An automatic acoustic bathroom monitoring system

A key element in caring for and treating elderly people who are exhibiting symptoms of dementia is understanding the patterns within their activities of daily living (ADL). One class of ADL of importance to both caregivers and doctors is personal hygiene activities within the bathroom, which typically cannot be observed directly for reasons of privacy. An automated bathroom activity monitoring system based on acoustics is described. The system is designed to recognize and classify different activities occurring within a bathroom based on sound. At the end of the day, it summarizes all detected activities and automatically generates a customized personal hygiene behavioral report for the benefit of caregivers and geriatric clinicians. This novel system consists of an infrared sensor which detects entry into the bathroom, a microphone sensor with its associated pre-amplifier circuitry and a laptop running our special software which detects and recognizes sound events in real-time using hidden Markov models (HMMs). Experiments were performed on an actual bathroom and preliminary results are encouraging with most sound recognition accuracy being above 87%.