Obtaining Lost Behavior Detection based on Inertial Sensor and Map Information

This paper presents the use of smartphone-embedded sensors to detect early Alzheimer’s disease (AD), where the most common symptoms are topographical disorientation and getting lost behavior. The objective of our research is to establish a correlation between diagnosis and navigation ability that might provide a doctor reliable evidence with which to diagnose incipient symptoms of patients diagnosed with AD who have cognitive impairment and abnormal motion behavior. A total of 59 subjects are available to participate in this research. 27 of them have been diagnosed with AD, and 32 are labeled as cognitively healthy subjects. We propose a new experiment to observe and detect getting lost behavior more accurately. The experiment is also used to extract sensor measurements and map information for analysis by using accelerometers to detect the temporal structure of motion behavior and track a subject’s trajectory. This information is used to observe how they make a decision on a route as recorded by the locations of smartphone signals received from a global navigation satellite system (GNSS) receiver. Finally, we implement different learning algorithms to distinguish between the AD and non-AD groups through validating the relationships and correlations in the classification results and a medical questionnaire.