基于机器学习的神经电生理在慢性疾病诊断中的应用

Application of neuroelectrophysiology incorporated with machine learning in the diagnosis of non-communicable diseases

摘要:

慢性疾病正逐步取代传染性疾病，成为威胁人类身体健康最为主要的因素，给家庭和社会带来沉重负担。因此，实现慢性疾病早期诊断意义重大。目前的临床诊断方法难以实现这一目标，因此需要一种新的慢性疾病诊断方法。心脑血管疾病、恶性肿瘤、糖尿病等常见的慢性疾病都会影响神经系统，造成神经电信号变化，因此神经电生理将会成为早期诊断慢性疾病的有效手段。来源于生物体不同部位的神经电信号组成复杂，将机器学习算法与神经电生理方法相结合，从而解决神经电信号包含信息复杂、难以分析这一痛点。本文对基于机器学习的神经电生理在慢性疾病诊断中的应用进行了系统全面的阐述。

Abstract:

Nowadays, the number of non-communicable diseases (NCDs) sufferers has increased dramatically surpassing that of infectious diseases, and NCDs have become the most important factor threatening the human health. This leads to heavy burdens on families and the society. Therefore, it is meaningful to early diagnose NCDs. However, it cannot be accomplished clinically at present, so a novel NCDs diagnosis method is necessary. Most NCDs like cardiovascular diseases, cancers and diabetes, can affect the nervous system and cause changes in nerve electrical signals. The state-of-the-art method for early diagnosing NCDs could be accomplished through neuroelectrophysiological methods. Considering the diversity and complication of neural signals, the widely-applied machine learning methods can be combined together with neuroelectrophysiological approaches to quantify the relationships between NCDs and neural signals. This paper generally and systematically review the neuroelectrophysiology incorporated with machine learning in the diagnosis of non-communicable diseases.