This paper proposes a logical correlation-based sleep scheduling mechanism (LCSSM) to implement energy-efficient wireless sensor networks (WSNs) in ambient-assisted homes (AAHs). LCSSM analyzes sensory data generated by different human behaviors to detect the logical correlations between sensor nodes in an AAH. By utilizing the particular logical correlations of an AAH to predict its usage status, LCSSM deactivates sensor nodes accordingly to save energy when they are not expected to sense any valuable event. Evaluation results based on real life-logs have validated that LCSSM not only reduces the energy consumption of WSNs significantly, but also retains their quality of sensing successfully, e.g., with a moderate assumption on the duty cycling ratio and hardware configuration of sensor nodes, LCSSM successfully senses 98.7% valuable events with an average of 37.0% usual energy consumption, and extends the life time of WSNs by 63.4%.