Table 10.
Performance of the elastic net models for predicting the Pittsburgh Sleep Quality Index global and component scores.
| Feature set | RMSEa | SD | R2 | SD | |||||
| PSQIb global score | |||||||||
|
|
Philips Actigraph sleep | 2.68 | 0.01 | –0.01 | 0.01 | ||||
|
|
Philips Actigraph activity | 2.67 | 0.01 | –0.01 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 2.67 | 0.01 | –0.01 | 0.01 | ||||
|
|
Garmin activity | 2.62 | 0.02 | 0.03 | 0.01 | ||||
|
|
Survey bidaily | 2.40c | 0.03 | 0.19 | 0.02 | ||||
|
|
Survey weekly | 2.41 | 0.02 | 0.18 | 0.02 | ||||
|
|
Survey (bidaily+weekly) | 2.41 | 0.03 | 0.18 | 0.02 | ||||
| PSQI subjective sleep quality | |||||||||
|
|
Philips Actigraph sleep | 0.76 | 0.01 | –0.01 | 0.01 | ||||
|
|
Philips Actigraph activity | 0.75 | 0.00 | –0.01 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 0.76 | 0.00 | –0.01 | 0.01 | ||||
|
|
Garmin activity | 0.76 | 0.00 | –0.01 | 0.01 | ||||
|
|
Survey bidaily | 0.73 | 0.01 | 0.05 | 0.02 | ||||
|
|
Survey weekly | 0.74 | 0.01 | 0.03 | 0.01 | ||||
|
|
Survey (bidaily+weekly) | 0.73 | 0.01 | 0.04 | 0.02 | ||||
| PSQI sleep latency | |||||||||
|
|
Philips Actigraph sleep | 1.64 | 0.01 | –0.01 | 0.01 | ||||
|
|
Philips Actigraph activity | 1.58 | 0.01 | 0.05 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 1.60 | 0.01 | 0.04 | 0.01 | ||||
|
|
Garmin activity | 1.64 | 0.01 | −0.01 | 0.01 | ||||
|
|
Survey bidaily | 1.62 | 0.01 | 0.02 | 0.01 | ||||
|
|
Survey weekly | 1.63 | 0.01 | −0.00 | 0.01 | ||||
|
|
Survey (bidaily+weekly) | 1.61 | 0.01 | 0.02 | 0.01 | ||||
| PSQI sleep duration | |||||||||
|
|
Philips Actigraph sleep | 1.43 | 0.02 | 0.13 | 0.02 | ||||
|
|
Philips Actigraph activity | 1.42 | 0.02 | 0.15 | 0.02 | ||||
|
|
Philips Actigraph (sleep+activity) | 1.44 | 0.01 | 0.12 | 0.02 | ||||
|
|
Garmin activity | 1.50 | 0.02 | 0.06 | 0.02 | ||||
|
|
Survey bidaily | 1.52 | 0.01 | 0.03 | 0.01 | ||||
|
|
Survey weekly | 1.55 | 0.01 | −0.01 | 0.01 | ||||
|
|
Survey (bidaily+weekly) | 1.52 | 0.01 | 0.03 | 0.01 | ||||
| PSQI habitual sleep efficiency | |||||||||
|
|
Philips Actigraph sleep | 7.38 | 0.02 | −0.00 | 0.01 | ||||
|
|
Philips Actigraph activity | 7.38 | 0.03 | −0.00 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 7.38 | 0.02 | −0.00 | 0.01 | ||||
|
|
Garmin activity | 7.38 | 0.02 | −0.00 | 0.01 | ||||
|
|
Survey bidaily | 7.38 | 0.03 | −0.00 | 0.01 | ||||
|
|
Survey weekly | 7.38 | 0.02 | −0.00 | 0.01 | ||||
|
|
Survey (bidaily+weekly) | 7.38 | 0.02 | −0.00 | 0.01 | ||||
| PSQI sleep disturbances | |||||||||
|
|
Philips Actigraph sleep | 3.76 | 0.08 | 0.05 | 0.04 | ||||
|
|
Philips Actigraph activity | 3.86 | 0.01 | −0.00 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 3.83 | 0.04 | 0.01 | 0.02 | ||||
|
|
Garmin activity | 3.85 | 0.02 | 0.00 | 0.01 | ||||
|
|
Survey bidaily | 3.34 | 0.04 | 0.25 | 0.02 | ||||
|
|
Survey weekly | 3.35 | 0.03 | 0.24 | 0.01 | ||||
|
|
Survey (bidaily+weekly) | 3.27 | 0.04 | 0.28 | 0.02 | ||||
| PSQI use of sleep medication | |||||||||
|
|
Philips Actigraph sleep | 1.35 | 0.00 | −0.00 | 0.01 | ||||
|
|
Philips Actigraph activity | 1.35 | 0.01 | −0.01 | 0.01 | ||||
|
|
Philips Actigraph (sleep+activity) | 1.35 | 0.01 | −0.01 | 0.01 | ||||
|
|
Garmin activity | 1.36 | 0.01 | −0.01 | 0.01 | ||||
|
|
Survey bidaily | 1.33 | 0.02 | 0.02 | 0.03 | ||||
|
|
Survey weekly | 1.32 | 0.01 | 0.03 | 0.02 | ||||
|
|
Survey (bidaily+weekly) | 1.34 | 0.02 | 0.02 | 0.02 | ||||
| PSQI daytime dysfunction | |||||||||
|
|
Philips Actigraph sleep | 0.96 | 0.01 | −0.04 | 0.01 | ||||
|
|
Philips Actigraph activity | 0.96 | 0.01 | −0.04 | 0.02 | ||||
|
|
Philips Actigraph (sleep+activity) | 0.97 | 0.01 | −0.04 | 0.01 | ||||
|
|
Garmin activity | 0.97 | 0.01 | −0.04 | 0.01 | ||||
|
|
Survey bidaily | 0.96 | 0.01 | −0.04 | 0.02 | ||||
|
|
Survey weekly | 0.95 | 0.03 | −0.01 | 0.06 | ||||
|
|
Survey (bidaily+weekly) | 0.96 | 0.02 | −0.03 | 0.04 | ||||
aRMSE: root mean square error.
bPSQI: Pittsburgh Sleep Quality Index.
cItalic values indicate significant correlation (P<.05).