The brain activity sees a functional decline in the aging process. The Electroencephalogram (EEG) signals of young and elderly people are obtained by the decision-making experiment to be used to quantitatively analyze the changes of the brain with age. This paper presents an entropy-based characterization method of the EEG, which can accurately predict the human brain age by the machine learning method. The results show that there is a rich performance with the power spectrum entropy (PSE) in the time-resolution ability and the effect of the accurate differentiation. The distribution of the entropy of the young people in a decision-making process has a greater intensity than that of the elderly. In other words, the amount of information of the brain generated by young people is larger than that of the elderly. In addition, the support vector machine (SVM) is superior to the random forest (RF) method, since the highest average accuracy (ACC) is 88.02% and is 2.66% higher than that of the RF method. It is also found that a great difference is observed in the responses of the decision-making, especially in the left EOG, temporal and central regions of the brain, which can be more easily classified by the classifiers.
[1] Swerdlow R H. Brain aging, Alzheimer's disease, and mitochondria[J]. Biochimica et Biophysica Acta (BBA)-Molecular Basis of Disease, 2011, 1812(12):1630-1639.
[2] Luders E, Cherbuin N, Gaser C. Estimating brain age using high-resolution pattern recognition:Younger brains in longterm meditation practitioners[J]. Neuroimage, 2016, 134:508-513.
[3] Vysata O, Kukal J, Prochazka A, et al. Age-related changes in EEG coherence[J]. Neurologia i Neurochirurgia Polska, 2014, 48(1):35-38.
[4] van Albada S J, Kerr C C, Chiang A K I, et al. Neurophysiological changes with age probed by inverse modeling of EEG spectra[J]. Clinical Neurophysiology, 2010, 121(1):21-38.
[5] Pace-Schott E F, Spencer R M C. Age-related changes in the cognitive function of sleep[M]//Green A M, Chapman E C, Kalaska J F, et al. Progress in Brain Research. Amsterdam:Elsevier, 2011:75-89.
[6] Cirelli L K, Bosnyak D, Manning F C, et al. Beat-induced fluctuations in auditory cortical beta-band activity:Using EEG to measure age-related changes[J]. Frontiers in Psychology, 2014, 5:742.
[7] Dosenbach N U F, Nardos B, Cohen A L, et al. Prediction of individual brain maturity using fMRI[J]. Science, 2010, 329(5997):1358-1361.
[8] Cole J H, Leech R, Sharp D J. Prediction of brain age suggests accelerated atrophy after traumatic brain injury[J]. Annals of Neurology, 2015, 77(4):571-581.
[9] Gur R C, Calkins M E, Satterthwaite T D, et al. Neurocognitive growth charting in psychosis spectrum youths[J]. JAMA Psychiatry, 2014, 71(4):366-374.
[10] Federmeier K D, Kutas M, Schul R. Age-related and individual differences in the use of prediction during language comprehension[J]. Brain and Language, 2010, 115(3):149-161.
[11] Meier T B, Desphande A S, Vergun S, et al. Support vector machine classification and characterization of age-related reorganization of functional brain networks[J]. Neuroimage, 2012, 60(1):601-613.
[12] 綦宏志, 程龙龙, 陈彬津, 等. 想象动作中动态脑电的信息熵研究[J]. 中国生物医学工程学报, 2007, 26(1):74-77. Yan Hongzhi, Cheng Longlong, Chen Binjin, et al. Information entropy of dynamic EEG in imaginary movements[J]. Chinese Journal of Biomedical Engineering, 2007, 26(1):74-77.
[13] 周酥. 基于功率谱信息熵的异常心音识别[J]. 中国医学物理学杂志, 2014, 31(3):4933-4935. Zhou Su. Abnormal heart sound recognition based on power spectrum information entropy[J]. Chinese Journal of Medical Physics, 2014, 31(3):4933-4935.
[14] 任亚莉. 基于功率谱熵和频带能量的运动意识任务分类研究[J]. 计算机应用与软件, 2010, 27(12):105-107. Ren Yali. Research on task classification of motion awareness based on power spectrum entropy and frequency band energy[J]. Journal of Computer Applications and Software, 2010, 27(12):105-107.
[15] 费成巍, 白广忱, 李晓颖, 等. 基于过程功率谱熵SVM的转子振动故障诊断方法[J]. 推进技术, 2012, 33(2):293-298. Fei Chengwei, Bai Guangchen, Li Xiaoying, et al. Rotor vibration fault diagnosis method based on process power spectral entropy SVM[J]. Propulsion Technology, 2012, 33(2):293-298.
[16] 王凯明, 钟宁, 周海燕, 等. 基于改进功率谱熵的抑郁症脑电信号活跃性研究[J]. 物理学报, 2014, 63(17):178701-178708. Wang Kaiming, Zhong Ning, Zhou Haiyan, et al. Study on the activity of EEG signals in depression based on improved power spectral entropy[J]. Acta Physica Sinica, 2014, 63(17):178701-178708.
[17] Sun Y. EEG signal analysis by using SVM and ELM[D]. Northridge:California State University, Northridge, 2015.
[18] Cortes C, Vapnik V. Support-vector networks[J]. Machine learning, 1995, 20(3):273-297.
[19] Goda K, Jalali B. Dispersive Fourier transformation for fast continuous single-shot measurements[J]. Nature Photonics, 2013, 7(2):102.
[20] Gray R M. Entropy and information theory[M]. New York:Springer Science & Business Media, 2011.
[21] Chang Y W, Hsieh C J, Chang K W, et al. Training and testing low-degree polynomial data mappings via linear SVM[J]. Journal of Machine Learning Research, 2010, 11:1471-1490.
[22] Biau G. Analysis of a random forests model[J]. Journal of Machine Learning Research, 2012, 13:1063-1095.
[23] Borra S, Ciaccio A D. Measuring the prediction error. A comparison of cross-validation, bootstrap and covariance penalty methods[J]. Computational statistics & data analysis, 2010, 54(12):2976-2989.
[24] Pedregosa F, Varoquaux G, Gramfort A, et al. Scikit-learn:Machine learning in Python[J]. Journal of machine learning research, 2011, 12:2825-2830.
