This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.
Polysomnography is the gold standard for measuring and detecting sleep patterns. In recent years, activity wristbands have become popular because they record continuous data in real time. Hence, comprehensive validation studies are needed to analyze the performance and reliability of these devices in the recording of sleep parameters.
This study compared the performance of one of the best-selling activity wristbands, the Xiaomi Mi Band 5, against polysomnography in measuring sleep stages.
This study was carried out at a hospital in A Coruña, Spain. People who were participating in a polysomnography study at a sleep unit were recruited to wear a Xiaomi Mi Band 5 simultaneously for 1 night. The total sample consisted of 45 adults, 25 (56%) with sleep disorders (SDis) and 20 (44%) without SDis.
Overall, the Xiaomi Mi Band 5 displayed 78% accuracy, 89% sensitivity, 35% specificity, and a Cohen κ value of 0.22. It significantly overestimated polysomnography total sleep time (
The Xiaomi Mi Band 5 can be potentially used to monitor sleep and to detect changes in sleep patterns, especially for people without sleep problems. However, additional studies are necessary with this activity wristband in people with different types of SDis.
ClinicalTrials.gov NCT04568408; https://clinicaltrials.gov/ct2/show/NCT04568408
RR2-10.3390/ijerph18031106
In recent years, technological advances have made it possible to carry out diverse daily tasks (eg, health management) more quickly, efficiently, and immediately [
Activity wristbands have become more popular among consumers because of their usefulness, affordability, and attractive design [
Activity wristbands can collect a myriad of daily health information in the user’s free-living environments [
Polysomnography (PSG) represents the reference method used to measure sleep patterns. Its objective is the diagnosis of sleep problems by assessing the quality and quantity of sleep [
On the basis of actigraphy, activity wristbands can combine movement signals from an accelerometer and heart rate (HR) variability from sensors to detect sleep-wake cycles [
Given the aforementioned limitations, sleep health entities, such as the American Academy of Sleep Medicine (AASM), consider it necessary to perform validation studies of activity devices to evaluate their performance and reliability against PSG, which is the gold standard [
Few studies have analyzed the sleep data performance of activity wristbands. Most of these studies focused on validation of the Fitbit [
Some of these validated devices can also identify sleep stages, although the Fitbit Charge 2 overestimates PSG N1+N2 stages of non-REM sleep and underestimates N3 stage of non-REM sleep [
According to the literature, it is necessary to continue testing the validity of wearable devices [
This study aimed to compare the performance of the Xiaomi Mi Band 5 in measuring the sleep-wake stages compared with PSG performed at a hospital sleep unit. The secondary objectives were (1) to determine the agreement between sleep measures from PSG and the Xiaomi Mi Band 5; (2) to assess the accuracy, specificity, and sensitivity for classifying sleep and wake stages by the Xiaomi Mi Band 5 compared with PSG; and (3) to determine the performance level of the Xiaomi Mi Band 5 for detecting sleep stages (wake,
This study was carried out from August 4, 2020, to December 10, 2021. Participants were recruited from the sleep unit of a hospital in A Coruña, Spain. These people were attending the sleep unit to participate in a PSG study, intended to detect possible sleep alterations, independent of this project. All participants had access to a study information sheet and provided written informed consent for their participation. The protocol study was registered with the ClinicalTrials.gov Protocol Registration and Results System (NCT04568408) and published in an international journal, where the design and recruitment process of the study are detailed [
A total of 58 people participated in the Xiaomi Mi Band 5 validation project. However, data from 13 (22%) of the 58 participants were not used in the sleep analysis owing to different factors, such as Xiaomi device malfunction (10/13, 77%), not meeting the inclusion criteria (2/13, 15%), and not performing a PSG (1/13, 8%). Therefore, the final sample comprised 45 people (n=23, 51%, men and n=22, 49%, women; aged 23-81 [mean 53.24 SD 15.44] years; BMI mean 27.86, SD 4.44 kg/m2). Of these 45 participants, 25 (56%) were diagnosed with SDis after they had undergone PSG. The sleep diagnoses were OSA syndrome (18/25, 72%), insomnia (4/25, 16%), narcoleptic syndrome (1/25, 4%), a combination of hypersomnia and narcoleptic syndrome (1/25, 4%), and a combination of sleep apnea syndrome and hypoventilation syndrome (1/25, 4%). For this reason, the sleep measures were also analyzed in 2 groups according to whether there were SDis. The demographic characteristics of the No SDis (20/45, 44%) and SDis (25/45, 56%) groups are shown in
Sample characteristics (n=45).
Characteristics | No SDisa group (n=20) | SDis group (n=25) | |||
|
|||||
|
Female | 12 (60) | 10 (40) | ||
|
Male | 8 (40) | 15 (60) | ||
Age (years), mean (SD) | 49.7 (14.22) | 56.08 (16.06) | |||
BMI (kg/m2), mean (SD) | 26.55 (3.67) | 28.91 (4.79) | |||
PSQIb, mean (SD) | 10.30 (4.66) | 9.8 (4.07) |
aSDis: sleep disorders.
bPSQI: Pittsburgh Sleep Quality Index.
All participants slept in the sleep unit for 1 night. On the day of the recording, participants did not drink liquids for 3 hours before PSG, and they attended the sleep unit a few hours earlier to become acquainted with the bedroom. The technical team members of the sleep unit were in charge of supervising, preparing the participants for PSG, and meeting their possible demands. Regarding PSG, the technicians placed the sensors with electrode gel on the participants and connected them to start the test. Participants wore the Xiaomi Mi Band 5 during the recording test, and its location on the wrist was noted by the technicians. During the registration, the technical team members supervised the PSG and the Xiaomi device worn by the participants, making notes and marking alterations that emerged throughout the night for subsequent analysis. The lights-off and lights-on times, temperature, and sound were controlled in the room by the technicians. PSG and Xiaomi data were collected and synchronized simultaneously. The data coincided with the lights-off and lights-on times, providing a record of approximately 8 hours of time in bed (TIB).
PSG was performed using the NicoletOne v44 Sleep Diagnostic System (Natus Medical Incorporated). This device uses several recordings that included an electroencephalogram (EEG; 6 leads: FP1/FP2, F3/F4, C3/C4, and O1/O2 referenced by the contralateral mastoid), a submental (P3 and P4) and bilateral anterior tibial (2 electrodes on each leg to assess leg movements) electromyogram (EMG), a bilateral electro-oculogram (EOG), and an electrocardiogram (ECG) [
The Xiaomi Mi Band 5 includes a 3-axis accelerometer, a 3-axis gyroscope, an HR sensor, and a photoplethysmography sensor to measure some biomedical parameters. This device contains updated software that continuously records daily activity (eg, steps, distance, activity time, and calories), sleep (eg,
The Xiaomi Mi Band 5 and PSG source data were originally available in different formats. The Xiaomi Mi Band 5 data were collected from the Zepp Life app manually and exported to an Excel sheet (Microsoft Corp) because it did not allow downloading of the raw data from the wristband. By contrast, PSG data consisted of CSV text files that contained the manually scored sleep stages and the corresponding reports with clinical sleep diagnostic parameters, available in Word format (Microsoft Corp), both exported using the NicoletOne software. Hence, to enable performance analysis, data were converted into a common format using the European Data Format + (EDF+) [
To carry out this process,
After this process, and using the corresponding EDF+ annotation files, different standard sleep parameters were calculated and compared [
Statistical analysis was performed using R software (version 4.1.2; R Foundation for Statistical Computing). The analysis was carried out with the total sample (n=45). Likewise, data were analyzed with the participants grouped depending on the existence of SDis to determine the differences between PSG and the Xiaomi Mi Band 5.
Summary measures of PSG and the Xiaomi Mi Band 5 equivalents were compared using the paired 2-tailed
Furthermore, the Bland-Altman method was used to determine the agreement between PSG and the Xiaomi Mi Band 5 for each sleep parameter. The mean difference (or bias) between the methods, the SD, the 95% CI, and the Bland-Altman 95% limits of agreement (mean observed difference ± 1.96 × SD of observed differences) were calculated. A positive bias indicates that the Xiaomi Mi Band 5 tends to underestimate a variable when compared with the gold standard (PSG). A negative bias indicates that a sleep variable is overestimated [
EBE analysis was performed according to the 2 levels of compliance as provided in the ANSI and CTA performance evaluation guidelines [
The study was approved by the A Coruña-Ferrol research ethics committee (2020/318).
The sleep parameters obtained from PSG and the Xiaomi Mi Band 5 were compared using the 2-tailed
Comparison of polysomnography (PSG) and Xiaomi Mi Band 5 sleep measures in the total sample.
|
PSG, mean (SD; 95% CI) | Xiaomi Mi Band 5, mean (SD; 95% CI) | Cohen |
|||
Lights on (hh:mm) | 07:01 (00:17; 06:56-07:10) | —b | — | — | — | — |
Lights off (hh:mm) | 23:36 (00:40; 22:29-01:03) | — | — | — | — | — |
Initial sleep onset (hh:mm) | 00:29 (01:31; 22:47-01:33) | 00:11 (01:05; 23:55-00:31) | 1.12 (44) | — | .27 | 0.208c |
TIBd (min) | 443.58 (44.98; 430.07 to457.1) | — | — | — | — | — |
TSPDe, (min) | 408.05 (57.13; 390.88-425.23) | 406.16 (51.27; 390.75-421.56) | 0.28 (44) | — | .78 | 0.250c |
TSTf (min) | 344.62 (79.58; 320.71-368.53) | 374.17 (72.42; 352.42-395.93) | — | −2.73 | .009 | −0.407c |
WASOg (min) | 63.42 (57.42; 46.17 to80.68) | 31.99 (51.64; 16.47-47.50) | — | 2.96 | .005 | 0.442c |
Awakenings (>5 min; number per night) | 3.64 (2.27; 4.33-2.96) | 0.69 (0.94; 0.97-0.40) | — | 7.72 | .001 | 1.15h |
SOLi (min) | 31.64 (33.97; 21.43-41.84) | 40.26 (42.59; 27.47-53.06) | — | −1.07 | .29 | −0.206c |
SEj (%) | 78.32 (16.56; 73.35-83.30) | 84.14 (15.70; 79.42-88.86) | — | −2.59 | .03 | −0.329c |
Time in N1 stage of non-REMk sleep (min) | 12.65 (9.21; 9.9-15.42) | — | — | — | — | — |
Time in N2 stage of non-REM sleep (min) | 202.14 (57.47; 184.87-219.40) | — | — | — | — | — |
Time in N1+N2 sleep (light sleep; min) | 214.79 (55.12; 198.23-231.35) | 244.62 (56.59; 227.61-261.62) | −2.49 (44) | — | .005 | −0.439c |
Time in N3 sleep (deep sleep; min) | 60.72 (29.41; 51.88-69.56) | 75.37 (31.73; 65.83-84.90) | −2.58 (44) | — | .01 | −0.385c |
Time in REM sleep (min) | 69.11 (28.08; 60.67-77.54) | 49.61 (30.7; 40.01-59.22) | 3.59 (44) | — | .001 | 0.536l |
Awake (min) | 94.86 (71.88; 116.46-73.27) | 66.51 (66.22; 86.40-46.61) | — | 2.61 | .01 | 0.390c |
aMann-Whitney Wilcoxon test.
bNot available.
cSmall effect.
dTIB: time in bed.
eTSPD: total sleep period duration.
fTST: total sleep time.
gWASO: wake after sleep onset.
hLarge effect.
iSOL: sleep onset latency.
jSE: sleep efficiency.
kREM: rapid eye movement.
lModerate effect.
Comparison of polysomnography (PSG) and Xiaomi Mi Band 5 sleep measures in the no sleep disorders (No SDis) and sleep disorders (SDis) groups.
|
|
PSG, mean (SD; 95% CI) | Xiaomi Mi Band 5, mean (SD; 95% CI) | Cohen |
|||
|
|||||||
|
No SDis group | 07:02 (00:11; 06:57-07:07) | —b | — | — | — | — |
|
SDis group | 07:01 (00:45; 06:52-07:10) | — | — | — | — | — |
|
|||||||
|
No SDis group | 23:32 (00:48; 23:09-23:54) | — | — | — | — | — |
|
SDis group | 23:42 (00:32; 23:28-23:55) | — | — | — | — | — |
|
|||||||
|
No SDis group | 00:37 (02:00; 23:41-01:33) | 00:10 (00:56; 23:43-00:37) | 0.91 (19) | — | .37 | 0.204c |
|
SDis group | 00:22 (01:02; 23:56-00:48) | 00:12 (01:13; 23:42-00:43) | 0.64 (24) | — | .52 | 0.230c |
|
|||||||
|
No SDis group | 446.66 (50.37; 423.09-470.24) | — | — | — | — | — |
|
SDis group | 441.12 (41.06; 424.16-458.07) | — | — | — | — | — |
|
|||||||
|
No SDis group | 442.80 (42.35; 402.97-442.62) | 410.56 (50.66; 386.84-434.27) | 2.12 (19) | — | .47 | 0.475c |
|
SDis group | 396.26 (65.11; 369.38-423.13) | 402.64 (52.51; 380.96-424.32) | −0.58 (24) | — | .57 | −0.508f |
|
|||||||
|
No SDis group | 363.55 (68.20; 331.63-395.47) | 378.26 (70.98; 345.04-411.49) | −0.83 (19) | — | .41 | −0.208c |
|
SDis group | 329.48 (85.97; 293.99-364.97) | 370.90 (74.84; 340.00-401.79) | — | −2.70 | .007 | −0.625f |
|
|||||||
|
No SDis group | 59.25 (56.44; 32.83-85.66) | 32.30 (54.31; 6.87-57.71) | — | 1.98 | .12 | 0.361c |
|
SDis group | 66.77 (59.14; 42.36-91.18) | 31.75 (50.53; 10.89-52.61) | — | 2.35 | .02 | 0.504f |
|
|||||||
|
No SDis group | 3.90 (2.38; 5.01-2.78) | 0.65 (0.98; 1.11-0.18) | — | 5.50 | <.001 | 1.20i |
|
SDis group | 3.44 (2.22; 4.35-2.52) | 0.72 (0.93; 1.10-0.33) | — | 5.35 | <.001 | 1.07i |
|
|||||||
|
No SDis group | 26.47 (15.19; 19.36-33.58) | 38.75 (32.46; 23.56-53.95) | — | −1.83 | .05 | −0.478c |
|
SDis group | 35.78 (43.50; 17.82-53.73) | 41.48 (49.88; 20.89-62.07) | — | −0.25 | .80 | −0.474c |
|
|||||||
|
No SDis group | 81.02 (13.67; 74.63-87.42) | 84.61 (15.24; 77.47-91.74) | — | −1.46 | .37 | −0.206c |
|
SDis group | 76.17 (18.56; 68.51-83.83) | 83.77 (16.36; 77.02-90.53) | — | −2.19 | .03 | −0.421c |
|
|||||||
|
No SDis group | 13.85 (10.05; 9.14-18.55) | — | — | — | — | — |
|
SDis group | 11.70 (8.56; 8.16-15.23) | — | — | — | — | — |
|
|||||||
|
No SDis group | 198.74 (52.72; 174.06-223.42) | — | — | — | — | — |
|
SDis group | 204.86 (61.95; 179.28-230.43) | — | — | — | — | — |
|
|||||||
|
No SDis group | 212.59 (48.09; 190.08-235.10) | 254.63 (52.78; 229.93-279.33) | −3.23 (19) | — | .004 | −0.724f |
|
SDis group | 216.56 (61.09; 191.34-241.77) | 236.59 (59.29; 212.12-261.07) | −1.34 (24) | — | .19 | −0.268c |
|
|||||||
|
No SDis group | 77.05 (20.68; 67.37-86.73) | 73.88 (30.54; 59.59-88.17) | 0.42 (19) | — | .67 | −0.345c |
|
SDis group | 47.66 (29.11; 35.64-59.67) | 76.55 (33.22; 62.84-90.27) | −3.97 (24) | — | <.001 | −0.796f |
|
|||||||
|
No SDis group | 73.90 (23.56; 62.88-84.94) | 49.85 (27.20; 37.12-62.58) | 2.66 (19) | — | .02 | 0.597f |
|
SDis group | 65.26 (31.17; 52.39-78.13) | 49.43 (35.89; 34.62-64.24) | 2.37 (24) | — | .03 | 0.475c |
|
|||||||
|
No SDis group | 85.72 (62.81; 115.12-56.33) | 71.01 (73.04; 105.21-36.83) | — | 0.83 | .42 | 0.218c |
|
SDis group | 102.18 (78.90; 134.75-69.61) | 60.77 (62.39; 86.53-35.01) | — | 3.13 | .005 | 0.625f |
aMann-Whitney Wilcoxon test.
bNot available.
cSmall effect.
dTIB: time in bed.
eTSPD: total sleep period duration.
fModerate effect.
gTST: total sleep time.
hWASO: wake after sleep onset.
iLarge effect.
jSOL: sleep onset latency.
kSE: sleep efficiency.
lREM: rapid eye movement.
There were also differences between the sleep groups. The Xiaomi Mi Band 5 significantly overestimated PSG
On average, the Bland-Altman agreement limits were exceeded every 4 and 2 participants for the total sample, especially in TST, WASO, awakenings, and SE measures. The participants with sleep problems were the ones who mainly exceeded these aggregation limits.
Bland-Altman parameters for the comparison between polysomnography and the Xiaomi Mi Band 5 in the total sample as well as the no sleep disorders (No SDis) and sleep disorders (SDis) groups.
|
Bias (SD; 95% CI) | Agreement limits | Number of participants exceeding the agreement limits, n (%) | |
|
||||
|
Total sample | 00:17 (01:44; 00:49 to 00:13) | −03:06 to 03:40 | 2 (4)a |
|
No SDis group | 00:27 (02:13; 01:29 to 00:35) | −01:36 to 02:30 | 2 (10)b |
|
SDis group | 00:09 (01:17; 00:41 to 00:21) | −02:22 to 02:40 | 0 (0)c |
|
||||
|
Total sample | −29.54 (72.54; −7.75 to 51.33) | −171.72 to 112.63 | 3 (7)a |
|
No SDis group | −14.71 (78.94; −51.66 to 22.23) | −169.44 to 140.01 | 1 (5)b |
|
SDis group | −41.41 (66.20; −68.73 to 14.08) | −171.18 to 88.36 | 2 (8)c |
|
||||
|
Total sample | 1.89 (44.97; −11.61 to 15.40) | −86.25 to 90.04 | 1 (2)a |
|
No SDis group | 12.24 (25.75; 0.19 to 24.29) | −38.24 to 62.73 | 0 (0)b |
|
SDis group | −6.38 (54.97; −29.07 to 16.31) | −114.14 to 101.37 | 1 (4)c |
|
||||
|
Total sample | 31.44 (71.13; 10.07 to 52.81) | −107.97 to 170.86 | 4 (9)a |
|
No SDis group | 26.96 (61.92; −8.00 to 61.93) | −119.47 to 173.40 | 1 (5)b |
|
SDis group | 35.03 (69.48; 6.35 to 63.71) | −101.15 to 171.21 | 3 (12)c |
|
||||
|
Total sample | 2.95 (2.57; 3.73 to 2.18) | −2.07 to 3.73 | 4 (9)a |
|
No SDis group | 3.25 (2.63; 4.48 to 2.01) | −1.91 to 8.41 | 3 (15)b |
|
SDis group | 2.72 (2.54; 3.77 to 1.67) | −2.26 to 7.70 | 1 (4)c |
|
||||
|
Total sample | −8.62 (53.76; −24.77 to 7.53) | −114.00 to 96.75 | 2 (4)a |
|
No SDis group | −12.28 (25.69; −24.31 to 0.26) | −62.65 to 38.08 | 0 (0)b |
|
SDis group | −5.7 (68.97; −34.17 to 22.77) | −140.88 to 129.48 | 2 (8)c |
|
||||
|
Total sample | −5.82 (17.67; −11.13 to 0.51) | −40.46 to 28.82 | 4 (9)a |
|
No SDis group | −3.85 (17.37; −11.71 to 4.54) | −37.63 to 30.46 | 2 (10)b |
|
SDis group | −7.60 (18.06; −15.06 to 0.14) | −43.00 to 27.80 | 2 (8)c |
|
||||
|
Total sample | −29.81 (67.98; −50.24 to 9.39) | −163.07 to 103.44 | 2 (4)a |
|
No SDis group | −42.04 (58.04; −69.20 to 14.87) | −155.81 to 71.72 | 0 (0)b |
|
SDis group | −20.03 (74.72; −50.88 to 10.81) | −166.49 to 126.42 | 2 (8)c |
|
||||
|
Total sample | −14.64 (59.97; −26.08 to 3.20) | −89.27 to 59.98 | 0 (0)a |
|
No SDis group | 3.17 (33.00; −12.28 to 18.62) | −61.53 to 67.86 | 0 (0)b |
|
SDis group | −28.89 (36.32; −43.88 to 13.90) | −100.08 to 36.32 | 0 (0)c |
|
||||
|
Total sample | 19.49 (36.40; 8.55 to 30.42) | −51.85 to 90.84 | 2 (4)a |
|
No SDis group | 24.05 (40.30; 5.20 to 42.92) | −54.93 to 103.05 | 0 (0)b |
|
SDis group | 15.83 (33.34; 2.07 to 29.60) | −49.52 to 81.20 | 2 (8)c |
|
||||
|
Total sample | 28.36 (72.69; 50.20 to 6.52) | −114.11 to 170.84 | 3 (7)a |
|
No SDis group | 14.71 (78.93; 51.65 to 22.23) | −140.00 to 169.43 | 2 (10)b |
|
SDis group | 41.41 (66.21; 14.08 to 68.74) | −88.36 to 171.18 | 1 (4)c |
an=45.
bn=20.
cn=25.
dTST: total sleep time.
eTSPD: total sleep period duration.
fWASO: wake after sleep onset.
gSOL: sleep onset latency.
hSE: sleep efficiency.
iREM: rapid eye movement.
Bland-Altman plots for initial sleep onset, total sleep time (TST), total sleep period duration (TSPD), wake after sleep onset (WASO), awakenings, sleep onset latency (SOL), sleep efficiency (SE), light sleep, deep sleep, rapid eye movement (REM) sleep, and awake time. The PSG-Xiaomi Mi Band 5 differences for sleep parameters (y-axis) are plotted as a function of the PSG-Xiaomi Mi Band 5 means (x-axis) for sleep parameters. Circles represent participants without sleep disorders (No SDis group; n=20), and triangles represent participants with sleep disorders (SDis group; n=25). Zero lines are marked and represent perfect agreement. The dotted lines represent the biases and Bland-Altman 95% limits of agreement (mean observed difference ± 1.96 × SD of observed differences). PSG: polysomnography.
Confusion matrix for the 2-way (wake vs sleep) epoch-by-epoch classification for the total sample.
|
Xiaomi Mi Band 5 | ||
|
Wake | Sleep | |
|
|||
|
Wake | 3019 | 5525 |
|
Sleep | 3238 | 27,786 |
aPSG: polysomnography.
Confusion matrix for the 2-way (wake vs sleep) epoch-by-epoch classification for the no sleep disorders group.
|
Xiaomi Mi Band 5 | ||
|
Wake | Sleep | |
|
|||
|
Wake | 1305 | 2117 |
|
Sleep | 1528 | 13,018 |
aPSG: polysomnography.
Confusion matrix for 2-way (wake vs sleep) epoch-by-epoch classification for the sleep disorders group.
|
Xiaomi Mi Band 5 | ||
|
Wake | Sleep | |
|
|||
|
Wake | 1714 | 3408 |
|
Sleep | 1710 | 14,786 |
aPSG: polysomnography.
According to the results, the Xiaomi Mi Band 5 correctly classified both sleep and wake epochs in 30,805 (77.85%) of the 39,568 available epochs. It correctly detected 27,786 (89.56%) of the 31,024 sleep epochs, thus resulting in 0.90 sensitivity for the sleep class, and it was able to identify 3019 (35.33%) of the 8544 wake stages, leading to a corresponding 0.35 specificity for the wake class. Because of the binary classification, we derive immediately the respective sensitivity and specificity values for the wake class as 0.35 and 0.90. On the basis of the Cohen κ value, the level of concordance between PSG and the Xiaomi Mi Band 5 was 0.22.
These results were similar in the No SDis and SDis groups. The Xiaomi Mi Band 5 had an accuracy of 0.80 and 0.76 in the No SDis group and the SDis group, respectively. The sensitivity for the sleep class was 0.89 for both groups, and the corresponding sensitivity was 0.38 for the No SDis group and 0.33 for the SDis group. The Cohen κ values for the Xiaomi Mi Band 5 were 0.27 in the No SDis group and 0.26 in the SDis group.
Confusion matrix for 4-way (wake, light sleep, deep sleep, and rapid eye movement [REM] sleep) epoch-by-epoch classification for the total sample.
|
Xiaomi Mi Band 5 | ||||||||
|
Wake | Light sleep | Deep sleep | REM | |||||
|
|||||||||
|
Wake | 3019 | 3957 | 1020 | 548 | ||||
|
Light sleep | 2236 | 11,366 | 3468 | 2274 | ||||
|
Deep sleep | 537 | 2653 | 1775 | 494 | ||||
|
REM | 465 | 4044 | 519 | 1193 |
aPSG: polysomnography.
Confusion matrix for 4-way (wake, light sleep, deep sleep, and rapid eye movement [REM] sleep) epoch-by-epoch classification for the no sleep disorders group.
|
Xiaomi Mi Band 5 | ||||||||
|
Wake | Light sleep | Deep sleep | REM | |||||
|
|||||||||
|
Wake | 1305 | 1630 | 272 | 215 | ||||
|
Light sleep | 872 | 5142 | 1438 | 1055 | ||||
|
Deep sleep | 422 | 1469 | 982 | 209 | ||||
|
REM | 234 | 1945 | 263 | 515 |
aPSG: polysomnography.
Confusion matrix for 4-way (wake, light sleep, deep sleep, and rapid eye movement [REM] sleep) epoch-by-epoch classification for the sleep disorders group.
|
Xiaomi Mi Band 5 | ||||||||
|
Wake | Light sleep | Deep sleep | REM | |||||
|
|||||||||
|
Wake | 1714 | 2327 | 748 | 333 | ||||
|
Light sleep | 1364 | 6224 | 2030 | 1219 | ||||
|
Deep sleep | 115 | 1184 | 793 | 285 | ||||
|
REM | 231 | 2099 | 256 | 678 |
aPSG: polysomnography.
Overall accuracy and Cohen κ statistics for 2-way and 4-way epoch-by-epoch classifications for the total sample as well as the no sleep disorders (No SDis) and sleep disorders (SDis) groups.
|
2-Way epoch-by-epoch classification | 4-Way epoch-by-epoch classification | |||
|
Accuracy, mean (SD) | Cohen κa, mean (SD) | Accuracy, mean (SD) | Cohen κ, mean (SD) | |
Total sample | 0.78 (0.13) | 0.22 (0.23) | 0.44 (0.10) | 0.12 (0.13) | |
No SDis | 0.80 (0.13) | 0.27 (0.21) | 0.45 (0.10) | 0.15 (0.12) | |
SDis | 0.76 (0.12) | 0.26 (0.25) | 0.43 (0.10) | 0.11 (0.14) |
aCohen κ: 0 to 0.2 (slight), 0.21 to 0.40 (fair), 0.41 to 0.60 (moderate), 0.61 to 0.80 (substantial), and >0.80 (almost perfect).
The Xiaomi Mi Band 5 is one of the most popular wristbands among consumers around the world [
This study investigated the agreement in sleep measures from PSG and the Xiaomi Mi Band 5. Overall, the Xiaomi Mi Band 5 had some limitations in the detection of several sleep measures. There were no significant differences detected among
The Bland-Altman analysis showed the biases between the Xiaomi Mi Band 5 and PSG in general, which ranged from 1.89 to 31.44 minutes. Unlike other devices, the Xiaomi Mi Band 5 more accurately estimated some summary measures of sleep compared with PSG (
Furthermore, 2 studies evaluated previous versions of the Xiaomi wristband against other devices [
The Bland-Altman 95% limits of agreement (
The Xiaomi Mi Band 5 showed an accuracy of 78% for identifying sleep and wake stages and a sensitivity of 89% for detecting sleep epochs. However, it showed a specificity of only 35%. These findings are similar to those of previous studies, highlighting that, in general, these devices have high accuracy and sensitivity but low specificity [
Furthermore, this study analyzed the Xiaomi Mi Band 5’s level of performance regarding the identification of sleep stages using a 4-way EBE classification. The device obtained an accuracy of 44% for this task. Specifically, the Xiaomi Mi Band 5 was more accurate in detecting wake (48%) and
Cohen κ corrects the agreement owing to chance between the Xiaomi Mi Band 5 and PSG. Overall, the obtained values hovered between 0.11 and 0.27 for the different 2-way (wake vs sleep) and 4-way stage classifications and patient groups. Thus, the levels of agreement between the Xiaomi Mi Band 5 and PSG were slight to fair. Conversely, other authors reported that the Fitbit Alta HR and the Fitbit Charge 2 devices had Cohen κ coefficients ranging from 0.52 to 0.66, indicating a moderate agreement with PSG [
Experts who have validated other devices have concluded the need to focus on people with sleep problems owing to their increased prevalence during the COVID-19 pandemic [
The estimations of these sleep measures are consistent with those made by the Jawbone or Fitbit devices in a group of people with SDis [
Moreover, both groups presented similar results regarding the performance of sleep and wake stage detection, but participants with SDis presented lower values. Overall, Cohen κ coefficients were also lower in the SDis group. Specifically, the Xiaomi Mi Band 5 misidentified more epochs in the SDis group than in the No SDis group. The device misclassified
Overall, the outcomes obtained from people without SDis were a bit more accurate in some sleep measures and sleep stages. Consistent with the literature, devices such as the Xiaomi Mi Band 5 may be an alternative for health management in people without SDis because the data are more reliable than in people with SDis [
However, the outcomes of the SDis group could have been influenced by the inclusion of multiple SDis (rather than a single disorder), with OSA being the most prevalent syndrome. Similar to the results of other studies, the performance of this activity wristband can be lower among people diagnosed with OSA. There are reports that devices such as the Xiaomi Mi Band 5 had worse outcomes in this population than in populations with other conditions [
There are some limitations that could have negatively influenced the main findings of this study. The first limitation is attrition: only 45 of the 58 participants completed the study. Participants did not complete the study owing to difficulties with the use of the wristband and data collection. Hence, the sample was heterogeneous, and the size of the groups (SDis and No SDis) was small. In addition, how the participants were monitored might have influenced the data because this was not done in the usual context in which people sleep. Moreover, participants should be followed up for more days for better assessment of the performance of the Xiaomi Mi Band 5.
This study includes other limitations. Specifically, the Xiaomi Mi Band 5 and other devices only combine movement and HR to classify sleep parameters, whereas PSG includes several sensors; therefore, the accuracy of the data collected by wearable devices could be lower than that of the data collected by PSG. These devices present more limitations in the detection of WASO,
In conclusion, the very popular Xiaomi Mi Band 5 may be an acceptable activity wristband in terms of quality and price. Moreover, its use can promote greater awareness of the importance of sleep and promote good healthy lifestyle habits so that people obtain more quality sleep. Likewise, this device could be considered a tool to monitor sleep and to screen changes in sleep patterns through which health professionals could determine the quality and quantity of people’s sleep. Specifically, it could be a potential tool for use in populations without SDis, especially to identify TST and
Summary of the main findings of the Xiaomi Mi Band 5 and other devices.
Analysis of data according to the type of sleep disorder.
American Academy of Sleep Medicine
American National Standards Institute
Consumer Technology Association
epoch-by-epoch
electrocardiogram
European Data Format +
electroencephalogram
electromyogram
electro-oculogram
heart rate
obstructive sleep apnea
polysomnography
rapid eye movement
sleep disorders
sleep efficiency
sleep onset latency
time in bed
total sleep period duration
total sleep time
wake after sleep onset
The authors would like to express their gratitude to the participants who kindly agreed to be part of this study, as well as to the clinical staff working at the sleep unit with whom they have worked closely. The authors disclose the receipt of the following financial support for the research, authorship, and publication of this paper: all economic costs involved in the study were borne by the research team. This work is supported in part by some grants from the European Social Fund 2014-2020: Centre for Information and Communications Technology Research (CITIC; Research Center of the Galician University System). Grant support for CITIC was provided by the Xunta de Galicia through a collaboration agreement between the Regional Ministry of Culture, Education, and Vocational Training and the Galician Universities for the work of the Research Centers of the Galician University System and the Handytronic chair. PCM obtained a scholarship to develop the PhD thesis (ED481A-2019/069). DAE received funding from the project ED431H 2020/10 of the Xunta de Galicia. In addition, this work is supported in part by structural support for the consolidation and structuring of competitive research units and other promotion actions in the universities of the Galician University System, in the public research organizations of Galicia, and other entities of the Galician R&D&I System for 2022 (ED431B-2022/39); “Quality of life for caregivers through a person-centered technological solution” (TED2021-130127A-I00) associated with the “projects oriented to the ecological transition and digital transition” in the framework of the state program to promote scientific-technical research and its transfer, according to the state plan for scientific, technical, and innovation research 2021-2023; and the Ministerio de Ciencia e Innovación R&D&I projects in the framework of the national programs of knowledge generation and scientific and technological strengthening of the R&D&I system and challenges of the call R&D&I 2019 oriented to society (PID2019-104323RB-C33).
The study was conceptualized by JP, PCM, and BG. JP, FJMM, PCM, DAE, and BG were responsible for the methodology. JP, FJMM, PCM, DAE, and BG were responsible for the investigation. PCM and MCMD wrote the original draft. JP, BG, DAE, LNR, and TP reviewed and edited the manuscript. PCM and MCMD were responsible for visualization. JP, BG, LNR, DAE, TP, and FJMM were responsible for supervision. JP, BG, LNR, TP, and FJMM were responsible for project administration. PCM was responsible for funding acquisition. All authors have read and approved the final version of the manuscript.
None declared.