No Detail Found!

Implementing multi-threaded autonomous anomaly detection (MAAD) in health tracking devices

Author(s)
MUHAMMAD YUNUS BIN IQBAL BASHEER Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
AZLIZA MOHD ALI Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
NURZEATUL HAMIMAH ABDUL HAMID Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
MUHAMMAD AZIZI MOHD ARIFFIN Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
ROZIANAWATY OSMAN Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
SHARIFALILLAH NORDIN Faculty of Computer & Mathematical Sciences, Universiti Teknologi MARA, Shah Alam, Selangor, Malaysia
Abstract
Nowadays, people commonly wear health tracking devices or smartwatches as regular trackers for their health. There are many brands available in the market that are offered different functionalities. The device produces real-time health data, which should be monitored autonomously. The data from these devices is the number of calories burned, steps taken, heart rate, sleeping pattern and exercises. Based on daily activities, these devices produce a vast amount of data that could be used for health monitoring purposes. Analyzing data patterns can also provide insights into detecting health anomaly data. This paper presents a multithreaded autonomous anomaly detection algorithm (MAAD) for use in health tracking devices. The data analysis will be done in incoming streaming data, which continuously changes dynamically in the pipeline. Hence, we propose an autonomous approach using MAAD to detect anomalies from incoming smartwatch data. Firstly, we conduct data pre-processing using data gathered from several wearable devices. Once the data were ready, we sent the data via the chosen internet of things (IoT) pipeline. The data is then received by MAAD, which can handle two different processes or threads simultaneously. The first thread runs incoming data, and the other performs anomaly detection. The result shows that MAAD performs better in detecting health tracking data anomalies. This algorithm is also faster than AAD and streaming TEDA when presented with streaming data. In the future, the algorithm can be applied to any brand of smartwatch or IoT device that can supply data continuously.
Publication Details
Page(s) 4595-4603
DOI DOI not available
Published Journal: Journal of Theoretical and Applied Information Technology, Volume: 100, Issue: 15, Year: 2022

Request the full text from PASTIC

Keywords
Anomaly, IoT, Autonomous, MAAD, Health Tracking Device
More from Journal of Theoretical and Applied Information Technology
0

Citations

0

Downloads

38

Views

Copyright ©  PASTIC National Center, Islamabad www.pastic.gov.pk