0.50 (0.16)

0.48 (0.16)

0.45 (0.16)

0.47 (0.16)

All

0.82 (0.11)

0.82 (0.11)

0.78 (0.10)

0.78 (0.12)

0.60 (0.18)

0.60 (0.17)

0.55 (0.20)

0.51 (0.19)

BRG

0.56 (0.01)

0.36 (0.01)

RG

0.51 (0.04)

0.34 (0.16)

Demographics

0.66 (0.12)

0.43 (0.16)

  1. The classification metrics are reported as mean (standard deviation) of the stratified group five-fold cross-validation iterations and 50 runs. The bold values denote the highest F1-score for each classifier.