Cognitive and brain function enhancement in Gen X group after personalized, AI supervised EEG-neurofeedback training

Abstract

Background: Interventions supporting medical care and enhancing quality of life in neurodegenerative or age-related cognitive decline are strongly needed. Electroencephalographic (EEG) neurofeedback can enable users to modulate their brain activity through real-time feedback. However, evidence for its clinical effectiveness remains inconclusive, partly due to limited personalization and insufficient task relevance in existing protocols. Objective: We tested whether personalized EEG neurofeedback supervised by deep neural networks (DNNs) can enhance cognitive performance in older adults. Methods: Fifty-seven healthy adults aged 41 to 64 (31 women), including a sham-feedback control group, completed a personalized neurofeedback protocol with DNNs fine-tuned to individual EEG patterns. The procedure included pre- and post-training assessments using a transitive reasoning task, three diagnostic sessions to adapt the DNN to each participant, and 10 neurofeedback sessions based on a gamified delayed-match-to-sample paradigm. Results: The training group showed robust gains across all three variants of the reasoning task (each p < .01), whereas the sham group improved only on the easiest variant. Groups did not differ at pretest; however, at posttest the training group outperformed the sham group on all task conditions (each p < .03), showing also a larger neural effort (lower alpha band power) and increased beta and gamma band connectivity (higher phase lag index). Conclusion: Personalized, task-oriented neurofeedback guided by individually fine-tuned DNNs can produce cognitive enhancement after relatively few sessions. The proposed Task-Pretrained, Subject-Finetuned Neurofeedback (TPSF-NF) framework is scalable to other cognitive domains in future research.