多源稀疏广泛转移学习用于通过语音诊断帕金森病
Yuchuan Liu1, Lianzhi Li2, Yu Rao3
1School of Intelligent Technology and Engineering, Chongqing University of Science and Technology, Chongqing, 401331, China. liuyc@cqust.edu.cn.
Medical & biological engineering & computing
|February 4, 2025
概括
一种新的多源稀疏广泛转移学习 (SBTL) 方法改善了帕金森病 (PD) 语音识别. 这种方法提高了诊断准确性和稳定性,即使语音数据有限,也有助于临床决策.
科学领域:
- 计算语言学 计算语言学
- 机器学习 机器学习
- 生物医学信号处理
背景情况:
- 使用语音分析诊断帕金森病 (PD) 提供了一种非侵入性数据收集方法.
- 在PD语音数据集中的有限样本大小阻碍了准确的识别模型的开发.
- 当在小型,专用数据集上训练模型时,过度装配仍然是一个挑战.
研究的目的:
- 引入一种新的多源稀疏宽转移学习 (SBTL) 方法,以增强PD语音识别.
- 为了解决小样本大小和过度匹配在PD语音数据分析中的局限性.
- 通过语言提高PD诊断的准确性和稳定性.
主要方法:
- 开发了一种多源稀疏广泛转移学习 (SBTL) 方法,灵感来自增量广泛学习.
- 使用稀疏网络预处理PD语音数据以识别内在的不变特征.
- 采用增量学习机制来评估转移的有效性,并为积极的知识转移调整模型结构.
主要成果:
- 在PD语音诊断中,SBTL在现有的转移学习方法上表现出显著的优势.
- 在精度方面取得了至少2.58%的改进,精度为5.71%,灵敏度为12%,F1得分为14.81%.
- 对众所周知的转移学习方法表现出相似的敏感性,同时在其他指标上保持卓越的表现.
结论:
- SBTL是一种有效,高效和稳定的PD语音识别多源传输学习方法.
- 提出的方法成功地平衡了学习能力和过度适应有限的PD语音数据.
- SBTL为临床医生在做PD诊断决策时提供了更准确的帮助.
相关概念视频
Parkinson's Disease: Overview
447
Neurodegenerative disorders are progressive diseases that cause irreversible damage and loss to neurons in specific brain areas. Examples of these disorders include Parkinson's disease, Alzheimer's disease, Multiple Sclerosis (MS), and Amyotrophic Lateral Sclerosis (ALS). These disorders share characteristics such as proteinopathies, selective neuronal vulnerability, and a complex interplay between genetic and environmental factors. The primary therapeutic goal for these conditions is...
447
Parkinson's Disease: Treatment
192
Neurodegenerative disorders, such as Parkinson's Disease (PD), involve the gradual and irreversible destruction of neurons in particular brain areas. These disorders exhibit standard features like proteinopathies, selective vulnerability of some neurons, and an interaction of intrinsic properties, genetics, and environmental influences in neural injury.
Parkinson's Disease is primarily a result of the loss of dopaminergic neurons in the substantia nigra pars compacta. The cornerstone of...
Parkinson's Disease is primarily a result of the loss of dopaminergic neurons in the substantia nigra pars compacta. The cornerstone of...
192


