The demand for assistive technology is expected to rise alongside the increasing care needs of aging populations. Ambient intelligence, particularly context awareness, offers potential benefits in addressing challenges faced by the healthcare system. In dementia care, context awareness can facilitate early detection of behavioural issues and proactive management of challenging behaviours. This paper presents a context-aware Cyber-Physical-Human System designed to monitor challenging behaviours in individuals with dementia. The system provides caregivers with interpretable situational information and enables them to personalize situational awareness. Designed for integration within nursing homes, it continuously monitors behaviour and influencing factors of behaviours in people with dementia. The system is built on three pillars: a digital twin of people and their environment, an Internet of Health Things (IoHT) infrastructure, and artificial agents. The integration of artificial agents enables near real-time calibration, ensuring situations are adequately communicated with caregivers. An initial implementation was evaluated in a Dutch nursing-home pilot, where IoT-integrated sensors recorded sound, light, climate conditions, activity, and heart rate. Caregivers contributed by conducting behavioural assessments using a custom-developed digital annotation application. Future work will focus on developing methods for real-time calibration by the proposed artificial agents to enhance the reliability of context awareness.