Smartphone Sensor Data Mining for Gait Abnormality Detection
Abstract
Today, smartphones are a ubiquitous part of daily life. We carry them everywhere with us, and they are involved in almost every aspect of our lives. These omnipresent devices are equipped with sensors that allow them to gather information about the world around them. Among these sensors are accelerometers and gyroscopes, which measure acceleration and rotation, such as that generated by walking. A smartphone, from its usual position in your pocket, is perfectly placed to capture this information. The WISDM Lab has shown that this information can determine the qualities, identity, or actions of an individual, but such technology might be used to diagnose injuries and neurological disorders as well. In collaboration with the Albert Einstein College of Medicine, we built a model from smartphone sensor data that can detect gait abnormalities, which are often symptomatic of neurological illnesses such as non-Alzheimer's dementia. As part of this project, medical students from Albert Einstein collected data using smartphones as part of their normal clinical gait assessment. The smartphones run a custom-built application that collects accelerator and gyroscope data and transmits it back to the WISDM server. The data was cleaned, converted to representative features, and analyzed using data mining algorithms to build a model. The performance of this model indicates the viability of smartphone sensor data as a tool for detecting gait abnormalities. Further research upon and deployment of the techniques developed in this thesis could result in an application that could be used to detect gait abnormalities or neurological illnesses themselves. Such technology would prove to be a valuable tool for gait monitoring in medical and commercial settings.
Subject Area
Information science|Computer science
Recommended Citation
Gallagher, Shaun, "Smartphone Sensor Data Mining for Gait Abnormality Detection" (2014). ETD Collection for Fordham University. AAI1568366.
https://research.library.fordham.edu/dissertations/AAI1568366
