A Big Data Analytical Framework for Sports Behavior Mining and Personalized Health Services

• Published in: 2017 Conference on Technologies and Applications of Artificial Intelligence (TAAI) pp. 178 - 183
• Main Authors: Tseng, Vincent S., Chou, Chih-Hsin, Yang, Kai-Qi, Tseng, Jerry C.C.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.12.2017
• Subjects: Activity Analysis; Big Data; Big Data Analytics; Data mining; Encoding; Heart rate; Heart Rate Analysis; Medical services; Personalized Health Services; Sports Behavior Analysis; Time series analysis
• DOI: 10.1109/TAAI.2017.47

Mobile healthcare has become an important trend in medical and healthcare domains. With the rapid development of wearable and sensing technologies, various health-related information can now be recorded, forming valuable big health data. Physical activities are considered to have a great impact on heart rate, and the analysis of heart rate data now is widely used in medical/healthcare researches. The analysis of exercise records and heart rate data have been used for the research of the exercise intensity in many institutes. Heart rate patterns refers to a symbol of health status of heart, which is based on the current rate, and other physiological parameters. An effective heart rate pattern discovering is very helpful for the healthcare and cardiovascular prevention. In this work, we aim to build a big data analytics framework for sports behavior mining and personalized health services. We analyzed users' exercise data including heart rate and GPS data, which were collected in a practical sports and social platform, to discover users' periodic sports patterns and the trend of heart rate change during exercise. Since the dataset is not only very huge but also growing very quickly, we adopt Apache Spark as the development framework to address this Velocity issue in Big Data. The analytical results can serve as important core for personalized healthcare applications. Moreover, we also group the individual result to discover the clustering result, which can be further applied for advanced healthcare applications.