Assessing the feasibility of replacing subjective questionnaire-based sleep measurement with an objective approach using a smartwatch

Citation: Gaiduk, M.; Seepold, R.; Martínez Madrid, N.; Ortega, J.A. Assessing the Feasibility of Replacing Subjective Questionnaire-Based Sleep Measurement with an Objective Approach Using a Smartwatch. Sensors 2023,23, 6145. https:// doi.org/10.3390/s23136145 Academic Editors: Emiliano Schena, Domenico Formica and Carlo Massaroni Received: 12 May 2023 Revised: 29 June 2023 Accepted: 3 July 2023 Published: 4 July 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). sensors Article Assessing the Feasibility of Replacing Subjective Questionnaire-Based Sleep Measurement with an Objective Approach Using a Smartwatch Maksym Gaiduk 1,* , Ralf Seepold 1, Natividad Martínez Madrid 2and Juan Antonio Ortega 3 1Department of Computer Science, HTWG Konstanz—University of Applied Sciences, 78462 Konstanz, Germany 2School of Informatics, Reutlingen University, 72762 Reutlingen, Germany 3School of Computer Engineering, University of Seville, 41012 Seville, Spain *Correspondence: [email protected] Abstract: In order to ensure sufficient recovery of the human body and brain, healthy sleep is indispensable. For this purpose, appropriate therapy should be initiated at an early stage in the case of sleep disorders. For some sleep disorders (e.g., insomnia), a sleep diary is essential for diagnosis and therapy monitoring. However, subjective measurement with a sleep diary has several disadvantages, requiring regular action from the user and leading to decreased comfort and potential data loss. To automate sleep monitoring and increase user comfort, one could consider replacing a sleep diary with an automatic measurement, such as a smartwatch, which would not disturb sleep. To obtain accurate results on the evaluation of the possibility of such a replacement, a field study was conducted with a total of 166 overnight recordings, followed by an analysis of the results. In this evaluation, objective sleep measurement with a Samsung Galaxy Watch 4 was compared to a subjective approach with a sleep diary, which is a standard method in sleep medicine. The focus was on comparing four relevant sleep characteristics: falling asleep time, waking up time, total sleep time (TST), and sleep efficiency (SE). After evaluating the results, it was concluded that a smartwatch could replace subjective measurement to determine falling asleep and waking up time, considering some level of inaccuracy. In the case of SE, substitution was also proved to be possible. However, some individual recordings showed a higher discrepancy in results between the two approaches. For its part, the evaluation of the TST measurement currently does not allow us to recommend substituting the measurement method for this sleep parameter. The appropriateness of replacing sleep diary measurement with a smartwatch depends on the acceptable levels of discrepancy. We propose four levels of similarity of results, defining ranges of absolute differences between objective and subjective measurements. By considering the values in the provided table and knowing the required accuracy, it is possible to determine the suitability of substitution in each individual case. The introduction of a “similarity level” parameter increases the adaptability and reusability of study findings in individual practical cases. Keywords: objective and subjective sleep measurement; sleep diary; sleep study; smartwatch; wearables 1. Introduction It is well known that sleep is an essential part of our lives, and it has a significant influence on our well-being and also on our health, which has been confirmed in numerous scientific studies [ 1 , 2 ]. Therefore, it is of great importance to ensure healthy sleep [ 3 ]. This means that both the duration and quality should be kept at a sufficient level. This would be important in recovering mental, brain, and body states, among other things [4]. In case a person gets sleep problems, it is crucial to identify the cause and then apply the appropriate treatment promptly. According to one of the classifications, the methods for analysing sleep with the aim of identifying possible sleep disorders can generally be divided Sensors 2023,23, 6145. https://doi.org/10.3390/s23136145 https://www.mdpi.com/journal/sensors Sensors 2023,23, 6145 2 of 15 into two groups: objective and subjective measurement [ 5 ]. In subjective measurement, the values perceived by the person are measured [ 6 , 7 ]. Questionnaires or sleep diaries are often used for this purpose [ 8 ]. In turn, objective measurement is based on electronic devices that measure physiological signals from the human body [ 9 ]. These signals are subsequently analysed. A classic example of an objective approach and, at the same time, the standard for sleep analysis is polysomnography (PSG), which is very accurate yet time-consuming and costly [10]. There are advantages and disadvantages to both of these approaches. Furthermore, it is important to note that both approaches have limitations and can be influenced by subjective factors. The crucial fact is that they are typically used to detect different sleep disorders. For example, objective measurement (e.g., PSG or polygraphy) is often used to diagnose sleep apnoea syndrome or for sleep stage scoring [ 11 , 12 ]. When detecting other sleep disorders, such as insomnia, a subjective measurement is often performed and further analysed [ 13 ]. One of the most commonly used methods is a sleep diary. It can be used to monitor and adjust therapy according to changes, for example, in cognitive behavioural therapy for insomnia (CBT-I) [14]. Subjective measurement requires daily action by the user due to the need to fill in a questionnaire manually. This may reduce the number of days on which the data are collected, which has been confirmed in other studies [ 15 ]. For this reason, it could be beneficial if the required data collection could be automated through an electronic device while maintaining the accuracy of a questionnaire. Ultimately, it is a standard approach with years of experience and numerous collected data sets to diagnose some sleep disorders, such as insomnia. Therefore, it would be vital to continue to have similar accuracy and validity. Several research studies have addressed the question of replacing subjective measurement in sleep medicine with objective measurement or vice versa, including wrist-worn devices and an electronic sleep diary [ 16 , 17 ]. Some have used a device placed on the bed and providing a high level of comfort to evaluate the measurement results [ 18 ]. Others have undertaken other types of devices that allow a comfortable and exact measurement such as wearables as the device to be compared [ 19 , 20 ]. With such a variety of different devices, it is necessary to perform an evaluation of each type in order to obtain reliable results due to different measurement approaches. While in [ 15 ] the device placed under the mattress was evaluated, the focus of this work is on a category of wearable devices. Furthermore, various devices have different types of sensors on the board, are placed differently, and, moreover, development is constantly producing new devices with new features and characteristics. Therefore, generalising the results cannot give an accurate answer as to whether a particular device could be used for an adequate substitution of subjective measurement in a specific area of sleep medicine. Additionally, the possibility of substitution depends on the type of subjective measurement, which means that the validity of the measurement made with the objective method depends on the questionnaire to be substituted. Consequently, it is not a generalisation but an individual assessment of the devices and their field of application that should enable precise results. The scientific reviews carried out in previous studies have also indicated that further research is needed to provide reliable and accurate results on the possibility of substituting one type of measurement for another [21,22]. This work’s main objective is to evaluate the possibility of measuring relevant sleeprelated parameters with a particular smartwatch (Samsung Galaxy Watch 4) as a sensorbased approach instead of a sleep diary in a real-life setting. This could lead to the solution of the problem with user comfort through the automation of the measurement process and ensure high data availability by excluding the possibility of forgetting to complete the questionnaire on time. Sensors 2023,23, 6145 3 of 15 2. Materials and Methods The specified research objectives were analysed in order to define the appropriate methods for their achievement. A detailed description of the selected methods is presented in the current section. 2.1. Subjective Sleep Diary-Based Measurement When measuring sleep using a questionnaire, individuals may subjectively perceive their sleep quality differently due to factors such as stress, mood, or cognitive biases. Nevertheless, in the practice of sleep medicine, a sleep diary is one of the most commonly used methods for subjective assessment of sleep [ 23 ]. The version used to conduct this study is the German version recommended by the German Society for Sleep Research and Sleep Medicine (DGSM) because this study was conducted in Germany and this version has already been evaluated and used in several trials [ 24 ]. A total of 18 items are asked in this sleep diary, some of which consist of several questions. The questionnaires should be filled out daily in the evening and in the morning. Some examples of the questions included and relevant for the parameters to be measured within the conducted study are according to [23]: •At what time did you go to bed? •At what time did you try to go to sleep? •How long did it take you to fall asleep? •How many times did you wake up, not counting the final wake-up? •How long did these awakenings last in total? •At what time did you finally wake up? •What time did you get out of bed? •How would you rate the quality of your sleep? In addition to the information extracted directly from the questionnaires, several additional sleep-relevant parameters can be determined by assessing the answers. Some examples are total sleep time (TST), total time in bed (TIB), and sleep efficiency (SE). The results of the sleep diary evaluation play an essential role in diagnosing and monitoring the evolution of some sleep disorders, e.g., in the case of insomnia [25]. 2.2. Objective Sensor-Based Measurement Various devices can be used in the objective measurement of sleep parameters. The accuracy of sleep stage identification, as well as other sleep-related characteristics, depends on the method of signal acquisition and can also vary between different algorithms applied to the captured signals, introducing a degree of subjectivity into the interpretation of sleep data. In addition, individual variations in sleep patterns and characteristics may affect the reliability of smartwatch sleep estimations in individual cases. If the goal is to enable comfortable use for users, it is desirable that the device does not disturb the user during sleep or disturb him as little as possible. To achieve this, there are different approaches as to which devices can be used and placed [ 26 ]. One viable option is a device placed under the mattress [ 27 ]. In this case, direct contact with the human body is not necessary, leading to a higher level of comfort. At the same time, accurate measurement is a challenging task due to the absence of a direct connection of a sensor to the human body. If direct contact with the body is not an exclusion criterion, another commonly used approach can find the application: the use of wearables [ 28 ]. Such devices can be comfortably worn on the human body. The category of wearables consists of several types of devices [ 29 ], one of which is a smartwatch facilitating comfortable yet accurate measurement [30]. An analysis was carried out to enable the selection of the appropriate device. Preliminary usability testing and research on existing scientifically conducted evaluations were performed for the following devices: Fitbit Sense, Samsung Galaxy Watch 4, Amazfit GTR 3 Pro, and Huawei Watch 3 Pro. Ultimately, a decision was made to consider the Samsung Galaxy Watch 4 for several reasons. The evaluation of previous models of the Samsung Sensors 2023,23, 6145 4 of 15 Watch that have assessed the measurement of physiological parameters with acceptable results has already been performed [ 31 ]. The sleep monitoring application [ 32 ] has also already been positively evaluated. Therefore, it can be assumed that the latest available version of the smartwatch would provide at least similar or even better results due to the use of the latest sensors and software system. This device can be used to measure the sleep parameters relevant to the study, such as the times of falling asleep and waking up, sleep efficiency, and sleep profile. The measurement results are either directly available in the corresponding mobile app or can be downloaded and analysed as part of the stored private data via the menu option in the mobile app. To set up the smartwatch, the mobile app can be used, which can be performed before the start of the study. This means that no further action is required before bedtime, increasing the convenience of use and minimising the number of potential problems during the study. The watch can be worn on any wrist during the day as well as at night, and the battery life is between 30 and 40 h, which is sufficient for the objectives of the planned study. 2.3. Sleep Characteristics Subjective sleep measurement aims to record and analyse relevant parameters. They are either taken directly from a questionnaire/sleep diary or calculated from the filledin points. The number of possible parameters is considerable; therefore, establishing a selection for the evaluation was necessary. One of the prerequisites was to evaluate the parameters relevant for insomnia diagnosis and therapy observation [ 33 ]. On the basis of the analysis carried out, several relevant parameters were selected for the evaluation of the possibility of replacing the sleep diary with an objective measurement: •Falling asleep time •Waking up time •Total sleep time (TST) •Sleep efficiency (SE) These parameters can be measured directly as a part of the objective measurement with a smartwatch or calculated from the collected data. At the same time, they can be measured by answering the questions in a sleep diary, as described in the Section 2.1. Thus, the set of sleep characteristics used for the first evaluation was determined. Since according to [ 34 ], there are different variants to calculate SE, and in order to eliminate possible misunderstandings in comprehending this article, we provide the formula used for its determination. The SE for subjective measurement was calculated in the study using the following Formula (1): SE = TST/((Waking up Time −Time trying go to Sleep)) ×100% (1) For the evaluation of the possibility of replacing subjective measurement with an objective one, the differences in measurement results between these two types of measurement must be calculated and analysed for every selected sleep characteristic. 2.4. Test Subjects The list of inclusion and exclusion criteria was developed during the preparation stage based on the performed analysis of the objectives. The following inclusion criteria were established: • Age: 20–60 years old (including adults without elderly due to possible differences in sleep patterns) • No significant sleep disorders are known. (To decrease the chance of disorders affecting the results, participants were asked to provide this information and no additional screening was performed) • Declaration of consent to participate in the study after explaining the study process and objectives (to follow ethical and legal requirements). Exclusion criteria defined for the study: Sensors 2023,23, 6145 5 of 15 • Chronic or acute clinically significant diseases occurring during the study. (No diagnostic tests were carried out in the study) •Drug/alcohol abuse •Significant perceived impairment of sleep due to the wear of smartwatches. The sample size of 100 suggested by Israel [ 35 ] for a precision level of 10% with a confidence level of 95% and p= 0.5 was used as the basis for the definition of the number of recordings required. In order to increase the explanatory power, this number was increased according to the aim of collecting at least 150 recordings, having several nights pro participant to compensate possible outliers in single recordings caused by abnormal daily routine or some exceptional night occurrences. For this purpose, the recruitment of subjects for the study was started at the HTWG Konstanz in Germany, trying to contact a similar number of male and female persons as well as considering different age groups within the determined range. For this step, 15 subjects were contacted by the time this article was prepared. A total of 13 of them expressed their willingness to participate in the study. Next, the inclusion/exclusion criteria were applied, and two individuals had to be excluded due to the presence of insomnia in one of them and significant perceived sleep disturbance due to wearing the smartwatch in another (feeling of discomfort and the perception of sleeping worse than before), which only became apparent after the first few days of the experiment. After applying the inclusion/exclusion criteria, the group of subjects consisted of eleven people with an average age of 32.7 years (SD = 10.1, median = 32 , range: 25 to 59 years) was available for the evaluation. The BMI of the participants was in the range of 20.5 to 29.9, with an average of 23.8 and median of 23.4. The number of male and female participants was similar (one more male participant) to facilitate a sex-independent analysis. During the study, participants were asked to maintain their usual lifestyle, and they remained in their homes for the execution of the experiment. Only healthy adult volunteers participated in the study. Furthermore, the subjects were not patients, children, or elderly people with no diagnostic purpose and no clinical trial was conducted. In addition, it is essential to mention that the focus of this study was to assess the feasibility of replacing subjective sleep measurement with an objective approach using a publicly available consumer smartwatch and not a medical device. A study description (including information on aims, procedures, risks, etc.) and an informed consent and publication forms were first prepared to provide potential participants with this information in advance and to give them sufficient time to decide whether to participate and to answer any questions they may have. It is important to note that none of the physiological signals, such as heart or respiratory signals, were directly processed by the study organisers and will under no circumstances be made publicly available or shared with third parties. In addition, subjects were given ID numbers and all data were completely anonymised so that ID and subject could not be matched as there is no key. Age and BMI data were collected separately and were not linked to IDs or names to ensure a high level of anonymisation. 2.5. Experiment Design A research team determined the design of the experiment conducted for the evaluation and was then followed by the study participants. The experimental procedure was introduced to the subjects on the first day, and smartwatches and sleep diaries were presented and explained. All the participants’ questions were addressed, and the devices were tried in use. Furthermore, all subjects gave their informed consent to be included in the study and publication of the results in an anonymised form prior to participation. The regular daily routine of the test persons should be continued in order to be able to carry out a real-life evaluation. The aim was to examine whether substituting the sleep diary with the smartwatch is practicable in a real-life use scenario. To obtain reliable data, test subjects were asked to wear the smartwatches during the day and night, or at least to put them on 20 min before going to bed to ensure correct measurement. They were not allowed to take them off earlier than 20 min after getting up to facilitate automatic scoring Sensors 2023,23, 6145 6 of 15 by the smartwatch software. The smartwatches had to be recharged whenever the battery level was below 35% to ensure their functionality throughout the night. If there were any issues with the device, the subjects had to contact the study organisers and discontinue using it until resolved. Each evening, participants were asked to complete the evening part of a sleep diary immediately after going to bed. Directly after getting up in the morning, they were requested to fill in another part of a sleep diary with follow-up questions. The duration of the experiment consisted of 17 nights per participant; however, 3 participants had only 16 nights of recording duration due to their unavailability on the last day. On the last day of the study, a discussion was held with the participants to obtain general feedback on the performed study. The sleep diaries were collected, and general (anonymised) information on age, height, and weight was gathered separately. The devices were also checked and picked up by the participants. 2.6. Statistical Analysis Statistical analysis was performed to assess the feasibility of replacing subjective questionnaire-based sleep measurement with an objective approach using a smartwatch. Several statistical measures were used, including the Bland–Altman plot, the intraclass correlation coefficient (ICC), and the mean, median, and standard deviation (SD) of the differences between the two measurement methods. To assess the agreement between the smartwatch and sleep diary measurements, a Bland–Altman plot was generated for each sleep characteristic [ 36 ]. The Bland–Altman plot is a graphical representation that can be used to examine the systematic bias and limits of agreement between two measurement methods. In this analysis, the differences between the smartwatch and sleep diary measurements were calculated for each participant and plotted on the y-axis, while the average of the two measurements was plotted on the x-axis. The graph includes a horizontal line representing the mean difference between the two methods and two additional lines, called limits of agreement, set at ± 1.96 times the standard deviation of the differences. These limits represent a range within which approximately 95% of the differences between the two methods should fall if there is no systematic bias. The Bland–Altman plot provides valuable insight into the agreement and comparability of the smartwatch and sleep diary measurements. It helps to identify potential sources of discrepancy between the two methods and allows a more comprehensive assessment of their interchangeability. By using this statistical method, the study aims to gain a deeper understanding of the feasibility of replacing subjective questionnaire-based sleep measurement with the objective approach using a smartwatch. In addition, the ICC was calculated using a two-way mixed effects model with absolute agreement and single rater/measurement [ 37 ]. The ICC is a statistical measure that assesses the proportion of the total variance in the measurements that can be attributed to true differences between subjects/methods. It is a measure of the consistency and agreement between the two methods, with values closer to 1 indicating greater reliability. As the result of a measurement should be a quantity that can be represented in a meaningful way by a real number, i.e., numerical data [ 38 ], the values for time of day (hh:mm) had to be converted to minutes (mm), and the calculation of the ICC for the SE, which is derived from time values and expressed as a percentage, and which naturally has a relatively high measurement variance with both methods, was not considered meaningful in the context of this study. Furthermore, the mean, median, and SD of the differences between the smartwatch and sleep diary measurements were calculated for each sleep characteristic. These descriptive statistics provide an overview of the size and variability of the observed differences between the two measurement methods. In addition, box plots were used to illustrate the differences between the smartwatch and sleep diary measurements for different sleep characteristics, including falling asleep time, waking up time, total sleep time, and sleep efficiency. A boxplot provides a concise Sensors 2023,23, 6145 7 of 15 visual summary of the data distribution and allows quick comparison between different measurement methods [39]. In each box plot, a rectangular box is drawn to represent the interquartile range (IQR) of the data. The bottom and top of the box represent the 25th and 75th percentiles, respectively, while the line inside the box represents the median. This median line is a measure of the typical difference between the two methods. The whiskers extend from the box to show the range of the data, excluding outliers. Outliers are shown as individual data points beyond the whiskers. These outliers may represent extreme differences between smartwatch and sleep diary measurements for some participants. 3. Results The total number of recorded experiment nights was equal to 184. However, ten nights had to be excluded from the evaluation due to missing or obviously incorrect data in the sleep diary (for example, when the waking up time was reported to be later than the getting out of bedtime), and eight nights were excluded due to missing or incomplete data from a smartwatch (for example, when the smartwatch lost power during the night). The distribution of excluded recordings between study participants was wide; therefore, no significant effect was expected on the final results. Ultimately, 166 overnight recordings were available for evaluation and have been analysed. This number of collected data allows us to obtain statistically significant outcomes and consequently draw substantive conclusions. The variation of the measured characteristics was also relatively high, as can be seen in Table 1, indicating that different sleep times were covered, which was important for a comprehensive analysis of the data due to the fact that sleep behaviour and consequently sleep times vary significantly among persons. Table 1shows the statistical data on the measured values of each characteristic for all 166 recordings, both for the subjective sleep diary measurement and the objective measurement by means of the smartwatch. Table 1. Descriptive statistics of the measured sleep characteristics for both the sleep diary and smartwatch measurements for 166 overnight recordings. Characteristic Mean Median SD IQR Subjective measurement with sleep diary Falling asleep time 00:21 00:19 00:54 01:07 Waking up time 07:27 07:29 01:03 01:24 Total Sleep Time 06:57 06:57 00:59 01:19 Sleep Efficiency 96.21% 97.79% 4.74% 3.24% Objective measurement with smartwatch Falling asleep time 00:21 00:16 00:54 01:06 Waking up time 07:36 07:30 01:06 01:22 Total Sleep Time 06:33 06:33 00:56 01:12 Sleep Efficiency 89.72% 90.00% 3.43% 4.00% For each parameter of interest, the absolute values of the differences between subjective and objective measurements were calculated for each of the 166 recording days. In the next step, statistical parameters, as well as visualisations for better comprehensibility of the evaluation results, were prepared as described in the Section 2.6. Figure 1presents box plots with calculated differences between two measurement approaches for three sleep characteristics: falling asleep time, waking up time, and total sleep time. It can be observed that the number of outliers is significantly higher for the time of falling asleep compared to the time of waking up. The dispersion of differences between the two measurement methods for a time of falling asleep and waking up is meaningfully lower than for total sleep time. Sensors 2023,23, 6145 8 of 15 Sensors 2023, 23, x FOR PEER REVIEW 8 of 16 Figure 1 presents box plots with calculated differences between two measurement approaches for three sleep characteristics: falling asleep time, waking up time, and total sleep time. It can be observed that the number of outliers is significantly higher for the time of falling asleep compared to the time of waking up. The dispersion of differences between the two measurement methods for a time of falling asleep and waking up is meaningfully lower than for total sleep time. Figure 1. Box plots presenting the differences between the smartwatch and sleep diary measurement for “Falling asleep time”, “Waking up time”, and “Total Sleep Time” for 166 overnight recordings. For the fourth evaluated sleep characteristic, sleep efficiency, the visualisation can be seen in Figure 2. The mean and median values of the differences are very close for this characteristic, being noticeable in the graphical representation. This indicates that the dispersion of results is lower than for other measured characteristics. Figure 2. Box plot presenting the differences between the smartwatch and sleep diary measurement for the characteristic “Sleep Efficiency” for 166 overnight recordings. The summary of the obtained within the statistical evaluation values for all four sleep characteristics is outlined in Table 2, providing a brief and coherent overview of the out- Figure 1. Box plots presenting the differences between the smartwatch and sleep diary measurement for “Falling asleep time”, “Waking up time”, and “Total Sleep Time” for 166 overnight recordings. For the fourth evaluated sleep characteristic, sleep efficiency, the visualisation can be seen in Figure 2. The mean and median values of the differences are very close for this characteristic, being noticeable in the graphical representation. This indicates that the dispersion of results is lower than for other measured characteristics. Sensors 2023, 23, x FOR PEER REVIEW 8 of 16 Figure 1 presents box plots with calculated differences between two measurement approaches for three sleep characteristics: falling asleep time, waking up time, and total sleep time. It can be observed that the number of outliers is significantly higher for the time of falling asleep compared to the time of waking up. The dispersion of differences between the two measurement methods for a time of falling asleep and waking up is meaningfully lower than for total sleep time. Figure 1. Box plots presenting the differences between the smartwatch and sleep diary measurement for “Falling asleep time”, “Waking up time”, and “Total Sleep Time” for 166 overnight recordings. For the fourth evaluated sleep characteristic, sleep efficiency, the visualisation can be seen in Figure 2. The mean and median values of the differences are very close for this characteristic, being noticeable in the graphical representation. This indicates that the dispersion of results is lower than for other measured characteristics. Figure 2. Box plot presenting the differences between the smartwatch and sleep diary measurement for the characteristic “Sleep Efficiency” for 166 overnight recordings. The summary of the obtained within the statistical evaluation values for all four sleep characteristics is outlined in Table 2, providing a brief and coherent overview of the out- Figure 2. Box plot presenting the differences between the smartwatch and sleep diary measurement for the characteristic “Sleep Efficiency” for 166 overnight recordings. The summary of the obtained within the statistical evaluation values for all four sleep characteristics is outlined in Table 2, providing a brief and coherent overview of the outcomes obtained as the result of the experiment carried out. The statistical values were selected according to the results of literature research on similar scientific studies with the aim of providing a meaningful overview. Sensors 2023,23, 6145 9 of 15 Table 2. Statistical characteristics calculated for the absolute values of differences between the smartwatch and sleep diary measurements for 166 overnight recordings. Characteristic Mean Median SD IQR Sample Standard Error Falling asleep time 00:13 00:07 00:15 00:15 00:01 Waking up time 00:12 00:06 00:12 00:17 00:00 Total Sleep Time 00:35 00:30 00:23 00:32 00:01 Sleep Efficiency 7.42% 6.77% 3.74% 5.09% 0.29% In order to facilitate a more in-depth analysis of the results obtained, we have prepared the graphical representation of the measurements using Bland–Altman plots for all four sleep characteristics, as shown in Figure 3. Sensors 2023, 23, x FOR PEER REVIEW 9 of 16 comes obtained as the result of the experiment carried out. The statistical values were selected according to the results of literature research on similar scientific studies with the aim of providing a meaningful overview. Table 2. Statistical characteristics calculated for the absolute values of differences between the smartwatch and sleep diary measurements for 166 overnight recordings. Characteristic Mean Median SD IQR Sample Standard Error Falling asleep time 00:13 00:07 00:15 00:15 00:01 Waking up time 00:12 00:06 00:12 00:17 00:00 Total Sleep Time 00:35 00:30 00:23 00:32 00:01 Sleep Efficiency 7.42% 6.77% 3.74% 5.09% 0.29% In order to facilitate a more in-depth analysis of the results obtained, we have prepared the graphical representation of the measurements using Bland–Altman plots for all four sleep characteristics, as shown in Figure 3. Figure 3. Bland–Altman plots for all analysed sleep characteristics for 166 nights: (a) Falling asleep time, (b) Waking up time, (c) Total sleep time, and (d) Sleep efficiency. The calculation of the ICC for “falling asleep time”, “waking up time”, and “TST”, both for each subject and for the 166 nights of recording, was performed after converting the time-of-day values (hh:mm) to minutes (mm) and gave the following results. For “falling asleep time”, the values range from 0.76 to 0.99 with the exception of one value of 0.68, along with a mean of 0.9, median of 0.93, SD = 0.096, and IQR = 0.057. The ICC for the sum of the 166 recordings is 0.93. For waking up time, the values range from 0.76 to 0.99 with a mean of 0.93, median of 0.95, SD = 0.075, and IQR = 0.086. For all 166 night recordings, the value is 0.96. This allows us to say that the ICC values for “falling asleep time” and “waking up time” indicate the reliability of the substitution of the sleep diary by the smartwatch measurement as “good” to “excellent” for individual subjects and “excellent” for Figure 3. Bland–Altman plots for all analysed sleep characteristics for 166 nights: ( a ) Falling asleep time, (b) Waking up time, (c) Total sleep time, and (d) Sleep efficiency. The calculation of the ICC for “falling asleep time”, “waking up time”, and “TST”, both for each subject and for the 166 nights of recording, was performed after converting the time-of-day values (hh:mm) to minutes (mm) and gave the following results. For “falling asleep time”, the values range from 0.76 to 0.99 with the exception of one value of 0.68, along with a mean of 0.9, median of 0.93, SD = 0.096, and IQR = 0.057. The ICC for the sum of the 166 recordings is 0.93. For waking up time, the values range from 0.76 to 0.99 with a mean of 0.93, median of 0.95, SD = 0.075, and IQR = 0.086. For all 166 night recordings, the value is 0.96. This allows us to say that the ICC values for “falling asleep time” and “waking up time” indicate the reliability of the substitution of the sleep diary by the smartwatch measurement as “good” to “excellent” for individual subjects and “excellent” for the complete set of recordings according to [ 40 ]. For the TST, which includes several components, the ICC calculation yielded the following values: interval between 0.53 and 0.94, mean of 0.76, median of 0.75, SD = 0.115, and IQR = 0.115, while the ICC for 166 recordings for this characteristic is 0.76, indicating “good reliability”.