
MotionDecipher: LLM-Assisted Clinical Interpretation of Human Motion Signals for Efficient Gait Analysis Accepted
ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp 2027) / Proceedings of the ACM on Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT 2026)
TLDR: This paper introduces MotionDecipher, an LLM-based framework that translates continuous human motion signals into clinical gait descriptions through a Human Motion Tokenizer, a Meta Feature Selector, and a Hybrid Context Constructor for In-Context Learning. Evaluated on gait lab data from 122 patients and 46 controls, MotionDecipher achieves 0.91 BERTScore (13% over GPT-5) and is ranked highest by clinical experts in Correctness, Completeness, and Conciseness.
DOI: https://doi.org/10.1145/3857982 (The paper will be officially published in December 2026; this DOI link will become active after publication.)
PDF coming soon
