首页> 外文会议>IEEE International Conference on Bioinformatics and Biomedicine >An Effective and Automatic Method to Aid the Diagnosis of Amyotrophic Lateral Sclerosis Using One Minute of Gait Signal
【24h】

An Effective and Automatic Method to Aid the Diagnosis of Amyotrophic Lateral Sclerosis Using One Minute of Gait Signal

机译:一种有效和自动的方法,可帮助使用一分钟信号诊断肌营养侧面硬化的诊断

获取原文

摘要

Amyotrophic Lateral Sclerosis (ALS) is a neurodegenerative disease that affects the nervous system responsible for muscle movement and eventually compromising one's ability to walk. Diagnosing ALS is a difficult task since no test can provide a definite diagnosis. In this sense, automatic methods that aid the diagnosis of ALS have an essential role in helping to reach a diagnose. However, most of the existing approaches that use gait dynamics are based on a 5-minute observation, which can be exhausting and demanding for a patient with ALS seeking the diagnosis. This paper proposes an automated method to aid the diagnosis of ALS using information obtained from one minute gait observation. The GaitNDD database, which provides gait data recorded for 5 minutes from people with ALS and from healthy subjects, was used to support and validate this study. Results are reported and evaluated for different machine learning classifiers. Features extracted from either 1-min or 5-min observations are evaluated. Our results show that 96.6% of accuracy was achieved for data derived from either the first or the 5-minute walking, with excellent sensitivity and specificity, thus showing that our method can help aid the diagnosis of ALS while reducing the time required for the walking experiment.
机译:肌萎缩的外侧硬化剂(ALS)是一种神经变性疾病,影响负责肌肉运动的神经系统,最终损害一个人的行走能力。诊断ALS是一项艰巨的任务,因为没有测试可以提供明确的诊断。从这个意义上讲,帮助诊断ALS的自动方法在帮助诊断方面具有重要作用。然而,使用步态动态的大多数现有方法基于5分钟的观察,这可能对寻求诊断的患者来说,这可能对患者进行耗尽和要求。本文提出了一种自动化方法,以帮助使用从一分钟步态观察获得的信息诊断ALS。 GAITNDD数据库提供从ALS和健康主题的人员录制5分钟的步态数据,用于支持和验证这项研究。报告并评估了不同机器学习分类器的结果。评估从1分钟或5分钟观察中提取的特征。我们的研究结果表明,从第一个或5分钟的步行源自源自敏感性和特异性,实现了96.6%的准确性,从而表明我们的方法可以帮助辅助ALS的诊断,同时减少行走所需的时间实验。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱:kefu@zhangqiaokeyan.com

京公网安备:11010802029741号 ICP备案号:京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有
  • 客服微信

  • 服务号