As the senior population in Canada ages, the need for costly healthcare monitoring increases and the expected workloads of many of the hospitals and nursing care facilities across the country begin to exceed the personnel available. There is therefore an increasing interest in smart home technology which can provide automated monitoring of the health and well-being of seniors. This thesis presents the design of a smart home system which uses unobtrusive pressure sensing and incorporates context-awareness in the monitoring of an occupant's functional mobility. To validate the design, clinical trials were conducted with both healthy adults and mobility-impaired seniors, thereby simulating the decline in an occupant's mobility over an extended period of time due to underlying conditions.
The proposed system design includes several algorithms to extract features from pressure sequences recorded during sit-to-stand (SiSt) transfers. The first was designed to detect and characterize regions of interest from pressure images recorded from sensors placed under a bed mattress. The algorithm allowed the observation of mobility-impaired cases as well as the automated classification of symmetry with an accuracy of 93%. An algorithm was also designed to analyze the sagittal center of pressure trajectory during SiSt transfers from a bed. This allowed the extraction of significant clinical features and enabled the automated detection of mobility impairment with an accuracy of 92%. A third algorithm measured the duration of the SiSt transfer using both bed and floor pressure sequences. The durations displayed significant differences between the healthy (2.31 s for young adults, 2.88 s for seniors) and mobility-impaired (3.57 s for post-stroke, 5.35 s for post-hip-fracture) participant groups. These results compared closely to those found in recently published literature using more obtrusive methods of data collection.
Various sources of context-awareness were introduced into the monitoring system design including context from pressure sensors embedded in the grab bars of a lavatory commode. A simulated context-aware environment was designed to demonstrate the function of the monitoring system using sequences collected from participants during the various clinical trials. The environment illustrated the effects of incorporating context-awareness through simulated scenarios of occupant behaviour and samples of corrected classification.