Methods for Gait Analysis in a Supportive Smart Home

Abstract

According to study, the elderly's mobility habits are closely tied to cognitive decline and other age-related health problems. Regular gait analysis may help with the early detection of various disorders, but the gathering of daily ambient data is difficult with current technology. The potential of ambient sensors on the market for estimating gait speed is examined in this thesis. The thesis's first section analyses data gathered from four motion sensors that were arranged in a straight line on the ceiling as used in some wide scale studies. The findings of this work indicate that the communications protocol limits the accuracy of gait speed estimation, which prompted the investigation of AI-enabled privacyrespecting cameras. Initial results showed the camera performance was limited by low and asynchronous frame rate, which led to significant error margins. A method is proposed that reduces this to 6% using techniques based on regression and interpolation.iii List of Acronyms2D 2-Dimensional 3D 3-Dimentional AD Alzheimer's' Disease AI Artificial Intelligence BE Back off Exponent BI Beacon Interval BLE Battery Life Extension CAP Contention Access Period CART Collaborative Aging Research using Technology CCA Clear Channel Assessment CDR Clinical Dementia Rating CFP Contention Free Period CNN Convolutional Neural Network CP CANDECOMP/PARAFAC decomposition CSI Channel State Information CSMA/CA Carrier Sense Multiple Access with Collision Avoidance CSV Comma-Separated Values xi EMG Electromyography GRF Ground Reaction Force GTS Guaranteed Time Slot HC Healthy Control IFS Intraframe Spacing IR Infra Red ITG Infrared Thermography MoCA The Montreal Cognitive assessment MCI Mild Cognitive Impairment PAN Personal Area Network PC Personal Computer PVC Polyvinyl Chloride RGB Red-Green-Blue RGBD Red-Green-Blue-Depth SD Superframe Duration ToF Time of Flight