| The information age has brought with it the desire to collect and analyze data on everything possible, including data related to personal health. The proliferation of small, low cost devices capable of measuring a wide array of data specific to a person's health has enabled an awareness of the state of a person's health in ways not previously possible. This thesis investigates the ability of a single, waist mounted tri-axial accelerometer to classify human movement and quantitatively measure parameters that could provide insight into a person's health and quality of life.;In the field of gait analysis, research as well as clinical trials has been done to investigate the relationship of gait with a person's risk of falling and diseases affecting the locomotor system such as Parkinson's disease. Typically a gait lab with expensive kinetic and kinematic systems is required in order to quantitatively measure gait parameters accurately, although pressure sensitive footswitches, and to a lesser extent, accelerometers, have been used as portable, low cost alternatives.;An algorithm to accurately determine the gait parameters of step count, step timing, and stride timing using the acceleration signals from a tri-axial accelerometer is developed. Investigation was done on the accelerometer sampling rates that are required for accurate gait parameter detection, as well as into a method for extracting gait parameters with no knowledge of the orientation of the accelerometer device with respect to the person wearing it. Accuracy of the accelerometer based gait analysis was verified using a GAITRite electronic walkway, a recognized method used for gait analysis. As well, several gait variability measures were implemented to quantify the amount of step-to-step and stride-to-stride variability that occurs during walking. Gait variability measures have been shown to be capable of distinguishing elderly fallers and non-fallers, as well as healthy gait from gait associated with diseases affecting the locomotor system such as Parkinson's disease. The gait analysis capability of a portable, very low cost, easy to use accelerometer device could add tremendous value to the health care system as a quantitative tool for assessing performance and in the detection, rehabilitation, management, and prevention of disorders affecting gait.;A framework based on a hierarchical binary processing tree is presented for classifying movements in real time using acceleration signals. Algorithms for determining common daily activities including the postures of standing, sitting, and lying, distinguishing between periods of activity and rest, identifying the activity of walking, and detecting falls are developed and evaluated through a prototype implementation. It was found that signals from a single tri-axial accelerometer can accurately classify the above mentioned human movements. These types of movement classifications can add value within a monitoring system by identifying trends in movement behaviour and abnormal events and in turn expose degradation in health or physical ability before it manifests into a more serious event. |
