PhD Scientific Days 2023

Budapest, 22-23 June 2023

Neurosciences II.

Detrended Cross-correlation Analysis Reveals Neurophysiological Correlates of Cognitive Deficits Related to Healthy Aging

Zalan Kaposzta MD 1, Frigyes Samuel Racz MD PhD 2
1 Semmelweis University, Department of Physiology, Budapest
2 University of Texas, Austin, Texas

Text of the abstract

Examining long-range cross-correlations between cortical regions can provide very valuable insight into cognitive functioning. Using our published novel real-time detrended cross-correlation analysis (rt-DCCA) formula to carry out such tests may be able to reveal changes naturally consequent of healthy aging as well as pathologic cognitive decline. We analysed EEG data obtained from 45 participants (25 young & 20 elderly, defined by <30 and >60 years of age respectively) during eyes-open and eyes-closed resting conditions (14 cortical regions, 256Hz sampling rate). This was followed by the Cambridge Neurophysiological Test Automated Battery (CANTAB) test, a collection of varied cognitive tasks. Processing the raw EEG data via rt-DCCA provided us with 62335 features. We applied feature selection via a Support Vector Machine classifier to select for the most distinct features. After selection, connectivity analysis was performed between the two groups to examine differences in emerging topological patterns. Cognitive scores were then compared between the two groups in the different utilized tasks. Finally, we explored if the connectivity features identified as the most discriminative between young and old groups show any relation with the cognitive scores obtained. We successfully differentiated the two groups via our calculated features with a test accuracy surpassing 89%, using 56 identified features; from these, the calculated network graphs showed significant stimulus dependent shifts in topology between the two groups. These changes correlated with significantly decreased task performance, and where changes in topology were not present, task performance also remained similar. Using our novel analysis technique, we examined feature differences between healthy young and elderly groups during resting state. We identified the most dissimilar features and used them to reconstruct a group specific network topology. Then, we identified core topological translations that can be attributed to natural aging. Finally, we showed how these changes impacted cognition.

Funding was provided by Semmelweis University’s School of PhD Studies (EFOP-3.6.3-VEKOP-16-2017-00009), and the New National Excellence Program of the Ministry for Culture and Innovation.