A Pattern-Mining Approach for Wearable Sensor-Based Remote Health Care

A Pattern-Mining Approach for Wearable Sensor-Based Remote Health Care

C. Sweetlin Hemalatha (VIT University, India) and V. Vaidehi (VIT University, India)
DOI: 10.4018/978-1-5225-3686-4.ch006
OnDemand PDF Download:
List Price: $37.50


Rapid advancement in Wireless Sensor Network (WSN) technology facilitates remote health care solutions without hindering the mobility of a person using Wearable Wireless Body Area Network (WWBAN). Activity recognition, fall detection and finding abnormalities in vital parameters play a major role in pervasive health care for making accurate decision on health status of a person. This chapter presents the proposed two pattern mining algorithms based on associative classification and fuzzy associative classification which models the association between the attributes that characterize the activity or health condition and handles the uncertainty in data respectively for an accurate decision making. The algorithms mine the data from WWBAN to detect abnormal health status of the person and thus facilitate remote health care. The experimental results on the proposed algorithms show that they work par with the popular traditional algorithms and predicts the activity class, fall or health status in less time compared to existing traditional classifiers.
Chapter Preview


Remote Patient Monitoring (RPM) facilitates monitoring of people in their respective locations (e.g. in home) and analyses the trends in physical and vital parameters, thus enabling early detection of occurrence of abnormal events. Recent advancement in Wireless Sensor Network (WSN) has led to the development of Wearable Wireless Body Area Network (WWBAN) which comprises of small miniaturized non-invasive and minimally invasive devices that can sense, process and communicate (Tripathi et al., 2011). Wearable sensors provide mobility freedom by allowing people to do their daily life activities while being monitored. This technology decreases the cost and increases the quality of health care services by enabling continuous monitoring and timely alert generation during emergency.

Human activity monitoring enables detection of serious health risks such as sudden falls of elderly people who are home alone or people with chronic disease in near real time. This requires continuous monitoring of human movements and classifying normal low-level activities from abnormal event like fall. Wearable motion sensors such as accelerometers, gyroscopes etc. have been widely used to track movements.

Monitoring of vital signs such as heart rate, respiration rate, blood pressure and blood oxygen saturation enables detection of life threatening emergencies such as heart attacks in near real time. These parameters are good indicators of health status. For example, high blood pressure is an important indicator of heart attack. Hence, there is a need for continuous monitoring of vital parameters and analysing the data for tracking the health status.

Besides wireless sensing technologies, data stream analysis play a vital role in emergency incident detection such as fall and health abnormality. Mining sensor data streams pose great challenges in data mining as large amount of data are generated continuously with high speed in real time. Mining sensor data stream possesses different characteristics compared to traditional database model (Agrawal et. Al., 1993) such as (1) Each data element should be examined only once. (2) Though data gets generated continuously, memory usage for mining data streams is limited. (3) Each data element should be processed faster. (4) The outputs generated by online classifier algorithms should be instantly available when user requested.

Most of the existing fall detection methods are based on classifiers constructed using traditional methods such as decision trees, Bayesian Networks (Li, 2008) Neural Networks (Chen et al., 2010), Support Vector Machine (Kaiquan et al., 2011), K-Nearest Neighbour (Wang et al., 2015) etc. These classifiers may miss to cover certain hidden and interesting patterns in the data and thus suffer high false positives rates.

This chapter presents the proposed classifier for recognizing low-level activities and detecting human fall based on mining frequent patterns in tri-axial accelerometer and physiological sensor data streams. The proposed approach addresses the problem of mining sensor streams using time-sensitive sliding window based pattern mining algorithm. A classifier model is built based on Associative Classification (AC) (Liu et al., 1998) that mines frequent bit patterns and extracts rules for recognizing human activities like sitting/standing, lying and walking with an ultimate aim to detect human fall events. In order to handle the uncertainty in the data, another classifier model is proposed which is based on Fuzzy Associative Classification (FAC) (Mangalapalli & Pudi, 2011) that integrates the accelerometer data and physiological data for mining patterns to determine the health status of a person.

The chapter presents the following contributions.

Complete Chapter List

Search this Book: