let's break down each column in this fitness tracker app data:

1. UserID: This column contains unique identifiers for each user of the fitness tracker app. Each row corresponds to a specific user's data.

2. Date: This column represents the date on which the data was recorded or collected. It's likely in a date format (e.g., YYYY-MM-DD).

3. Steps: This column records the number of steps the user took on the given date. Steps are a common metric used by fitness trackers to measure physical activity.

4. Total_Distance: This column indicates the total distance covered by the user on the given date, likely measured in a unit such as kilometers or miles. It might be calculated based on steps taken and stride length.

5. Tracker_Distance: This column represents the distance recorded by the fitness tracker device itself, which could include steps as well as other factors like GPS data.

6. Logged_Activities_Distance: This column contains additional distance covered during specific activities that the user manually logged into the app. For example, if the user went for a run and entered the distance manually, it would be recorded here.

7. Very_Active_Distance: This column indicates the distance covered during activities classified as "very active," such as running, intense cardio, or high-intensity interval training.

8. Moderately_Active_Distance: This column represents the distance covered during activities classified as "moderately active," which may include brisk walking, cycling, or light jogging.

9. Light_Active_Distance: This column indicates the distance covered during activities classified as "light activity," such as casual walking, household chores, or light stretching.

10. Sedentary_Active_Distance: This column represents the distance covered while engaged in sedentary activities, such as sitting or lying down. It could be used to track inactive periods.

11. Very_Active_Minutes: This column records the number of minutes the user spent engaging in activities classified as "very active," typically high-intensity exercises that significantly elevate heart rate.

12. Fairly_Active_Minutes: This column contains the number of minutes spent engaging in activities classified as "fairly active," which are moderately intense activities that raise heart rate but are not as vigorous as "very active" activities.

13. Lightly_Active_Minutes: This column indicates the number of minutes spent engaging in activities classified as "lightly active," which include low-intensity activities that contribute to overall movement but do not significantly elevate heart rate.

14. Sedentary_Minutes: This column records the amount of time the user spent in sedentary behavior, such as sitting or lying down, without engaging in physical activity.

15. Calories_Burned: This column represents an estimate of the number of calories the user burned throughout the day based on their activity levels and other factors like age, weight, and gender. It's often calculated using algorithms that take into account activity data and user profile information.

Wearable Activity Tracker Data

SQLite database of wearable activity tracker users (N=88) data collected for 4 months.

The data has been collected between May 15th 2020 and September 15th 2020 in Switzerland, and is part of the data originally used in this study.

The data was collected with a Fibit Inspire HR.

• Age of participants: min. age: 18 y.o., max age: 31 y.o., mean age: 21.1 y.o., std age: 2.2 y.o.
• Gender of participants: women: 67%, men: 33%

For each of the 88 users, we collected:

• step count for every minute
• heart rate for every minute
• resting heart rate
• sleep data
  • fall asleep time
  • duration
  • sleep quality
  • restless times
  • restless duration
• all automatically detected activities
  • type (e.g., walking, swimming)
  • duration
  • time
• gender (as declared in the Fitbit profile)

This dataset contains Axivity AX3 wrist-worn activity tracker data that were collected from 151 participants in 2014-2016 around the Oxfordshire area. Participants were asked to wear the device in daily living for a period of roughly 24 hours, amounting to a total of almost 4,000 hours. Vicon Autograph wearable cameras and Whitehall II sleep diaries were used to obtain the ground truth activities performed during the period (e.g. sitting watching TV, walking the dog, washing dishes, sleeping), resulting in more than 2,500 hours of labelled data. Accompanying code to analyse this data is available at https://github.com/activityMonitoring/capture24. The following papers describe the data collection protocol in full: i.) Gershuny J, Harms T, Doherty A, Thomas E, Milton K, Kelly P, Foster C (2020) Testing self-report time-use diaries against objective instruments in real time. Sociological Methodology doi: 10.1177/0081175019884591; ii.) Willetts M, Hollowell S, Aslett L, Holmes C, Doherty A. (2018) Statistical machine learning of sleep and physical activity phenotypes from sensor data in 96,220 UK Biobank participants. Scientific Reports. 8(1):7961. Regarding Data Protection, the Clinical Data Set will not include any direct subject identifiers. However, it is possible that the Data Set may contain certain information that could be used in combination with other information to identify a specific individual, such as a combination of activities specific to that individual ("Personal Data"). Accordingly, in the conduct of the Analysis, users will comply with all applicable laws and regulations relating to information privacy. Further, the user agrees to preserve the confidentiality of, and not attempt to identify, individuals in the Data Set.

Data were collected from 113 participants, who shared their physical activity (PA) data using privately owned smart watches and activity trackers from Garmin and Fitbit. This data set consists of two data files: "data.csv" and "data raw.csv": The first file ("data.csv") contains daily averages for steps, total energy expenditure (TEE), activity energy expenditure (AEE), moderate-to-vigorous physical activity (MVPA), light PA (LPA), moderate PA (MPA), vigorous PA (VPA), and sedentary time, grouped by month. In addition, daily averages for the whole year of 2019 and 2020 are included. Finally, separate variables for the first and second half of March 2020 (pre- and post COVID-19 lockdown in Norway) are included. The second file ("data raw.csv") contains raw daily values for steps, TEE, AEE, MVPA, LPA, MPA, VPA, sedentary time, and non-wear time.

This dataset contains a list of 423 consumer-based wrist-worn activity trackers and smart watches, capable of collecting and estimating physical activity levels in individuals, using accelerometer and other sensors. Data were collected by automatic and manual searches through six online and offline databases, as well as manual collecting of data from company web sites. Data were collected in 2017, and contains all identified devices released between 2011 (earliest identified device) and July 2017. For each device, 12 attributes are included. See list in the ReadMe file.

****Content**** This dataset is based on participants to a survey to record activity level, monitoring sleep, heart rate, BMI and fat. The company collected the survey of thirty participant during a period of Two months from (March to May) in the year 2016. Each Participants were assigned an ID number.

https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F13292089%2Fc0ff406f475eeb2d39c75042ffbc8960%2FBellabeat_image.webp?generation=1685466898589032&alt=media" alt=""> ** Questions related to Bellabeat devices:**

What are some trends in smart device usage? How could these trends apply to Bellabeat customers? How could these trends help influence Bellabeat marketing strategy?

Reference: The case study of this FitBit Fitness Tracker Data stored on Kaggle and made available by Mobius. https://www.kaggle.com/code/julenaranguren/bellabeat-case-study Starter Kernel(s) ANASTASIIA CHEBOTINA https://www.kaggle.com/code/chebotinaa/bellabeat-case-study-with-r EILYN ELISA CAMPOS RODRIGUEZ https://www.kaggle.com/code/eilynelisacampos/bellabeat-case-study-20230512 AZIM https://www.kaggle.com/code/azimasare/bellabeat-marketing-strategy JULEN ARANGUREN https://www.kaggle.com/code/julenaranguren/bellabeat-case-study

This dataset contains wrist-worn activity tracker data collected from 151 participants for a period of roughly 24hs in natural settings, annotated using wearable cameras and sleep diaries.

This is a fitness tracker product dataset consisting of different products from various brands with their specifications, ratings and reviews for the Indian market. The data has been collected from e-commerce websites namely Flipkart and Amazon using web scraping technique.

Inspiration This dataset could be used to find answers to some interesting questions like - 1. Is there a significant demand for fitness trackers in the Indian market? 2. Information on the top 5 brands for fitness bands and smart watches 3. Is there a correlation between the prices and product specifications, ratings, etc. 4. Different types of fitness trackers, their price segments for different users

This dataset contains 451 samples with 16 attributes. There are some missing values in this dataset.

Objective: Nighttime wakening with asthma symptoms is a key to assessment and therapy decisions, with no gold standard objective measure. The study aims were to (1) determine the feasibility, (2) explore equivalence, and (3) test concordance of a consumer-based accelerometer with standard actigraphy for measurement of sleep patterns in women with asthma as an adjunct to self-report. Methods: Panel study design of women with poorly controlled asthma from a university-affiliated primary care clinic system was used. We assessed sensitivity and specificity, equivalence and concordance of sleep time, sleep efficiency, and wake counts between the consumer-based accelerometer Fitbit Charge™ and Actigraph wGT3X+. We linked data between devices for comparison both automatically by 24-hour period and manually by sleep segment. Results: Analysis included 424 938 minutes, 738 nights, and 833 unique sleep segments from 47 women. The fitness tracker demonstrated 97% sensitivity and 40% specificity to identify sleep. Between device equivalence for total sleep time (15 and 42-minute threshold) was demonstrated by sleep segment. Concordance improved for wake counts and sleep efficiency when adjusting for a linear trend. Conclusions: There were important differences in total sleep time, efficiency, and wake count measures when comparing individual sleep segments versus 24-hour measures of sleep. Fitbit overestimates sleep efficiency and underestimates wake counts in this population compared to actigraphy. Low levels of systematic bias indicate the potential for raw measurements from the devices to achieve equivalence and concordance with additional processing, algorithm modification, and modeling. Fitness trackers offer an accessible and inexpensive method to quantify sleep patterns in the home environment as an adjunct to subjective reports, and require further informatics development.

Introduction: Consumer-wearable activity trackers are small electronic devices engineered to monitor and record fitness and health-related measures. The purpose of this systematic review is to examine the validity and reliability of commercial wearables in measuring step count, heart rate, and energy expenditure. Method: We extracted information about commercial wearable devices (e.g., price, size, battery life, sensors, measurements, algorithms) using an Internet search conducted from November 2016- January 2017. From this search we identified devices to be included in the review. Database searches were conducted in PubMed, Embase, and SPORTDiscus, and only included articles published in the English language up to May 2019. Studies were excluded if they did not identify the device used and if they did not examine the validity and/or reliability of a device. Studies including the general population and all special populations were included. We operationalized validity as criterion (as compared to other measures) and construct (degree to which device is measuring what it purports) validity. Reliability measures focused on intradevice and interdevice reliability. Results: We included 158 publications examining 9 different commercial wearable device brands. Fitbit was by far the most studied brand. In lab-based settings Fitbit, Apple, and Samsung appeared to measure steps accurately. Heart rate was more variable with Apple Watch, Garmin was the most accurate and Fitbit tended towards underestimation. For energy expenditure, no brand was accurate. We also examined validity between devices within a specific brand. Conclusion: Activity trackers are still an emerging market and the devices are constantly being upgraded and redesigned to new models, suggesting the need for more current reviews and research.

This dataset contains Axivity AX3 activity tracker (accelerometer) data collected from 39 participants in 2019-2020 around the Oxfordshire area. Calibrated and resampled traxial acceleration data is included, captured during unscripted, free living in healthy adult volunteers (aged 18 and above) with no lower limb injury within the previous 6 months and who were able to walk without an assistive device. Participants wore four triaxial accelerometers concurrently (AX3, Axivity, Newcastle, UK), two placed side-by-side on the dominant wrist and two clipped to the dominant-side hip at the midsagittal plane. Accelerometers were synchronised using the Open Movement GUI software (v.1.0.0.42), with one recording at 100 Hz and the other at 25 Hz at each body location. Foot-facing video was captured using an action camera (Action Camera CT9500, Crosstour, Shenzhen, China) mounted at the participant’s beltline. From the synchronised camera data, a step is annotated in each CSV file by a single "1" at the approximate time of heel strike.

A full data description is available in README.txt upon download.

LifeSnaps Dataset Documentation

Ubiquitous self-tracking technologies have penetrated various aspects of our lives, from physical and mental health monitoring to fitness and entertainment. Yet, limited data exist on the association between in the wild large-scale physical activity patterns, sleep, stress, and overall health, and behavioral patterns and psychological measurements due to challenges in collecting and releasing such datasets, such as waning user engagement, privacy considerations, and diversity in data modalities. In this paper, we present the LifeSnaps dataset, a multi-modal, longitudinal, and geographically-distributed dataset, containing a plethora of anthropological data, collected unobtrusively for the total course of more than 4 months by n=71 participants, under the European H2020 RAIS project. LifeSnaps contains more than 35 different data types from second to daily granularity, totaling more than 71M rows of data. The participants contributed their data through numerous validated surveys, real-time ecological momentary assessments, and a Fitbit Sense smartwatch, and consented to make these data available openly to empower future research. We envision that releasing this large-scale dataset of multi-modal real-world data, will open novel research opportunities and potential applications in the fields of medical digital innovations, data privacy and valorization, mental and physical well-being, psychology and behavioral sciences, machine learning, and human-computer interaction.

The following instructions will get you started with the LifeSnaps dataset and are complementary to the original publication.

Data Import: Reading CSV

For ease of use, we provide CSV files containing Fitbit, SEMA, and survey data at daily and/or hourly granularity. You can read the files via any programming language. For example, in Python, you can read the files into a Pandas DataFrame with the pandas.read_csv() command.

Data Import: Setting up a MongoDB (Recommended)

To take full advantage of the LifeSnaps dataset, we recommend that you use the raw, complete data via importing the LifeSnaps MongoDB database.

To do so, open the terminal/command prompt and run the following command for each collection in the DB. Ensure you have MongoDB Database Tools installed from here.

For the Fitbit data, run the following:

mongorestore --host localhost:27017 -d rais_anonymized -c fitbit 

For the SEMA data, run the following:

mongorestore --host localhost:27017 -d rais_anonymized -c sema 

For surveys data, run the following:

mongorestore --host localhost:27017 -d rais_anonymized -c surveys 

If you have access control enabled, then you will need to add the --username and --password parameters to the above commands.

Data Availability

The MongoDB database contains three collections, fitbit, sema, and surveys, containing the Fitbit, SEMA3, and survey data, respectively. Similarly, the CSV files contain related information to these collections. Each document in any collection follows the format shown below:

{
  _id: 

Objectives: Activity trackers hold the promise to support people in managing their health through quantified measurements about their daily physical activities. Monitoring personal health with quantified activity tracker-generated data provides patients with an opportunity to self-manage their health. Many activity tracker user studies have been conducted within short time frames, however, which makes it difficult to discover the impact of the activity tracker’s novelty effect or the reasons for the device’s long-term use. This study explores the impact of novelty effect on activity tracker adoption and the motivation for sustained use beyond the novelty period.

Materials and Methods: This study uses a mixed-methods approach that combines both quantitative activity tracker log analysis and qualitative one-on-one interviews to develop a deeper behavioral understanding of 23 Fitbit device users who have used their trackers for at least two months (range of use = 69 - 1073 days).

Res...

Background: A self-monitoring approach utilizing fitness trackers that provide feedback regarding physical activities has been recently applied to rehabilitation patients to promote voluntary walking activities. Although this approach has been proven to increase physical activity, it is uncertain whether the intervention improves walking ability.Aim: This review investigated whether the additional self-monitoring approach using activity trackers would improve walking ability in any type of rehabilitation setting.Methods: A systematic search was performed in four databases [PubMed (MEDLINE), The Cochrane Library, SPORTDiscus, and Cumulative Index to Nursing and Allied Health Literature] to identify studies that examined the self-monitoring approach combined with rehabilitative intervention vs. the same rehabilitative intervention only in participants with any unhealthy conditions. Two review authors independently assessed the eligibility of all the retrieved English literature published from 2009 to 2019, then discussed the final inclusion. The risk of bias was assessed referring to the criteria of the Cochrane Risk of Bias tool. The key findings were synthesized using narrative synthesis. In addition, a quantitative synthesis was conducted when more than two studies investigating the same disease were identified.Results: Eleven randomized controlled trials satisfied the eligibility criteria, nine of which had a lower risk of bias. The types of diseases included stroke, chronic obstructive pulmonary disease (COPD), cancer, Parkinson's disease, hemophilia, peripheral artery disease, post-total knee arthroplasty, and geriatric rehabilitation. Eight studies reported measures of walking endurance and four reported measures of gait speed. In the quantitative synthesis of two studies investigating COPD, there was a significant between-group difference in terms of changes in the 6-min walking distance from the baseline, which was favorable to the additional self-monitoring intervention group (mean difference: 13.1 m; 95% confidence interval, 1.8–24.5; 2 studies, 124 participants; p = 0.02; I2 = 0%). Other available data revealed no consistent evidence regarding effectiveness of the intervention.Conclusions: The findings indicate that there is little evidence suggesting the effectiveness of the self-monitoring approach in improving walking ability in rehabilitation settings. However, a weak recommendation for patients with stable COPD was implicated in the quantitative synthesis. Further research would be required to explore the best indications for this self-monitoring approach.Systematic Review Registration: CRD 42020157695.

Here are a few use cases for this project:

1. Medical Training: The ESP3903 model could be used as a learning tool in medical schools or nursing programs. Students could apply this technology to study arrhythmia cases, understand different patterns and gain practical knowledge.

2. AI-Powered Health Monitoring Devices: The model can be incorporated into wearable devices like smartwatches or fitness trackers to monitor users' heart rates, identify potential abnormal heart rhythms, and alert the user or their healthcare provider to possible arrhythmias.

3. Telemedicine and Remote Diagnosis: Using this computer vision model, doctors could diagnose arrhythmias remotely by analyzing the patient's cardiogram data. This would be particularly beneficial for patients living in remote areas or those unable to travel due to health conditions.

4. Emergency Medical Support: In emergency situations, paramedics could use a device powered by ESP3903 to quickly identify potential arrhythmia in patients and take immediate actions if required. It could also help in making informed decisions about the urgency of a situation.

5. Research Purposes: Scientists studying arrhythmia patterns may use the ESP3903 model for analyzing large datasets of cardiographic images. It could help identify patterns, correlations, or anomalies in the data, contributing to better understanding or discovery of new findings in cardiology.

Supplementary files for article "Use and consequences of exercise tracking technology on exercise psychopathology and mental health outcomes in adolescents"Exercise tracking technology use is associated with exercise psychopathology in adolescents; however, research is yet to identify components of such technology that can predict maladaptive exercise at this age. This research assessed the relationship between exercise tracking technology use and exercise psychopathology in adolescents. Development of a new measure of exercise tracking behaviours/attitudes was also conducted. Adolescents (N = 327; aged 12–15, mean = 13.64 years (SD = .95); n = 168 girls) participated in this multi-phase study. Following factor analysis to develop and validate the new measure, relationships between exercise tracking behaviours/attitudes and compulsive exercise were explored. Key components of such technology (e.g. pressure to achieve exercise-related goals) were significantly associated with higher compulsive exercise in adolescents. However, using technology to simply monitor their own exercise behaviours was significantly associated with positive exercise and mental wellbeing outcomes. Prospective research should assess how exercise tracking can predict exercise psychopathology changes and mental wellbeing throughout adolescent development.© The Authors, CC BY 4.0

Here are a few use cases for this project:

1. Home Surveillance: The "d_new_a" model can be used in home security systems to identify whether a visitor is a human or a pet. The system can send different type of alerts based on this information, for example, a more urgent alert for unknown humans and a low-priority alert for stray pets.

2. Pet Monitoring: The model can be applied for smart pet monitoring systems. When the system detects dog activities, it could be programmed to perform certain actions like opening a pet door or dispensing a treat.

3. Urban Planning: City planners can use this model to analyze the pattern of human and dog traffic in public areas. This can help identify the need for facilities such as dog parks or pedestrian zones.

4. Retail Stores: In retail environments, the model can help understand customer behavior along with their pets, thereby, providing insights for pet-friendly product placement and marketing strategies.

5. Health and Fitness: The model can be incorporated into fitness trackers or applications for outdoors activities. It can differentiate between when the user is working out alone versus with their pet dog for more accurate health and fitness tracking.

The dataset is provided in the form of an excel files with 5 tabs. The first three excel tabs constitute demonstration data on the set up of consumer wearable device for exposure and health monitoring in population studies while the two last excel tabs include the full dataset with actual data collected using the consumer wearable devices in Cyprus and Greece respectively during the Spring of 2020. The data from the last two tabs were used to assess the compliance of asthmatic schoolchildren (n=108) from both countries to public health intervention levels in response to COVID-19 pandemic (lockdown and social distancing measures), using wearable sensors to continuously track personal location and physical activity. Asthmatic children were recruited from primary schools in Cyprus and Greece (Heraklion district, Crete) and were enrolled in the LIFE-MEDEA public health intervention project (Clinical.Trials.gov Identifier: NCT03503812). The LIFE-MEDEA project aimed to evaluate the efficacy of behavioral recommendations to reduce exposure to particulate matter during desert dust storm (DDS) events and thus mitigate disease-specific adverse health effects in vulnerable groups of patients. However, during the COVID-19 pandemic, the collected data were analysed using a mixed effect model adjusted for confounders to estimate the changes in 'fraction time spent at home' and 'total steps/day' during the enforcement of gradually more stringent lockdown measures. Results of this analysis were first presented in the manuscript titled “Use of wearable sensors to assess compliance of asthmatic children in response to lockdown measures for the COVID-19 epidemic” published by Scientific Reports (https://doi.org/10.1038/s41598-021-85358-4). The dataset from LIFE-MEDEA participants (asthmatic children) from Cyprus and Greece, include variables: Study ID, gender, age, study year, ambient temperature, ambient humidity, recording day, percentage of time staying at home, steps per day, callendar day, calendar week, date, lockdown status (phase 1, 2, or 3) due to COVID-19 pandemic, and if the date was during the weekend (binary variable). All data were collected following approvals from relevant authorities at both Cyprus and Greece, according to national legislation. In Cyprus, approvals have been obtained from the Cyprus National Bioethics Committee (EEBK EΠ 2017.01.141), by the Data Protection Commissioner (No. 3.28.223) and Ministry of Education (No 7.15.01.23.5). In Greece, approvals have been obtained from the Scientific Committee (25/04/2018, No: 1748) and the Governing Board of the University General Hospital of Heraklion (25/22/08/2018).

Overall, wearable sensors, often embedded in commercial smartwatches, allow for continuous and non-invasive health measurements and exposure assessment in clinical studies. Nevertheless, the real-life application of these technologies in studies involving many participants for a significant observation period may be hindered by several practical challenges. Using a small subset of the LIFE-MEDEA dataset, in the first excel tab of dataset, we provide demonstration data from a small subset of asthmatic children (n=17) that participated in the LIFE MEDEA study that were equipped with a smartwatch for the assessment of physical activity (heart rate, pedometer, accelerometer) and location (exposure to indoor or outdoor microenvironment using GPS signal). Participants were required to wear the smartwatch, equipped with a data collection application, daily, and data were transmitted via a wireless network to a centrally administered data collection platform. The main technical challenges identified ranged from restricting access to standard smartwatch features such as gaming, internet browser, camera, and audio recording applications, to technical challenges such as loss of GPS signal, especially in indoor environments, and internal smartwatch settings interfering with the data collection application. The dataset includes information on the percentage of time with collected data before and after the implementation of a protocol that relied on setting up the smartwatch device using publicly available Application Lockers and Device Automation applications to address most of these challenges. In addition, the dataset includes example single-day observations that demonstrate how the inclusion of a Wi-Fi received signal strength indicator, significantly improved indoor localization and largely minimised GPS signal misclassification (excel tab 2). Finally excel tab 3, shows the tasks Overall, the implementation of these protocols during the roll-out of the LIFE MEDEA study in the spring of 2020 led to significantly improved results in terms of data completeness and data quality. The protocol and the representative results have been submitted for publication to the Journal of Visualised experiments (submission: JoVE63275). The Variables included in the first three excel tabs were the following: Participant ID (Unique serial number for patient participating in the study), % Time Before (Percentage of time with data before protocol implementation), % Time After (Percentage of time with data after protocol implementation), Timestamp (Date and time of event occurrence), Indoor/Outdoor (Categorical- Classification of GPS signals to Indoor and Outdoor and null(missing value) based on distance from participant home), Filling algorithm (Imputation algorithm), SSID (Wireless network name connected to the smartwatch), Wi-Fi Signal Strength (Connection strength via Wi-Fi between smartwatch and home’s wireless network. (0 maximum strength), IMEI (International mobile equipment identity. Device serial number), GPS_LAT (Latitude), GPS_LONG (Longitude), Accuracy of GPS coordinates (Accuracy in meters of GPS coordinates), Timestamp of GPS coordinates (Obtained GPS coordinates Date and time), Battery Percentage (Battery life), Charger (Connected to the charger status).

Important notes on data collection methodology: Global positioning system (GPS) and physical activity data were recorded using LEMFO-LM25 smartwatch device which was equipped with the embrace™ data collection application. The smartwatch worked as a stand-alone device that was able to transmit data across 5-minute intervals to a cloud-based database via Wi-Fi data transfer. The software was able to synchronize the data collected from the different sensors, so the data are transferred to the cloud with the same timestamp. Data synchronization with the cloud-based database is performed automatically when the smartwatch contacts the Wi-Fi network inside the participants’ homes. According to the study aims, GPS coordinates were used to estimate the fraction of time spent in or out of the participants' residences. The time spent outside was defined as the duration of time with a GPS signal outside a 100-meter radius around the participant’s residence, to account for the signal accuracy in commercially available GPS receivers. Additionally, to address the limitation that signal accuracy in urban and especially indoor environments is diminished, 5-minute intervals with missing GPS signals were classified as either “indoor classification” or “outdoor classification” based on the most recent available GPS recording. The implementation of this GPS data filling algorithm allowed replacing the missing 5-minute intervals with estimated values. Via the described protocol, and through the use of a Device Automation application, information on WiFi connectivity, WiFi signal strength, battery capacity, and whether the device was charging or not was also made available. Data on these additional variables were not automatically synchronised with the cloud-based database but had to be manually downloaded from each smartwatch via Bluetooth after the end of the study period.

This study examined data and strategic skill-related problems that activity tracker users experience and the extent to which these problems vary by gender, age and educational attainment. A performance test (N = 100) was conducted to study problems experienced during actual use of activity trackers. Video data of participants’ screen actions were analyzed by coding for skill-related problems regarding data retrieval and interpretation, and goal setting and decision making. The results revealed that both data and strategic skill-related problems are experienced by all users, but are particularly prominent amongst elderly and less educated users. Problems were mostly related to retrieving the correct data. Additionally, substantial problems were experienced in every facet of strategic use. Altogether, data and strategic skills are underdeveloped for the beneficial use of activity trackers. Moreover, the differences in the problems experienced among users cause widening digital inequalities.

Objectives There is considerable promise for using commercial wearable devices for measuring physical activity at the population level. The objective of this study was to examine whether commercial wearable devices could accurately predict lying, sitting, and different physical activity intensity in a lab based protocol. Methods We recruited a convenience sample of 46 participants (26 women) to wear three devices, a GENEActiv, and Apple Watch Series 2, a Fitbit Charge HR2. Participants completed a 65-minute protocol with 40-minutes of total treadmill time and 25-minutes of sitting or lying time. Indirect calorimetry was used to measure energy expenditure. The outcome variable for the study was the activity class; lying, sitting, walking self-paced, 3 METS, 5 METS, and 7 METS. Minute-by-minute heart rate, steps, distance, and calories from Apple Watch and Fitbit were included in four different machine learning models. Results Our analysis dataset included 3656 and 2608 minutes of Apple Watch and Fitbit data, respectively. We test decision trees, support vector machines, random forest, and rotation forest models. Rotation forest models had the highest classification accuracies at 82.6% for Apple Watch and 89.3% for Fitbit. Classification accuracies for Apple Watch data ranged from 72.5% for sitting to 89.0% for 7 METS. For Fitbit, accuracies varied between 86.2 for sitting to 92.6% for 7 METS. Conclusion This study demonstrated that commercial wearable devices, Apple Watch and Fitbit, were able to predict physical activity type with a reasonable accuracy. The results support the use of minute by minute data from Apple Watch and Fitbit combined machine learning approaches for scalable physical activity type classification at the population level.