Year: 2018
Type: conference-paper
Abstract: Wearables enable measuring physical activities and heart rate using accelerometer and heart rate sensor.However, the output of these sensors is often aggregated into a general activity status without detailed analysis and the link between activity recognition and heart rate measurement is often missing in health apps.This paper compares heart rate measurements during physical activity of three wearable devices: a specialized sports device with chest strap, a fitness tracker, and a smart watch.Due to unintended shifts of the device with respect to the wrist, the fitness tracker and smart watch have difficulties to measure sudden variations in heart rate.During the physical activity, movements of the user's wrist were measured using the accelerometer of the wearable.Correlations in the data patterns of the heart rate sensor and accelerometer are identified.Both sensors are used as input for personal recommendations for physical activities with a rule based filter.These recommendations are tailored to the user's physical capabilities and preferences by matching them to a user profile that is learned from the user's data.Combining the insights from heart rate sensor and accelerometer may allow to improve the accuracy of detecting physical activities, estimate the intensity of an activity, and generate more accurate recommendations.
Authors: Toon De Pessemier, Enias Cailliau, Luc Martens
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W2926579890