Ambulatory sleep scoring using accelerometers-distinguishing between nonwear and sleep/wake states

An mHealth Intervention Promoting Physical Activity and Healthy Eating in a Family Setting (SMARTFAMILY): Randomized Controlled Trial

BACKGROUND

Numerous smartphone apps are targeting physical activity (PA) and healthy eating (HE), but empirical evidence on their effectiveness for the initialization and maintenance of behavior change, especially in children and adolescents, is still limited. Social settings influence individual behavior; therefore, core settings such as the family need to be considered when designing mobile health (mHealth) apps.

OBJECTIVE

The purpose of this study was to evaluate the effectiveness of a theory- and evidence-based mHealth intervention (called SMARTFAMILY [SF]) targeting PA and HE in a collective family-based setting.

METHODS

A smartphone app based on behavior change theories and techniques was developed, implemented, and evaluated with a cluster randomized controlled trial in a collective family setting. Baseline (t0) and postintervention (t1) measurements included PA (self-reported and accelerometry) and HE measurements (self-reported fruit and vegetable intake) as primary outcomes. Secondary outcomes (self-reported) were intrinsic motivation, behavior-specific self-efficacy, and the family health climate. Between t0 and t1, families of the intervention group (IG) used the SF app individually and collaboratively for 3 consecutive weeks, whereas families in the control group (CG) received no treatment. Four weeks following t1, a follow-up assessment (t2) was completed by participants, consisting of all questionnaire items to assess the stability of the intervention effects. Multilevel analyses were implemented in R (R Foundation for Statistical Computing) to acknowledge the hierarchical structure of persons (level 1) clustered in families (level 2).

RESULTS

Overall, 48 families (CG: n=22, 46%, with 68 participants and IG: n=26, 54%, with 88 participants) were recruited for the study. Two families (CG: n=1, 2%, with 4 participants and IG: n=1, 2%, with 4 participants) chose to drop out of the study owing to personal reasons before t0. Overall, no evidence for meaningful and statistically significant increases in PA and HE levels of the intervention were observed in our physically active study participants (all P>.30).

CONCLUSIONS

Despite incorporating behavior change techniques rooted in family life and psychological theories, the SF intervention did not yield significant increases in PA and HE levels among the participants. The results of the study were mainly limited by the physically active participants and the large age range of children and adolescents. Enhancing intervention effectiveness may involve incorporating health literacy, just-in-time adaptive interventions, and more advanced features in future app development. Further research is needed to better understand intervention engagement and tailor mHealth interventions to individuals for enhanced effectiveness in primary prevention efforts.

TRIAL REGISTRATION

German Clinical Trials Register DRKS00010415; https://drks.de/search/en/trial/DRKS00010415.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR2-10.2196/20534.

Determination of cut-off points for the Move4 accelerometer in children aged 8-13 years

BACKGROUND

To assess physical activity (PA) there is a need of objective, valid and reliable measurement methods like accelerometers. Before these devices can be used for research, they need to be calibrated and validated for specific age groups as the locomotion differs between children and adults, for instance. Therefore, the aim of the present study was the calibration and validation of the Move4 accelerometer for children aged 8-13 years.

METHODS

53 normal weighted children (52% boys, 48%girls) aged 8-13 years (mean age = 10.69 ± 1.46, mean BMI = 17.93 kg/m- 2, 60th percentile), wore the Move4 sensor at four different body positions (thigh, hip, wrist and the Move4ecg including heart rate measurement at the chest). They completed nine activities that considered the four activity levels (sedentary behavior (SB), light PA (LPA), moderate PA (MPA) and vigorous PA (VPA)) within a test-retest design. Intensity values were determined using the mean amplitude deviation (MAD) as well as the movement acceleration intensity (MAI) metrics. Determination of activities and energy expenditure was validated using heart rate. After that, cut-off points were determined in Matlab by using the Classification and Regression Trees (CART) method. The agreement for the cut-off points between T1 and T2 was analyzed.

RESULTS

MAD and MAI accelerometer values were lowest when children were lying on the floor and highest when running or doing jumping jacks. The mean correlation coefficient between acceleration values and heart rate was 0.595 (p = 0.01) for MAD metric and 0.611 (p = 0.01) for MAI metric, indicating strong correlations. Further, the MAD cut-off points for SB-LPA are 52.9 mg (hip), 62.4 mg (thigh), 86.4 mg (wrist) and 45.9 mg (chest), for LPA-MPA they are 173.3 mg (hip), 260.7 mg (thigh), 194.4 mg (wrist) and 155.7 mg (chest) and for MPA-VPA the cut-off points are 543.6 mg (hip), 674.5 mg (thigh), 623.4 mg (wrist) and 545.5 mg (chest). Test-retest comparison indicated good values (mean differences = 9.8%).

CONCLUSION

This is the first study investigating cut-off points for children for four different sensor positions using raw accelerometer metrics (MAD/MAI). Sensitivity and specificity revealed good values for all positions. Nevertheless, depending on the sensor position, metric values differ according to the different involvement of the body in various activities. Thus, the sensor position should be carefully chosen depending on the research question of the study.

Generalizability and performance of methods to detect non-wear with free-living accelerometer recordings

Wearable physical activity sensors are widely used in research and practice as they provide objective measures of human behavior at a low cost. An important challenge for accurate assessment of physical activity behavior in free-living is the detection non-wear. Traditionally, heuristic algorithms that rely on specific interval lengths have been employed to detect non-wear time; however, machine learned models are emerging. We explore the potential of detecting non-wear using decision trees that combine raw acceleration and skin temperature, and we investigate the generalizability of our models, traditional heuristic algorithms, and recently developed machine learned models by external validation. The Decision tree models were trained using one week of data from thigh- and hip-worn accelerometers from 64 children. External validation was performed using data from wrist-worn accelerometers of 42 adolescents. For non-wear episodes longer than 60 min, the heuristic algorithms performed the best with F1-scores above 0.96. However, regarding episodes shorter than 60 min, the best performing method was the decision tree model including the six most important predictors with F1 scores above 0.74 for all sensor locations. We conclude that for classifying non-wear time, researchers should carefully select an appropriate method and we encourage the use of external validation when reporting on machine learned non-wear models.

A wearable device for at-home obstructive sleep apnea assessment: State-of-the-art and research challenges

In the last 3 years, almost all medical resources have been reserved for the screening and treatment of patients with coronavirus disease (COVID-19). Due to a shortage of medical staff and equipment, diagnosing sleep disorders, such as obstructive sleep apnea (OSA), has become more difficult than ever. In addition to being diagnosed using polysomnography at a hospital, people seem to pay more attention to alternative at-home OSA detection solutions. This study aims to review state-of-the-art assessment techniques for out-of-center detection of the main characteristics of OSA, such as sleep, cardiovascular function, oxygen balance and consumption, sleep position, breathing effort, respiratory function, and audio, as well as recent progress in the implementation of data acquisition and processing and machine learning techniques that support early detection of severe OSA levels.

Detecting Sleep and Nonwear in 24-h Wrist Accelerometer Data from the National Health and Nutrition Examination Survey

INTRODUCTION

Estimating physical activity, sedentary behavior, and sleep from wrist-worn accelerometer data requires reliable detection of sensor nonwear and sensor wear during both sleep and wake.

PURPOSE

This study aimed to develop an algorithm that simultaneously identifies sensor wake-wear, sleep-wear, and nonwear in 24-h wrist accelerometer data collected with or without filtering.

METHODS

Using sensor data labeled with polysomnography ( n = 21) and directly observed wake-wear data ( n = 31) from healthy adults, and nonwear data from sensors left at various locations in a home ( n = 20), we developed an algorithm to detect nonwear, sleep-wear, and wake-wear for "idle sleep mode" (ISM) filtered data collected in the 2011-2014 National Health and Nutrition Examination Survey. The algorithm was then extended to process original raw data collected from devices without ISM filtering. Both algorithms were further validated using a polysomnography-based sleep and wake-wear data set ( n = 22) and diary-based wake-wear and nonwear labels from healthy adults ( n = 23). Classification performance (F1 scores) was compared with four alternative approaches.

RESULTS

The F1 score of the ISM-based algorithm on the training data set using leave-one-subject-out cross-validation was 0.95 ± 0.13. Validation on the two independent data sets yielded F1 scores of 0.84 ± 0.60 for the data set with sleep-wear and wake-wear and 0.94 ± 0.04 for the data set with wake-wear and nonwear. The F1 score when using original, raw data was 0.96 ± 0.08 for the training data sets and 0.86 ± 0.18 and 0.97 ± 0.04 for the two independent validation data sets. The algorithm performed comparably or better than the alternative approaches on the data sets.

CONCLUSIONS

A novel machine-learning algorithm was designed to recognize wake-wear, sleep-wear, and nonwear in 24-h wrist-worn accelerometer data that are applicable for ISM-filtered data or original raw data.

Putting Temperature into the Equation: Development and Validation of Algorithms to Distinguish Non-Wearing from Inactivity and Sleep in Wearable Sensors

Long-term monitoring of real-life physical activity (PA) using wearable devices is increasingly used in clinical and epidemiological studies. The quality of the recorded data is an important issue, as unreliable data may negatively affect the outcome measures. A potential source of bias in PA assessment is the non-wearing of a device during the expected monitoring period. Identification of non-wear time is usually performed as a pre-processing step using data recorded by the accelerometer, which is the most common sensor used for PA analysis algorithms. The main issue is the correct differentiation between non-wear time, sleep time, and sedentary wake time, especially in frail older adults or patient groups. Based on the current state of the art, the objectives of this study were to (1) develop robust non-wearing detection algorithms based on data recorded with a wearable device that integrates acceleration and temperature sensors; (2) validate the algorithms using real-world data recorded according to an appropriate measurement protocol. A comparative evaluation of the implemented algorithms indicated better performances (99%, 97%, 99%, and 98% for sensitivity, specificity, accuracy, and negative predictive value, respectively) for an event-based detection algorithm, where the temperature sensor signal was appropriately processed to identify the timing of device removal/non-wear.

Women with abuse-related PTSD sleep more fitfully but just as long as healthy controls: an actigraphic study

STUDY OBJECTIVES

Subjective reports of sleep impairments are common in individuals with posttraumatic stress disorder (PTSD), but objective assessments of sleep have yielded mixed results.

METHODS

We investigated sleep via actigraphy and e-diary on 6 consecutive nights in a group of 117 women with PTSD after childhood abuse (CA; PTSD group), a group of 31 mentally healthy women with a history of CA (healthy trauma controls, HTC group) and a group of 36 non-traumatized mentally healthy women (healthy controls, HC group).

RESULTS

The PTSD group reported lower sleep quality, more nights with nightmares, and shorter sleep duration than both HTC and HC. Actigraphic measures showed more and longer sleep interruptions in the PTSD group compared to HTC and HC, but no difference in sleep duration. While the PTSD group underestimated their sleep duration, both HTC and HC overestimated their sleep duration. HTC did not differ from HC regarding sleep impairments.

CONCLUSIONS

Sleep in women with PTSD after CA seems to be more fragmented but not shorter compared to sleep patterns of mentally healthy control subjects. The results suggest a stronger effect of PTSD psychopathology on sleep compared to the effect of trauma per se.

Manual Annotation of Time in Bed Using Free-Living Recordings of Accelerometry Data

With the emergence of machine learning for the classification of sleep and other human behaviors from accelerometer data, the need for correctly annotated data is higher than ever. We present and evaluate a novel method for the manual annotation of in-bed periods in accelerometer data using the open-source software Audacity^®, and we compare the method to the EEG-based sleep monitoring device Zmachine^® Insight+ and self-reported sleep diaries. For evaluating the manual annotation method, we calculated the inter- and intra-rater agreement and agreement with Zmachine and sleep diaries using interclass correlation coefficients and Bland–Altman analysis. Our results showed excellent inter- and intra-rater agreement and excellent agreement with Zmachine and sleep diaries. The Bland–Altman limits of agreement were generally around ±30 min for the comparison between the manual annotation and the Zmachine timestamps for the in-bed period. Moreover, the mean bias was minuscule. We conclude that the manual annotation method presented is a viable option for annotating in-bed periods in accelerometer data, which will further qualify datasets without labeling or sleep records.

Drivers of productivity: Being physically active increases yet sedentary bouts and lack of sleep decrease work ability

Physical behavior (ie, physical activity, sedentary behavior, and sleep) is a crucial lifestyle factor for preventing and managing diseases across the lifespan. However, less is known about potential work-related psychological and cognitive outcomes such as productivity. The present study examined within-person associations between physical behavior and self-perceived work ability. To investigate the degree to which physical behavior parameters influence self-perceived work ability in everyday life, we conducted an Ambulatory Assessment study in 103 university students over 5 days. Physical behavior was assessed continuously via a multi-sensor system. Self-perceived work ability was assessed repeatedly up to six times per day on smartphones. We employed multilevel modeling to analyze the within-person effects of physical behavior on self-perceived work ability. Physical activity intensity (MET) (β = 0.15 ± 0.06, t = 2.59, p = 0.012) and sit-to-stand transitions (β = 0.07 ± 0.03, t = 2.44, p = 0.015) were positively associated with self-perceived work ability. Sedentary bouts (≥20 min) (β = -0.21 ± 0.08, t = -2.74, p = 0.006) and deviation from a recommended sleep duration (ie, 8 h) (β = -0.1 ± 0.04, t = -2.38, p = 0.018) were negatively associated with self-perceived work ability. Exploratory analyses supported the robustness of our findings by comparing various time frames. Total sedentary time and sleep quality were not associated with self-perceived work ability. Regular sleep durations, breaking up sedentary time through sit-to-stand transitions, and higher intensities of physical activity may be important for the regulation of self-perceived work ability in university students' daily lives.

Comparison of Self-Reported and Device-Based Measured Physical Activity Using Measures of Stability, Reliability, and Validity in Adults and Children

Quantification of physical activity (PA) depends on the type of measurement and analysis method making it difficult to compare adherence to PA guidelines. Therefore, test-retest reliability, validity, and stability for self-reported (i.e., questionnaire and diary) and device-based measured (i.e., accelerometry with 10/60 s epochs) PA was compared in 32 adults and 32 children from the SMARTFAMILY study to examine if differences in these measurement tools are systematic. PA was collected during two separate measurement weeks and the relationship for each quality criteria was analyzed using Spearman correlation. Results showed the highest PA values for questionnaires followed by 10-s and 60-s epochs measured by accelerometers. Levels of PA were lowest when measured by diary. Only accelerometry demonstrated reliable, valid, and stable results for the two measurement weeks, the questionnaire yielded mixed results and the diary showed only a few significant correlations. Overall, higher correlations for the quality criteria were found for moderate than for vigorous PA and the results differed between children and adults. Since the differences were not found to be systematic, the choice of measurement tools should be carefully considered by anyone working with PA outcomes, especially if vigorous PA is the parameter of interest.

Sleep classification from wrist-worn accelerometer data using random forests

Accurate and low-cost sleep measurement tools are needed in both clinical and epidemiological research. To this end, wearable accelerometers are widely used as they are both low in price and provide reasonably accurate estimates of movement. Techniques to classify sleep from the high-resolution accelerometer data primarily rely on heuristic algorithms. In this paper, we explore the potential of detecting sleep using Random forests. Models were trained using data from three different studies where 134 adult participants (70 with sleep disorder and 64 good healthy sleepers) wore an accelerometer on their wrist during a one-night polysomnography recording in the clinic. The Random forests were able to distinguish sleep-wake states with an F1 score of 73.93% on a previously unseen test set of 24 participants. Detecting when the accelerometer is not worn was also successful using machine learning ([Formula: see text]), and when combined with our sleep detection models on day-time data provide a sleep estimate that is correlated with self-reported habitual nap behaviour ([Formula: see text]). These Random forest models have been made open-source to aid further research. In line with literature, sleep stage classification turned out to be difficult using only accelerometer data.

Emfit Bed Sensor Activity Shows Strong Agreement with Wrist Actigraphy for the Assessment of Sleep in the Home Setting

PURPOSE

Wrist-worn actigraphy via research-grade devices, a well-established approach to the assessment of rest-activity, is limited by poor compliance, battery life, and lack of direct evidence for time spent physically in the bed. A non-invasive bed sensor (Emfit) may provide advantages over actigraphy for long-term sleep assessment in the home. This study compared sleep-wake measurements between this sensor and a validated actigraph.

PATIENTS AND METHODS

Thirty healthy subjects (6 to 54 years) underwent simultaneous monitoring with both devices for 14 days and filled out a daily sleep diary. Parameters included bed entry time, sleep start, sleep end, bed exit time, rest interval duration, and wake after sleep onset (WASO). The agreement between the two devices was measured using Bland-Altman plots and inter-class correlation coefficients (ICC). In addition, sensitivity, specificity, and accuracy were obtained from epoch-by-epoch comparisons of Emfit and actigraphy.

RESULTS

Fifteen percent of the subjects reported that wearing the actigraph was a burden. None reported that using the bed sensor was a burden. The minimal detectable change between Emfit and actigraphy was 11 minutes for bed entry time, 14 minutes for sleep start, 14 minutes for sleep end, 10 minutes for bed exit time, 20 minutes for rest interval duration, and 110 minutes for WASO. Inter-class correlation coefficients revealed an excellent agreement for all sleep parameters (ICC=0.99, 95% CI 98-99) except for WASO (ICC=0.46, 95% CI 0.33-0.56). Sensitivity, specificity, and accuracy were 0.62, 0.93, and 0.88, respectively. Kappa correlation analysis revealed a moderate correlation between the two devices (κ=0.55, p<0.0001).

CONCLUSION

Emfit is an acceptable alternative to actigraphy for the estimation of bed entry time, sleep start, sleep end, bed exit time, and rest interval duration. However, WASO estimates are poorly correlated between the two devices. Emfit may offer methodological advantages in situations where actigraphy is challenging to implement.

Accuracy of Sedentary Behavior-Triggered Ecological Momentary Assessment for Collecting Contextual Information: Development and Feasibility Study

Background

Sedentary behavior has received much attention in the scientific community over the past decade. There is growing evidence that sedentary behavior is negatively associated with physical and mental health. However, an in-depth understanding of the social and environmental context of sedentary behavior is missing. Information about sedentary behavior, such as how everyday sedentary behavior occurs throughout the day (eg, number and length of sedentary bouts), where, when, and with whom it takes place, and what people are doing while being sedentary, is useful to inform the development of interventions aimed at reducing sedentary time. However, examining everyday sedentary behavior requires specific methods.

Objective

The purpose of this paper is (1) to introduce sedentary behavior–triggered Ecological Momentary Assessment (EMA) as a methodological advancement in the field of sedentary behavior research and (2) to examine the accuracy of sedentary behavior–triggered EMA in 3 different studies in healthy adults. Moreover, we compare the accuracy of sedentary behavior–triggered EMA to simulations of random-trigger designs.

Methods

Sedentary behavior–triggered EMA comprises a continuous assessment of sedentary behavior via accelerometers and repeated contextual assessments via electronic diaries (ie, an application on a smartphone). More specifically, the accelerometer analyzes and transfers data regarding body position (a sitting or lying position, or an upright position) via Bluetooth Low Energy (BLE) to a smartphone in real time and triggers the deployment of questionnaires. Each time a participant spends a specified time (eg, 20 minutes) in a sedentary position, the e-diary triggers contextual assessments. To test the accuracy of this method, we calculated a percentage score for all triggered prompts in relation to the total number of bouts that could trigger a prompt.

Results

Based on the accelerometer recordings, 29.3% (5062/17278) of all sedentary bouts were classified as moderate-to-long (20-40 minutes) and long bouts (≥ 41 minutes). On average, the accuracy by participant was 82.77% (3339/4034; SD 21.01%, range 71.00-88.22%) on the study level. Compared to simulations of random prompts (every 120 minutes), the number of triggered prompts was up to 47.9% (n=704) higher through the sedentary behavior–triggered EMA approach. Nearly 40% (799/2001) of all prolonged sedentary bouts (≥ 20 minutes) occurred during work, and in 57% (1140/2001) of all bouts, the participants were not alone.

Conclusions

Sedentary behavior–triggered EMA is an accurate method for collecting contextual information on sedentary behavior in daily life. Given the growing interest in sedentary behavior research, this sophisticated approach offers a real advancement as it can be used to collect social and environmental contextual information or to unravel dynamic associations. Furthermore, it can be modified to develop sedentary behavior–triggered mHealth interventions.

Does Cardiorespiratory Fitness Buffer Stress Reactivity and Stress Recovery in Police Officers? A Real-Life Study

High levels of cardiorespiratory fitness have the potential to buffer against physical and mental health impairments, which can result from exposure to occupational stress. Police officers are especially at risk of high psychosocial stress; therefore, effective intervention strategies are warranted. Given this background, the purpose of the present study was to examine whether police officers with different levels of cardiorespiratory fitness differ with regard to their (a) physiological stress reactivity during acute real-life stress situations, and (b) physiological recovery related to acute and chronic work stress. In total, 201 police officers took part in this study (M = 38.6 years, SD = 10.1, 35.8% females). Officers were contacted eight times on a smartphone during their workday, and asked to report their current level of positive and negative affect, as well as feelings of stress and anger. Physiological stress responses and recovery (heart rate variability) were assessed using Movisens EcgMove3 devices. The Åstrand bicycle ergometer test was used to assess participants' cardiorespiratory fitness. Chronic work stress was assessed using the effort-reward imbalance model and the job strain model. Multilevel modeling was used to test buffering effects of cardiorespiratory fitness on physiological stress reactivity. Linear regression was applied to test stress-buffering effects of cardiorespiratory fitness on physiological recovery. Results showed lowered physiological stress reactivity to acute work stress in officers with higher levels of cardiorespiratory fitness. However, these results were not consistent, with no effects occurring for feelings of anger, positive affect, and negative affect. Chronic work stress (effort-reward imbalance) was related to lower physiological recovery. Cardiorespiratory fitness was positively related to physiological recovery. Data did not support interactions between work stress and cardiorespiratory fitness on physiological recovery. To some extent, cardiorespiratory fitness seems to have the potential to buffer stress reactivity in police officers in acute stress situations. Therefore, we encourage promoting fitness programs which aim to enhance cardiorespiratory fitness in stressful occupations such as law enforcement. Improvements in cardiorespiratory fitness might further enhance physiological recovery from chronic work stress, which is thought to improve cardiovascular health.