Long-term monitoring of COPD using wearable sensors

Abstract

Activity recognition can provide important contextual information for the diagnosis and treatment of several medical conditions. In COPD patients, measurement of long term physical activity level, combined with physiological parameters such as heart rate and respiration rate can be used for early detection of exacerbations. Using wearable sensors, we can achieve this goal by continuously monitoring the daily activities of COPD patients. Due to low computation power of wearable sensors, typical activity monitoring systems are designed to store or wirelessly transfer raw data from the sensors to a more powerful PC-class computer for classification. While this approach preserves the original data at the highest resolution, it is highly resource-intensive and therefore reduces the lifetime of the wearable sensors due to required storage space, bandwidth, and battery capacity. In this demo, we present an optimized activity monitoring system for COPD patients that performs feature extraction on wearable sensors. Such implementation minimizes the number of radio packets sent by the wearable sensors and eliminates the need to store raw sensor data.