In this study, a model for identifying the levels of physical activity (PA) with a wrist-worn accelerometer-based device has been proposed. The levels of identified PA have been categorized into rest/sleep, sedentary, light, moderate, and vigorous activity states by analyzing the data collected from 10 normal subjects. An activity-based sleep duration detection algorithm has been proposed and implemented thereafter to further distinguish activities between short period of rest and sleep. The model and method proposed in this study could be further used to monitor subject's daily PA status and sleep quality assessment in the future for various home-based healthcare applications.