Abstract：
BACKGROUND: Emerging and advanced technologies in the field of Artificial Intelligence (AI) represent promising methods to predict and diagnose neurodegenerative diseases, such as dementia. By using multimodal approaches, Machine Learning (ML) seems to provide a better understanding of the pathological mechanisms underlying the onset of dementia. The purpose of this review was to discuss the current ML application in the field of neuropsychology and electrophysiology, exploring its results in both prediction and diagnosis for different forms of dementia, such as Alzheimer\'s disease (AD), Vascular Dementia (VaD), Dementia with Lewy bodies (DLB), and Frontotemporal Dementia (FTD).
METHODS: Main ML-based papers focusing on neuropsychological assessments and electroencephalogram (EEG) studies were analyzed for each type of dementia.
RESULTS: An accuracy ranging between 70 % and 90 % or even more was observed in all neurophysiological and electrophysiological results trained by ML. Among all forms of dementia, the most significant findings were observed for AD. Relevant results were mostly related to diagnosis rather than prediction, because of the lack of longitudinal studies with appropriate follow-up duration. However, it remains unclear which ML algorithm performs better in diagnosing or predicting dementia.
CONCLUSIONS: Neuropsychological and electrophysiological measurements, together with ML analysis, may be considered as reliable instruments for early detection of dementia.
METHODS: Main ML-based papers focusing on neuropsychological assessments and electroencephalogram (EEG) studies were analyzed for each type of dementia.
RESULTS: An accuracy ranging between 70 % and 90 % or even more was observed in all neurophysiological and electrophysiological results trained by ML. Among all forms of dementia, the most significant findings were observed for AD. Relevant results were mostly related to diagnosis rather than prediction, because of the lack of longitudinal studies with appropriate follow-up duration. However, it remains unclear which ML algorithm performs better in diagnosing or predicting dementia.
CONCLUSIONS: Neuropsychological and electrophysiological measurements, together with ML analysis, may be considered as reliable instruments for early detection of dementia.
摘要：
背景：人工智能（AI）领域的新兴和先进技术代表了预测和诊断神经退行性疾病的有前途的方法，比如痴呆症。通过使用多模态方法，机器学习（ML）似乎可以更好地了解痴呆症发作的病理机制。这篇综述的目的是讨论当前ML在神经心理学和电生理学领域的应用，探索其对不同形式痴呆的预测和诊断结果，如阿尔茨海默病(AD)，血管性痴呆（VaD），路易体痴呆（DLB），和额颞叶痴呆（FTD）。
方法：针对每种类型的痴呆，分析了基于ML的主要论文，重点是神经心理学评估和脑电图（EEG）研究。
结果：在ML训练的所有神经生理学和电生理学结果中观察到70-90％甚至更多的准确性。在所有形式的痴呆症中，最显著的发现是AD.相关结果主要与诊断有关，而不是与预测有关。由于缺乏具有适当随访时间的纵向研究。然而,目前尚不清楚哪种ML算法在诊断或预测痴呆方面表现更好.
结论：神经心理学和电生理学测量，与ML分析一起，可以被认为是早期发现痴呆症的可靠工具。
方法：针对每种类型的痴呆，分析了基于ML的主要论文，重点是神经心理学评估和脑电图（EEG）研究。
结果：在ML训练的所有神经生理学和电生理学结果中观察到70-90％甚至更多的准确性。在所有形式的痴呆症中，最显著的发现是AD.相关结果主要与诊断有关，而不是与预测有关。由于缺乏具有适当随访时间的纵向研究。然而,目前尚不清楚哪种ML算法在诊断或预测痴呆方面表现更好.
结论：神经心理学和电生理学测量，与ML分析一起，可以被认为是早期发现痴呆症的可靠工具。
