Sensors | Free Full-Textual content | Evaluating Accuracy in 5 Industrial Sleep-Monitoring Gadgets In comparison with Analysis-Grade Actigraphy and Polysomnography
Writer Contributions
Conceptualization, methodology, software program, validation, formal evaluation, investigation, assets, information curation, writing—unique draft preparation, writing—reviewing and modifying, visualization, supervision, and mission administration: Ok.A.Ok.; conceptualization, methodology, investigation, and information curation: J.C.; information curation and investigation: M.B.; information curation and investigation: C.H.; information curation and investigation: A.B.; conceptualization, methodology, software program, assets, writing—unique draft preparation, writing—reviewing and modifying, visualization, supervision, and mission administration: R.M.C.S. All authors have learn and agreed to the printed model of the manuscript.
Determine 1.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for whole sleep time. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat; (F) Phillips Actiwatch. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 1.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for whole sleep time. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat; (F) Phillips Actiwatch. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 2.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for wake after sleep onset (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat; (F) Phillips Actiwatch. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 2.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for wake after sleep onset (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat; (F) Phillips Actiwatch. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 3.
Bland-Altmann plots depicting device-measured and PSG-measured settlement for gentle sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 3.
Bland-Altmann plots depicting device-measured and PSG-measured settlement for gentle sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 4.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for deep sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 4.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for deep sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 5.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for REM sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 5.
Bland–Altmann plots depicting device-measured and PSG-measured settlement for REM sleep. (A) Fitbit Encourage; (B) Fitbit Versa; (C) Garmin Vivosmart 4; (D) Oura Ring; (E) Withings Sleep Mat. Knowledge are proven in minutes. Pink strains = imply bias, gray strains = higher and decrease limits of settlement, dots = people.
Determine 6.
Imply absolute % error comparisons between device-measured and PSG-measured sleep variables. (A) Whole sleep time; (B) wake after sleep onset; (C) gentle sleep; (D) deep sleep; (E) REM sleep. Knowledge are proven as imply ± normal error.
Determine 6.
Imply absolute % error comparisons between device-measured and PSG-measured sleep variables. (A) Whole sleep time; (B) wake after sleep onset; (C) gentle sleep; (D) deep sleep; (E) REM sleep. Knowledge are proven as imply ± normal error.
Determine 7.
Abstract determine depicting absolute bias in sleep measurements for every gadget. TST = whole sleep time, WASO = wake after sleep onset, inexperienced bins = absolute bias < 30, yellow bins = absolute bias between 31 and 50, purple bins = absolute bias > 51, bins with dotted outlined = proportional bias.
Determine 7.
Abstract determine depicting absolute bias in sleep measurements for every gadget. TST = whole sleep time, WASO = wake after sleep onset, inexperienced bins = absolute bias < 30, yellow bins = absolute bias between 31 and 50, purple bins = absolute bias > 51, bins with dotted outlined = proportional bias.
Determine 8.
Abstract determine depicting intra-class correlation (ICC) values and 95% confidence intervals of settlement between device- and PSG-measured sleep variables. TST = whole sleep time, WASO = wake after sleep onset, gentle inexperienced bins = 0.90 > ICC > 0.75; good accuracy, yellow bins = 0.75 > ICC > 0.50; reasonable accuracy, purple bins = ICC < 0.50; poor accuracy.
Determine 8.
Abstract determine depicting intra-class correlation (ICC) values and 95% confidence intervals of settlement between device- and PSG-measured sleep variables. TST = whole sleep time, WASO = wake after sleep onset, gentle inexperienced bins = 0.90 > ICC > 0.75; good accuracy, yellow bins = 0.75 > ICC > 0.50; reasonable accuracy, purple bins = ICC < 0.50; poor accuracy.
Desk 1.
Details about price, location, and obtainable measurements for every gadget.
Desk 1.
Details about price, location, and obtainable measurements for every gadget.
| Price | Location | Measurements | |
|---|---|---|---|
| Fitbit Encourage HR | USD 97 | Wrist | Exercise, Sleep, Coronary heart Fee |
| Fitbit Versa 2 | USD 200 | Wrist | Exercise, Sleep, Coronary heart Fee, SpO2 |
| Garmin Vivosmart | USD 110 | Wrist | Exercise, Sleep, Coronary heart Fee, SpO2 |
| Oura Ring | USD 300 | Finger | Exercise, Sleep, Coronary heart Fee |
| Withings | USD 80 | Below Mattress | Exercise, Sleep, Coronary heart Fee |
| Actiwatch | USD 800 | Wrist | Exercise, Sleep |
Desk 2.
Demographics of the members included on this examine.
Desk 2.
Demographics of the members included on this examine.
| Whole | F:M | White | Asian | Hispanic | Black | DNR | |
|---|---|---|---|---|---|---|---|
| N | 53 | 31:22 | 29 | 11 | 7 | 3 | 3 |
Desk 3.
Abstract of participant traits.
Desk 3.
Abstract of participant traits.
| Age (Years) |
Top (m) |
Weight (kg) |
BMI (kg/m2) |
MEQ * | PSQI | |
|---|---|---|---|---|---|---|
| Imply (SD) | 22.5 (3.5) | 1.7 (0.10) | 69.4 (16.6) | 23.7 (5.0) | 45.4 (8.4) | 5.9 (2.8) |
| Median | 22.0 | 1.7 | 65.8 | 22.7 | 45.5 | 6.0 |
| Vary | 18–34 | 1.5–1.9 | 49.9–131 | 16.9–44.4 | 26–64 | 1–17 |
Desk 4.
Abstract statistics for sleep phases calculated by polysomnography in minutes.
Desk 4.
Abstract statistics for sleep phases calculated by polysomnography in minutes.
| TST | WASO | NREM1 | NREM2 | NREM3 | REM | |
|---|---|---|---|---|---|---|
| Imply (SD) | 431 (54.7) | 33.1 (22.6) | 15.7 (10.0) | 226 (38.4) | 113 (35.2) | 76.4 (25.2) |
Desk 5.
Gadget failures.
Desk 5.
Gadget failures.
| Gadget | % of Knowledge Misplaced |
N Errors Whole |
Knowledge Synch |
Poor Match | Low Battery |
|---|---|---|---|---|---|
| Fitbit Encourage | 5.6% | 3 | 1 | 2 | 0 |
| Fitbit Versa | 18.8% | 10 | 9 | 1 | 0 |
| Vivosmart | 5.6% | 3 | 2 | 1 | 0 |
| Oura Ring | 7.5% | 4 | 3 | 0 | 1 |
| Withings | 11.3% | 6 | 6 | 0 | 0 |
| Actiwatch | 15.1% | 8 | 1 | 7 | 0 |
Desk 6.
Settlement between gadgets and PSG for whole sleep time.
Desk 6.
Settlement between gadgets and PSG for whole sleep time.
| Gadget (min) |
PSG (min) |
Bias | LOA | ICC | |
|---|---|---|---|---|---|
| Encourage | 424.5 (66.4) | 433.5 (52.7) | −8.96 (29.7) | 0.15 | 0.84 (0.73–0.91) |
| Versa | 423.7 (57.7) | 434.9 (56.6) | −11.2 (27.6) | 0.14 | 0.86 (0.75–0.93) |
| Vivosmart | 476.6 (87.9) | 435.0 (53.5) | 41.7 (77.5) | 0.46 | 0.38 (0.09–0.60) |
| Oura Ring | 420.7 (70.6) | 434.6 (54.1) | −13.9 (41.1) | 0.23 | 0.77 (0.61–0.86) |
| Withings | 443.6 (58.9) | 436.0 (55.9) | 7.6 (30.8) | 0.16 | 0.85 (0.74–0.91) |
| Actiwatch | 449.0 (71.1) | 435.4 (56.7) | 13.6 (54.4) | 0.26 | 0.63 (0.41–0.79) |
Desk 7.
Bias and limits of settlement between device- and PSG-measured wake after sleep onset.
Desk 7.
Bias and limits of settlement between device- and PSG-measured wake after sleep onset.
| Gadget (min) |
PSG (min) |
Bias | LOA | ICC | |
|---|---|---|---|---|---|
| Encourage | 54.8 (20.1) | 32.7 (22.2) | 42.2 + −0.62 × ref | 0.99 | 0.27 (−0.06–0.55) |
| Versa | 58.5 (21.9) | 33.5 (24.3) | 45.5 + −0.61 × ref | 1.01 | 0.27 (−0.07–0.56) |
| Vivosmart | 7.2 (10.1) | 33.1 (22.9) | 4.0 + −0.9 × ref | 2.00 | 0.08 (−0.08–0.27) |
| Oura Ring | 58.8 (26.6) | 33.6 (23.2) | 39.6 + −0.43 × ref | 1.01 | 0.32 (−0.05–0.61) |
| Withings | 32.2 (33.2) | 33.6 (23.9) | 25.4 + −0.8 × ref | 2.00 | 0.15 (−0.15–0.43) |
| Actiwatch | 35.9 (21.1) | 34.7 (23.5) | 21.4 + −0.58 × ref | 1.10 | 0.47 (0.19–0.68) |
Desk 8.
Bias and limits of settlement between device- and PSG-measured gentle sleep.
Desk 8.
Bias and limits of settlement between device- and PSG-measured gentle sleep.
| Gadget (min) |
PSG (min) |
Bias | LOA | ICC | |
|---|---|---|---|---|---|
| Encourage | 256.8 (73.0) | 243.0 (42.2) | 13.7 (63.6) | 0.79 | 0.42 (0.16–0.63) |
| Versa | 246.2 (72.2) | 244.9 (43.1) | 1.3 (65.3) | 0.80 | 0.40 (0.11–0.64) |
| Vivosmart | 263.3 (51.7) | 243.0 (41.3) | 155 + −0.56 × ref | 0.41 | 0.32 (0.05–0.55) |
| Oura Ring | 193.5 (64.3) | 243.0 (42.2) | −49.5 (59.0) | 0.60 | 0.29 (−0.03–0.55) |
| Withings | 235.9 (58.5) | 243.6 (42.7) | 170 + −0.73 × ref | 0.51 | 0.19 (−0.11–0.47) |
Desk 9.
Bias and limits of settlement between device- and PSG-measured deep sleep.
Desk 9.
Bias and limits of settlement between device- and PSG-measured deep sleep.
| Gadget (min) |
PSG (min) |
Bias | LOA | ICC | |
|---|---|---|---|---|---|
| Encourage | 81.5 (52.3) | 117.7 (30.4) | 99.9 + −1.16 × ref | 1.10 | 0.00 (−0.21–0.24) |
| Versa | 92.2 (56.3) | 117.2 (30.8) | 109.3 + −1.14 × ref | 0.96 | 0.00 (−0.28–0.29) |
| Vivosmart | 120.3(110) | 117.9 (30.3) | 2.3 (108.2) | 1.99 | 0.14 (−0.15–0.40) |
| Oura Ring | 142.5(40.5) | 118.5 (30.9) | 94.9 + −0.6 × ref | 0.67 | 0.29 (0.01–0.53) |
| Withings | 122.8(48.1) | 121.1 (30.2) | 1.7 (48.6) | 0.93 | 0.31 (0.01–0.56) |
Desk 10.
Bias and limits of settlement between device- and PSG-measured REM sleep.
Desk 10.
Bias and limits of settlement between device- and PSG-measured REM sleep.
| Gadget (min) |
PSG (min) |
Bias | LOA | ICC | |
|---|---|---|---|---|---|
| Encourage | 86.8 (27.4) | 75.6 (24.2) | 11.2 (18.3) | 0.52 | 0.70 (0.40–0.84) |
| Versa | 85.8 (34.6) | 76.5 (26.6) | 9.3 (27.6) | 0.69 | 0.57 (0.32–0.75) |
| Vivosmart | 115.6 (81.8) | 76.3 (25.0) | 39.2 (82.3) | 2.00 | 0.06 (−0.16–0.31) |
| Oura Ring | 81.8 (38.9) | 75.9 (25.4) | 5.81 (37.7) | 1.11 | 0.34 (0.05–0.57) |
| Withings | 85.8 (41.0) | 74.2 (24.7) | 11.6 (39.7) | 1.93 | 0.30 (0.02–0.55) |