Deriving kinematic quantities from accelerometer readings for assessment of functional upper limb motions

Abstract Wearable accelerometers are lightweight, affordable, and allow for even smaller form factors than 9D inertial measurement units. They are therefore a promising tool for assessing the quality of movement of patients during daily life activities. While generic signal features such as signal power and frequency content are widely used, the derivation of kinematic (angular and spatial) quantities remains a challenge. We consider a chain of body segments, such as the arm, equipped with 3D accelerometers and propose a method for calculation of the inclination and relative height of the distal segment. For validation of the method against an optical motion capture system, we consider a setup with accelerometers on the forearm and the upper arm of a subject, who performs a sequence of drinking motions and pick-and-place motions. We obtain a root-mean-square deviation of about 2.5 cm for the wrist height relative to the shoulder and about 6° for the inclination angles of the forearm. We conclude that the proposed method yields measurements of kinematic quantities that are accurate enough for classification of functional versus non-functional motions or well-performed motions versus incomplete motions.