Data Collection and Feature Extraction for a Smart Ward Application

Agitated behavior is common in people with dementia, and one of the challenges faced by doctors and caregivers is the detailed and continuous monitoring of such people. With the use of modern sensor and networking technology, and system directed observation, objective outcomes may be expected and with removal of the manpower requirement. This paper presents a context aware patient data collection system (CAPDCS), which is done on the fly with multiple sensors of different modalities deployed in the hospital ward. These features extracted from each sensor modality are used in data fusion algorithms. We also present an approach for multi-modalities data fusion based on Bayesian network.