.png){kind=icon}
.png){kind=icon}
Return
A personalized rehabilitation design method via knowledge graph-based multi-source data integration
.png){kind=icon}
PRE
AI
DOI：10.1016/j.aei.2025.104005.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
The efficacy of rehabilitation products is critically dependent on personalization, but conventional design methods struggle to effectively integrate heterogeneous multi-source data, such as patient physiological signals and behavioral feedback. This disconnect often results in poor therapeutic outcomes and a lack of adaptation to individual patient needs. To address this challenge, this study proposes a method that leverages a knowledge graph to integrate physiological, behavioral, and product parameter data. We developed a rehabilitation system combining virtual reality (VR) and an upper limb module to capture user data, and constructed a dedicated rehabilitation product design knowledge graph (RPD-KG) to unify these sources. An enhanced RippleNet algorithm was then implemented to provide personalized recommendations for design variables based on the RPD-KG. Experimental results demonstrate the method’s effectiveness, achieving a recommendation accuracy (ACC) of 0.7949 and an area under curve (AUC) of 0.8587. This research offers two key contributions. First, we propose a novel, knowledge graph-driven framework for multi-source data integration tailored for rehabilitation design. This framework utilizes a dedicated RPD-KG to semantically unify physiological, behavioral, and product data. Second, based on this integrated data, we develop and validate a personalized recommendation mechanism that effectively translates user-specific information into tailored design configurations.
Keywords:
Knowledge graph
Product design
Data-driven design
fNIRS
Journal
IF:
9.9
Papers:
4.0K
Citations:
1.7W