phat 16(7): e1

Research Article

A smart phone based gait monitor system

Download1060 downloads
  • @ARTICLE{10.4108/eai.28-9-2015.2261519,
        author={Dong Qin and Ming-Chun Huang},
        title={A smart phone based gait monitor system},
        journal={EAI Endorsed Transactions on Pervasive Health and Technology},
        keywords={gait analysis, wearable computing, fall detection, gait model},
  • Dong Qin
    Ming-Chun Huang
    Year: 2015
    A smart phone based gait monitor system
    DOI: 10.4108/eai.28-9-2015.2261519
Dong Qin1, Ming-Chun Huang1,*
  • 1: Department of Electrical Engineering and Computer Science Case Western Reserve University
*Contact email:


Gait is a person’s manner of walking. Analysis of gait can be used in many areas like healthcare, therapy, sports training and characteristic recognition. The goal of this paper is to present a smart phone based system to collect and calculate gait parameters. These parameters which consists of steps, step length, velocity, cadence, motion intensity and walking regularity were collected by the inertial sensor in the smartphone. A prototype of gait parameter collection and visualization system has been built to collect accelerometer data from the smartphone, providing a reliable algorithm to calculate several gait parameters closely related to walking activity. The system also contains a fall detection function. Once the user suffers from fall, an alarm message will be send to another. Experiment has been done on 4 subjects for testing the stability and accuracy of the system. The experiment result has been compared with the real data. It shows a high accuracy and reliability for counting steps (error<5.47%) and walking duration (error<4.55%). Based on the gait monitor system, an anomaly data detection method is presented. Four independent gait parameters (cadence and motion intensity in three axis) are chosen from previous results during normal activity and their mean and standard deviation are calculated individually. If the latest data deviate from the normal activity model too far, this data is defined as an abnormal event.