In order to support the development of our products and turn towards cutting edge projects, the company keeps a constant watch on scientific discoveries and technological improvements.

EEG Biomarkers in Alzheimer’s disease


Alzheimer’s Disease (AD) is the most common type of dementia in humans. It is a neurodegenerative disorder during which neural tissue is gradually degraded (death of neurons) leading to progressive loss of cognitive functions such as memory.  The disease was first described in 1901 by the german psychiatrist and neuropathologist Alois Alzheimer for whom the disease is named.  The pathophysiology of AD remains largely unknown to date, even though both genetic and environmental factors, such as exposure to heavy metals, are suspected to contribute to its development.

Background data on Alzheimer’s Disease (AD) Pathophysiology

First symptoms usually appear in the first half of the sixth decades with a large predominance in men. They include, the loss of short term memory (amnesia) leading to minor distractions and possibly depression that gradually increases with the disease progression. The early stage of the disease can not easily be distinguished from Mild Cognitive Impairment (MCI) that has another etiology. Interestingly, the long-term memory is preserved so that childhood memory  for instance may remain until late stage of the condition. The condition then usually spreads to associative cortex: frontal and temporal-parietal leading to a worsening of the psychiatric variables: confusion, irritability, aggression, mood and emotions, executive functions and language. At later stages, the disease spreads to the autonomous nervous system – the “deepest” part of the brain that controls life functions – eventually causing death. The life expectancy from diagnosis varies from 3 to 8 years, which makes prognosis largely unreliable.


Worldwide, about 26m people were diagnosed with AD in 2005 (Feri 2012), which could increase four-fold or more in 2050  yielding to an astonishing approximated 1.2% of the world population (Brookmeyer 2007). In developed countries, AD is already one of the most important financial burden for society (Bonin-Guillaume 2005, Meek 1998).  Likewise, according to Business Insights (Advances in Alzheimer’s Disease Drug Discovery, 2011), “an estimated 35.6 million people worldwide had dementia in 2010. This number will double every 20 years, reaching 65.7 million in 2030 and 115.4 million in 2050.” The costs associated with managing the disease are staggering: “The total estimated worldwide costs of dementia, including direct and indirect costs of care, were $604bn in 2010.”

Diagnosis and treatment

To date, the diagnosis of Alzheimer’s disease mostly relies on questionnaires, tests and neuroanatomical imagery, which first reveals the atrophy of the internal lobe of the hippocampus. Alzheimer’s Disease is a slow progressing terminal condition, meaning that no treatment reversing the condition exists. Different strategies can be implemented to slow the progression of the disease: cognitive exercise and diet, which is often suggested and implemented by caretakers and relatives, who ultimately bear the real burden. The chemical treatments are also extremely few (see Table 1 below) with limited promises and non-negligible side effects. Eventually, palliative care can be recommended.

Drug treatment of Alzheimer’s Disease relies on two approved pharmacological classes (NMDA antagonists and Cholinesterase inhibitors) which claim limited efficacy on the symptoms of Alzheimer’s disease. To date, no pharmacological treatment has demonstrated prevention or slowing down of the disease leading to a large unmet medical need still to be addressed.

Recent trends in EEG diagnosis (since 2011)

Given the importance of AD for public health, there is a strong rationale for the development of reliable and operator-independent diagnosis or assessment tools. If such tools could provide good calibration properties (the ability to assign patients to the right range of predicted risk), they would also turn out extremely useful for the monitoring of the evolution of the disease, which can be used to assess the efficacy of a given treatment. Consequently, the literature dealing with such topic is large and we present here its most significant subset for the four years between 2010 and 2014.


First of all, recent work still features databases of relatively small sample size (n<30; Sankari2011, Ahmadlou2011, Simons2015), but there is a growing evidence of the diagnosis ability of EEG-based techniques, which is derived from larger database (n>100; wall 2011, Staudinger2011, Poil2015, Ommundsen2011, Moretti2012).


In terms of features, most studies include some kind of frequency analysis from which features are then extracted using complexity (Ahmadlou2011, Garn2014, Staudinger2011, Simons2015), information theory (Dauwels2012, Garn2014, Morison2013), wavelet (Sankari2012, Ahmadlou2011, Ghorbarian2013), and connectivity (Sankari2012).


The modeling techniques implemented are traditional techniques from biostatistics such as ANOVA (Sankari2011) and machine learning predictive models: logistic regression (Waal 2011, Trambaiolli2011, Poil2013), linear discriminant analysis (Ahmadlou2011), support vector machines (Mcbride2014, Staudinger2014, Trambaiolli2011), neural networks (Sankari2011, Ahmadlou2011, Staudinger2014), and regression trees (Ghorbarian2013, Trambaiolli2011).


The most predictive variables identified in these studies are: beta fractal dimension (Ahmadlou2011), left temporo-central-parietal decrease in wavelet coherence in delta bands (Sankari2011), auto mutual-information at T7 (Garn2014), delta bands (Ghorbarian2013), beta bands (Poil2013), T4-F4-Fp2 (Simons2015), and alpha III to alpha II ratio (Moretti2012). Thankfully, results from the smaller studies remains relatively consistent with accuracies reported in the nineties in most cases (Ahmadlou2011, Sankari2011, Dawnels2010, Trambaiolli2011, Poil2013, Simons2015). Interestingly, a correlation can be seen between performance and sample size, which is a relatively common phenomenon in the field since complex machine learning techniques applied to small sample size is particularly prone to over-fitting. Thankfully, even studies on largest populations report a significant discriminatory effect (Staudinger2011, Ommundsen2011) and identified markers that remain relatively consistent.


Given the heterogeneity of populations, features, and techniques used, it is difficult to draw any strong conclusion about which technique (or set of feature) performs the best. Also, it may eventually that there may not exist a “one-size fits all” test but that many different algorithms may ferret out different subtly different sub-types of the disease sub-types, with comparable outcomes. These results suggests that the discriminatory information might spread over different location, frequency bands, and features, which builds a strong rationale for the use of advanced machine learning techniques and diversified tests, applied to very large and consistent databases, which Mensia Technologies currently implements.

A step towards EEG-based treatment

As for many other psychiatric conditions for which quantitative EEG has been successfully applied for diagnosis and monitoring, neurofeedback stands as a promising treatment opportunity. This is particularly true for Alzheimer’s disease where existing chemical and behavioral therapies are merely delaying the progression of the disease. Consequently, the last few years have seen the emergence of pilot clinical trials evaluating the efficacy of neurofeedback with sometimes astonishing results. In particular, Lee et al. report on fifteen AD patient a statistically significant effect on immediate memory, visuo-spatial construction, attention, and delayed memory, which was also notably associated with the absence of adverse effect (Lee2014). Mensia Technologies strongly believes in the potential of these techniques and offers its real-time neurophysiology cloud platform to implement these neuromarkers that are currently evaluated in Asia.


Ferri, Cleusa P., et al. “Global prevalence of dementia: a Delphi consensus study.” The Lancet 366.9503 (2006): 2112-2117.

Brookmeyer, Ron, et al. “Forecasting the global burden of Alzheimer’s disease.”Alzheimer’s & dementia 3.3 (2007): 186-191.

Bonin-Guillaume, Sylvie, et al. “Impact économique de la démence.” La Presse Médicale 34.1 (2005): 35-41. Meek, Patrick D., E. Kristin McKeithan, and Glen T. Schumock. “Economic considerations in Alzheimer’s disease.” Pharmacotherapy: The Journal of Human Pharmacology and Drug Therapy 18.2P2 (1998): 68-73.

Ahmadlou, Mehran, Hojjat Adeli, and Anahita Adeli. “Fractality and a Wavelet-chaos-Methodology for EEG-based Diagnosis of Alzheimer Disease.” Alzheimer Disease & Associated Disorders 25.1 (2011): 85-92.

Sankari, Ziad, Hojjat Adeli, and Anahita Adeli. “Wavelet coherence model for diagnosis of Alzheimer disease.” Clinical EEG and Neuroscience 43.4 (2012): 268-278.

Sankari, Ziad, and Hojjat Adeli. “Probabilistic neural networks for diagnosis of Alzheimer’s disease using conventional and wavelet coherence.” Journal of neuroscience methods 197.1 (2011): 165-170.

Dauwels, Justin, et al. “A comparative study of synchrony measures for the early diagnosis of Alzheimer’s disease based on EEG.” NeuroImage 49.1 (2010): 668-693.

Sankari, Ziad, Hojjat Adeli, and Anahita Adeli. “Intrahemispheric, interhemispheric, and distal EEG coherence in Alzheimer’s disease.” Clinical Neurophysiology 122.5 (2011): 897-906.

Ahmadlou, Mehran, Hojjat Adeli, and Anahita Adeli. “New diagnostic EEG markers of the Alzheimer’s disease using visibility graph.” Journal of neural transmission 117.9 (2010): 1099-1109.

de Waal, Hanneke, et al. “EEG abnormalities in early and late onset Alzheimer’s disease: understanding heterogeneity.” Journal of Neurology, Neurosurgery & Psychiatry 82.1 (2011): 67-71.

Garn, Heinrich, et al. “Electroencephalographic complexity markers explain neuropsychological test scores in Alzheimer’s disease.” Biomedical and Health Informatics (BHI), 2014 IEEE-EMBS International Conference on. IEEE, 2014.

McBride, Joseph C., et al. “Spectral and complexity analysis of scalp EEG characteristics for mild cognitive impairment and early Alzheimer’s disease.”Computer methods and programs in biomedicine 114.2 (2014): 153-163.

Staudinger, Tyler, and Robi Polikar. “Analysis of complexity based EEG features for the diagnosis of Alzheimer’s disease.” Engineering in Medicine and Biology Society, EMBC, 2011 Annual International Conference of the IEEE. IEEE, 2011.

Morison, Gordon, Zoë Tieges, and Kerry Kilborn. “Multiscale permutation entropy analysis of the EEG in early stage alzheimer’s patients.” Biomedical Engineering and Sciences (IECBES), 2012 IEEE EMBS Conference on. IEEE, 2012.

Ghorbanian, Parham, et al. “Identification of resting and active state EEG features of Alzheimer’s disease using discrete wavelet transform.” Annals of biomedical engineering 41.6 (2013): 1243-1257.

Trambaiolli, Lucas R., et al. “EEG spectro-temporal modulation energy: a new feature for automated diagnosis of Alzheimer’s disease.” Engineering in Medicine and Biology Society, EMBC, 2011 Annual International Conference of the IEEE. IEEE, 2011.

Poil, Simon-Shlomo, et al. “Integrative EEG biomarkers predict progression to Alzheimer’s disease at the MCI stage.” Frontiers in aging neuroscience 5 (2013).

Simons, Samantha, Daniel Abasolo, and Michael Hughes. “Investigation of Alzheimer’s Disease EEG Frequency Components with Lempel-Ziv Complexity.” 6th European Conference of the International Federation for Medical and Biological Engineering. Springer International Publishing, 2015.

Trambaiolli, Lucas R., et al. “Improving Alzheimer’s disease diagnosis with machine learning techniques.” Clinical EEG and Neuroscience 42.3 (2011): 160-165.

Ommundsen, Nina, Knut Engedal, and A. R. Øksengård. “Validity of the quantitative EEG statistical pattern recognition method in diagnosing Alzheimer’s disease.” Dementia and geriatric cognitive disorders 31.3 (2011): 195-201.

Moretti, D. V., et al. “Specific EEG changes associated with atrophy of hippocampus in subjects with mild cognitive impairment and Alzheimer’s disease.” International journal of Alzheimer’s disease 2012 (2012).

Lee, Tih-Shih, et al. “A Brain-Computer Interface Based Cognitive Training System for Healthy Elderly: A Randomized Control Pilot Study for Usability and Preliminary Efficacy.” PloS one 8.11 (2013): e79419.