| 摘 要: 针对膝关节置换术患者的术后康复过程中需要由康复医师完成定量评估,但这种传统方法的诊疗效率低的问题,提出一种基于麻雀搜索算法-长短期记忆神经网络(Sparrow Search Algorithm-Long Short-Term Memory, SSA-LSTM)的步态机能评估方法。该方法旨在通过高效、精确的量化评估,辅助康复医师更好地指导患者术后恢复。首先,提取了正常人和患者之间的步态时空参数,设计了回归评价指标;其次,利用麻雀搜索算法优化长短期记忆神经网络,构建了步态机能模型用于分析和评估。结果表明,相比传统回归模型,优化后模型的决定系数有所提升,平均绝对误差降低了25%,为膝关节置换术患者术后康复的步态分析提供了一种科学有效的量化评估方法。 |
| 关键词: 麻雀搜索算法;长短期记忆神经网络;步态机能评估;膝关节置换术;时空步态参数;惯性测量单元 |
|
中图分类号: TP391
文献标识码: A
|
|
| Research on Gait Function Evaluation Method Based on SSA-LSTM After Knee Arthroplasty |
|
SUI Zhihang, GU Minming
|
(School of In f ormation Science and Engineering, Zhejiang Sci-Tech University, Hangzhou 310018, China)
202130605276@mails.zstu.edu.cn; guminming@zstu.edu.cn
|
| Abstract: The quantitative evaluation of knee arthroplasty patients is completed by rehabilitation doctors during the postoperative recovery. However, this traditional methods have low diagnostic efficiency. Therefore, a gait function evaluation method based on Sparrow Search Algorithm-Long Short-Term Memory (SSA-LSTM) is proposed. This approach aims to assist rehabilitation physicians in better guiding patients' recovery through efficient and accurate quantitative assessments. Firstly, gait spatiotemporal parameters between normal individuals and patients are extracted and regression evaluation metrics are designed accordingly. Next, the Sparrow Search Algorithm is utilized to optimize the Long Short-Term Memory neural network and a gait function model is constructed for analysis and evaluation. The results indicate that compared to traditional regression models, the optimized model shows an improvement in the coefficient of determination and a 25% reduction in the mean absolute error, providing a scientifically effective and quantitative evaluation method for gait analysis in postoperative rehabilitation of knee arthroplasty patients. |
| Keywords: SSA; LSTM neural network; gait function evaluation; knee arthroplasty; Spatiotemporal gait parameters; IMU |
