Sources of error during inertial sensing of human movement: a critical review of the fundamentals

Abstract

Inertial assessments of human movement have potential to support diagnosis and treatment of neuromuscular disorders in healthcare settings. Despite the potential advantages, uptake and acceptance by healthcare professionals are still a challenge, as inertial measurement units are prone to measurement errors due to inherent limitations with the technology. As such, full exploitation is limited to a small group of highly qualified personnel. For usage to be more ubiquitous, standard practices for acquiring high-quality data are required and should include methods for error avoidance, detection, identification, quantification, and mitigation. In this paper, a critical review of sources of error was conducted, from which a taxonomic error classification framework was developed. From this review, it has become apparent which sources of error carry the highest risk for impacting data quality. Methods for error mitigation have been identified, along with limitations and areas for improvement. This framework is intended to serve as a useful reference for both proficient and non-proficient users to ensure all sources of error are considered when developing and interpreting IMU-based assessments. It also provides a foundation for developing standard practices to help users efficiently and reliably acquire high-quality data, which is imperative for uptake and acceptance in healthcare settings.