Using machine learning methods to predict physical activity types with Apple Watch and Fitbit data using indirect calorimetry as the criterion.

Abstract

Background 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 intensity level of other activities in a lab-based protocol. Methods We recruited a convenience sample of 49 participants (23 men and 26 women) to wear three devices, an Apple Watch Series 2, a Fitbit Charge HR2, and and iPhone 6S. Participants completed a 65-minute protocol consisting of 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, and running 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 dataset included 3656 and 2608 minutes of Apple Watch and Fitbit data, respectively. We tested 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 types with a reasonable accuracy. The results support the use of minute-by-minute data from Apple Watch and Fitbit combined with machine learning approaches for scalable physical activity type classification at the population level.

Measuring Energy Expenditure and Heart Rate during Maximum Aerobic Testing with the Apple Watch Series 7

“…Like the MHR outcomes, when comparing MEE on both the AW7 and the PARVO, it also shows that there is no difference between the variables, which is in agreement with previously performed studies 21,22,25 . Contrary to the results MEE found, most of the studies (performed with past models) do not suggest the MEE of AW to be valid as the measurement errors are too large [6][7][8][9]13,19,23,34 .…”

“…However, the error rate was still too high to consider the AW a valid instrument for measuring one's MEE during exercise [6][7][8][9]13,17,19,23 . On the contrary, studies indicated that the AW series 1 and 2 are valid measurements of MEE during running on the treadmill when compared to the indirect calorimetry 25,26 . It is well supported by previous findings that the MEE estimations of AW have poor accuracy, however, it is important to note that there has only been research done with Apple Watches up to series 6.…”

“…The metabolic rate is another prediction based on a person's gender, weight, and height, which can be modified on an iPhone app that connects to the AW 5 . However, the outcomes of previous studies show that the reading accuracies of MEE on AW varies between the series as well as the intensity level of exercise performed when comparing to an indirect calorimetry [6][7][8][9]13,14,[17][18][19][21][22][23][24][25] . Most studies included multiple wrist-worn wearables to determine which device consistently produced the lowest error rate for measuring MEE during various activities [7][8][9]14,[17][18][19][22][23][24] .…”

“…However, the outcomes of previous studies show that the reading accuracies of MEE on AW varies between the series as well as the intensity level of exercise performed when comparing to an indirect calorimetry [6][7][8][9]13,14,[17][18][19][21][22][23][24][25] . Most studies included multiple wrist-worn wearables to determine which device consistently produced the lowest error rate for measuring MEE during various activities [7][8][9]14,[17][18][19][22][23][24] . The results showed that the AW provided the most accurate results across multiple studies compared to all other wearables tested [7][8][9]18,19,23,24 .…”

An Exploratory Sequential Sentiment Analysis of Online Learning During the Movement Control Order in Malaysia

“…The study selected 42 students from the Faculty of Computer and Mathematical Sciences, Universiti Teknologi MARA Kelantan as the target group. Convenience sampling (Fuller et al, 2020;Jin et al, 2019) of the target population, namely students who were involved with group projects in a PBL environment was used in this research. The data collected, which comprised 150 comments from the 42 students who were using WhatsApp messages was used to perform the sentiment analysis.…”