Table 1
Characteristics of the Participants.
| Total | Day Shift | Night Shift | Rotate Shift | p Value | |
|---|---|---|---|---|---|
| N | 14 | 5 | 5 | 4 | |
| Age | 44.21 [22, 67] | 47.4 [30, 64] | 36.8 [22, 67] | 49.5 [42, 63] | 0.398 |
| BMI | 27.6 ± 6.71 | 27.2 ± 7.33 | 28 ± 8.65 | 27.6 ± 4.81 | 0.984 |
| Female sex – no. (%) | 12 (85.7%) | 5 (100%) | 5 (100%) | 2 (50%) | |
| Race – no. (%) | 0.352 | ||||
| White | 11 (78.6%) | 3 (60%) | 5 (100%) | 3 (75%) | |
| Black | 3 (21.4%) | 2 (40%) | 0 (0%) | 1 (25%) | |
| Hispanic ethnicity | 0 | 0 | 0 | 0 | |
| Education | 0.599 | ||||
| High School | 1 (7.1%) | 0 (0%) | 1 (20%) | 0 (0%) | |
| College 2yrs | 7 (50%) | 2 (40%) | 2 (40%) | 3 (75%) | |
| College 4yrs | 6 (42.9%) | 3 (60%) | 2 (40%) | 1 (25%) | |
| Household Income | 0.866 | ||||
| 30–50k | 3 (21.4%) | 1 (20%) | 1 (20%) | 1 (25%) | |
| 50–100k | 7 (50%) | 3 (60%) | 3 (60%) | 1 (25%) | |
| >100k | 4 (28.6%) | 1 (20%) | 1 (20%) | 2 (50%) | |
| Comorbidity | |||||
| Hypertension | 2 (14.3%) | 1 (20%) | 0 (0%) | 1 (25%) | |
| Diabetes | 1 (7.1%) | 0 (0%) | 0 (0%) | 1 (25%) | |
| Metabolic syndrome | 2 (14.3%) | 0 (0%) | 0 (0%) | 2 (50%) | |
| Asthma | 1 (7.1%) | 0 (0%) | 1 (20%) | 0 (0%) | |
| Sleep History | |||||
| Sleep disorder | 2 (14.3%) | 1 (20%) | 0 (0%) | 1 (25%) | |
| Restless leg syndrome | 1 (7.1%) | 1 (20%) | 0 (0%) | 0 (0%) | |
| OSA | 2 (14.3%) | 1 (20%) | 0 (0%) | 1 (25%) | |
| Problem with sleep | 6 (42.9%) | 2 (40%) | 2 (40%) | 2 (50%) | |
| Snore | 5 (35.7%) | 3 (60%) | 1 (20%) | 1 (25%) | |
| Medication | |||||
| Benzodiazepine | 2 (14.3) | 1 (20%) | 1 (20%) | 0 (0%) |
Table 2
Correlation between daily totals of eating factors of different subjects. Strong correlations (highlighted) between calorie intake and fat/carbs/sugar are shown. As expected, BMI is moderately negatively correlated with daily activity. The number of bites is only moderately correlated with calories and other nutrients, with protein and fiber being slightly higher (as expected since meat/vegetables require more bites).
| BMI | Daily average number of bites | Daily average calorie intake | Daily average fat intake | Daily average carboh ydrate intake | Daily average protein intake | Daily average fiber intake | Daily average sugar intake | Daily average sodium intake | |
|---|---|---|---|---|---|---|---|---|---|
| Daily activity | –0.5392 | –0.1122 | –0.2569 | –0.2516 | –0.2022 | –0.4257 | –0.2835 | 0.2194 | –0.0135 |
| BMI | 0.2073 | 0.1963 | 0.2781 | 0.0237 | 0.4364 | 0.1309 | –0.1290 | 0.2619 | |
| Daily average number of bites | 0.4451 | 0.3876 | 0.4037 | 0.4967 | 0.4812 | 0.4456 | 0.2673 | ||
| Daily average calorie intake | 0.9247 | 0.9095 | 0.7577 | 0.7499 | 0.8097 | 0.5273 | |||
| Daily average fat intake | 0.7303 | 0.7342 | 0.8352 | 0.7870 | 0.7440 | ||||
| Daily average carbohydrate intake | 0.5291 | 0.4997 | 0.6786 | 0.1910 | |||||
| Daily average protein intake | 0.7831 | 0.6085 | 0.5452 | ||||||
| Daily average fiber intake | 0.7236 | 0.7409 | |||||||
| Daily average sugar intake | 0.6918 |
Figure 1
Violin plot of daily total bite-counter data for different groups. Dashed green lines represent mean. Solid yellow lines represent median. There are significant differences between bites for shift workers.
Figure 2
Violin plots of daily-average bite-counter data between working and non-working day for all groups (top) and for different groups (bottom sub-figures). Dashed green lines represent mean. Solid yellow lines represent median.
Figure 3
Violin plot of Fat Secret App daily calorie totals for different groups.
Figure 4
Violin plots of Fat Secret App calorie daily totals between working and non-working day for different groups.
Figure 5
Violin plot of snack calorie totals for different groups.
Figure 6
Violin plots of the snack calorie totals between working and non-working day for different groups.
Figure 7
Violin plot of activity data for different groups.
Figure 8
Violin plots of activity data between working and non-working day for different groups.
Figure 9
Violin plot of daily sleep hours for different groups.
Figure 10
Violin plots of daily sleep hours between working and non-working day for different groups.
Figure 11
Box plots of combined activity and sleep hours for different groups. Color red indicates working day, and color blue indicates non-working day. The saturated bars on the right are activity data (scale mentioned on the right) and washy bars on the left are sleep data (scale mentioned on the left).
Figure 12
P-Value after filtering out days with NaN density more than different threshold. Each x mark is the P-Value of KW test when days with NaN density larger than a certain threshold are filtered out. Orange line is the first order regression. The rightmost x mark corresponds to threshold = 1, which means that nothing is filtered out and even days with 100% NaNs are present in the analysis. As it can be seen, when these days are present, P-Value becomes small (p = 0.02).
Table 3
Missing data density and period of dominant frequency for each subject.
| Subjects | Labels | Missing Data Density | Period of dominant frequency of missing data (in hour) |
|---|---|---|---|
| Subject 1 | Night-Shift | 0.03016 | 30.54394 |
| Subject 2 | Day-Shift | 0.117265 | 111.9972 |
| Subject 4 | Night-Shift | 0.021869 | 23.925 |
| Subject 7 | Rotating-Shift | 0.10975 | 335.9917 |
| Subject 8 | Day-Shift | 0.021241 | 3.418793 |
| Subject 15 | Night-Shift | 0.057566 | 111.9972 |
| Subject 17 | Rotating-Shift | 0.284729 | 167.9958 |
| Subject 18 | Day-Shift | 0.042702 | 23.9756 |
| Subject 20 | Night-Shift | 0.136834 | 335.9917 |
| Subject 22 | Day-Shift | 0.310928 | 23.94524 |
| Subject 23 | Day-Shift | 0.805476 | 111.9972 |
| Subject 24 | Rotating-Shift | 0.047756 | 23.98095 |
| Subject 25 | Night-Shift | 0.051077 | 18.63565 |
| Subject 26 | Rotating-Shift | 0.42769 | 24.00641 |