科研动态期刊摘要
基于深度学习的步态分析预测多系统健康表型
Nature Medicine · 2026-11-10
摘要
《自然·医学》,在线发表:2026年11月9日;doi:10.1038/s41591-026-04725-8 我们利用3,328名深度表型化成年人的三维骨骼运动数据开发了一个步态基础模型,这些数据是在五项运动任务中通过深度相机记录的。嵌入显著预测了3,210个表型目标中的1,887个,并在男性所有18个身体系统和女性18个中的17个中提供了预测增益,同时还改善了临床诊断和药物使用的预测。
English
Deep learning-based gait analysis predicts multisystem health phenotypes
Nature Medicine, Published online: 09 November 2026; doi:10.1038/s41591-026-04725-8 We developed a gait foundation model from three-dimensional skeletal motion data of 3,328 deeply phenotyped adults, recorded using a depth camera during five motor tasks. Embeddings significantly predicted 1,887 of 3,210 phenotypic targets and provided predictive gains across all 18 body systems in men and 17 out of 18 in women, while also improving prediction of clinical diagnoses and medication use.