Unobtrusively monitoring older adults in their homes could reduce the physical and cognitive impacts of aging. The problem of autonomous extraction of nocturnal movement times and respiratory rates using a pressure sensor array in bed was investigated.
Four segmentation methods were assessed for movement localization. A new movement detection segmentation algorithm accurately identified over 85% of movements. Six methods were evaluated for the extraction of breathing signals, including a recommended cascade that increased signal to noise ratio by 4.45 decibels. A proposed weighted voting algorithm was compared to two existing methods of data fusion. Finally, a reliability metric for validity evaluation was also presented.
Through use of the proposed methods, respiratory rates and movement times were reliably estimated from participants who slept with a pressure sensor array below their mattress. With these parameters available, decision algorithms could be developed to alert a caregiver when intervention is necessary.