将机器学习和深度学习模型进行比较,以预测帕金森病的认知进展
Edgar A Bernal1, Shu Yang2,3, Konnor Herbst2
1FLX AI, Rochester, New York, USA.
Clinical and translational science
|November 8, 2024
概括
深度学习模型,特别是时间融合变压器 (TFT),在预测帕金森病 (PD) 的认知衰退方面表现出卓越的表现. 这些先进的方法优于传统的模型,用于识别随时间推移的认知进展.
科学领域:
- 神经科学是一个神经科学.
- 人工智能在医学中的应用
- 计算生物学 计算生物学
背景情况:
- 帕金森病 (PD) 的认知衰退是高度可变的,需要准确的预测模型.
- 现有的PD认知进展的概率模型存在局限性,深度学习方法仍未得到充分研究.
- 对认知状态变化的早期和准确的预测对于管理PD进展至关重要.
研究的目的:
- 将传统的序列模型与深度学习技术的有效性进行比较,以预测患有PD和没有PD的人的认知进展.
- 评估浅马尔科夫,深度反复 (LSTM) 和非反复 (TFT) 模型在预测认知状态转换方面的表现.
主要方法:
- 利用了来自帕金森氏症进展标记计划 (PPMI) 数据库的数据,包括临床,人口和认知评估数据.
- 对比浅层马尔科夫,长短期记忆 (LSTM) 和时间融合变压器 (TFT) 模型,每年预测多达三年的认知状态 (正常认知,轻度认知障碍,痴呆症).
- 采用组合方法,结合模型输出和使用逆概率加权 (IPW-) F1分数评估性能.
主要成果:
- 与马尔科夫 (0.349) 和LSTM (0.414) 模型相比,时间融合变压器 (TFT) 模型表现出优异的预测性能 (IPW-F1 = 0.468).
- 集成马尔科夫,LSTM和TFT模型的整体方法进一步提高了预测准确性 (IPW-F1 = 0.502).
- TFT模型在预测较罕见的认知状态方面表现特别强,例如轻度认知障碍 (IPW-F1 = 0.496) 和痴呆症 (IPW-F1 = 0.533).
结论:
- 序列深度学习模型,特别是TFT,在预测帕金森病中临床显著的认知转变方面表现出色.
- TFT处理长期依赖和复杂数据的能力使其在退行性疾病预测方面非常有效.
- 对深度学习序列模型的进一步研究有必要用于预测神经退行性疾病中的认知变化.
更多相关视频
07:26Characterizing the Relationship Between Eye Movement Parameters and Cognitive Functions in Non-demented Parkinson's Disease Patients with Eye Tracking
Published on: September 26, 2019
7.8K
09:47Author Spotlight: Advancing Alzheimer's Research – Exploring Early Detection and Multi-Omics Approaches
Published on: December 15, 2023
960
相关概念视频
Parkinson's Disease: Overview
482
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...
482
Neural Regulation
39.1K
Digestion begins with a cephalic phase that prepares the digestive system to receive food. When our brain processes visual or olfactory information about food, it triggers impulses in the cranial nerves innervating the salivary glands and stomach to prepare for food.
39.1K
Parkinson's Disease: Treatment
222
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...
222
