Diverse data mapping for sleep status estimation: From intelligent band to passive infrared motion sensors

Wearable electronic products measure activity information, it is helpful to improve quality of healthy living. For example, Microsoft band measures heart rate, step, walking distance, and sleep status. However, in some situations wearable devices make people feel uncomfortable, such as elder with dementia. This paper we propose a scheme to estimate sleep status via ambient sensor, i.e., motion sensor, instead of wearable device. The experimental environment deploys motion sensors. Furthermore, participant wears intelligent band in order to collect data of sleep status. We perform training process to sensing data, and desired output in training process is data of sleep status acquired from intelligent band. The experiment results demonstrate that our scheme achieves accuracies of 50% ~ 75% in sleep status estimation via motion sensor.