Prediction of Nocturia in Live Alone Elderly Using Unobtrusive In-Home Sensors

Nocturia, or the need to void (or urinate) one or more times in the middle of night time sleeping, represents a significant economic burden for individuals and healthcare systems. Although it can be diagnosed in the hospital, most people tend to regard nocturia as a usual event, resulting in underreported diagnosis and treatment. Data from self-reporting via a voiding diary may be irregular and subjective especially among the elderly due to memory problems. This study aims to detect the presence of nocturia through passive in-home monitoring to inform intervention (e.g., seeking diagnosis and treatment) to improve the physical and mental health of community-dwelling elderly living alone. With continuous and objective data from motion sensors installed in each zone of the apartment (bedroom, living room, kitchen, and bathroom) and a contact sensor on the main door from 39 elderly, we derive a sensor-based nocturia classification model, where nocturia labeling is done based on psychosocial survey data. Our evaluation of the model reveals that (i) the use of sensor-derived features (e.g., bedroom and living room occupancy and activity level as well as going out patterns) beyond nocturia events and (ii) the extraction and use of usual sleep location as a feature improves the classification performance, where perfect accuracy can be achieved with support vector machine. Further analysis on the survey findings also reveals that elderly with nocturia are more likely to have poor sleep quality, and suffer from conditions related to physical frailty. Our findings lend support to the efficacy of passive in-home monitoring as a digital biomarker for detection of nocturia and related conditions in live-alone elderly.