Tri-Axial Accelerometer-Based Recognition of Daily Activities Causing Shortness of Breath in COPD Patients

Background: The 6 min walk test is widely used to measure the severity of COPD, but this imposes an extensive burden on patients and medical staff. As alternative test, measurement activity amount during daily activities that cause COPD feel shortness breath can be acquired using wearable devices, measurements between healthy people compared. Accordingly, based machine learning, we first evaluated accuracy accelerometers in recognizing such activities. Methods: Forty-six participants wore tri-axial wrist hip, performed nine activities: changing clothes, sitting, standing, lying a supine position, brushing teeth, moving luggage, going up/down stairs, running, walking. Features were extracted from 10 s windows 31 datasets input into machine-learning classifier model. Results: wrist+hip recognized walking, running. Sitting other movements not appropriately recognized. Conclusions: Sedentary as sitting difficult recognize. However, running recognizable. These tend patients. Thus, findings confirm feasibility algorithm could determine by comparing for each these with those same age.