Closed-Loop Neuromodulation for Brain Fatigue: From Real-Time Biomarkers to Adaptive Intervention

Brain fatigue is a debilitating condition whose molecular complexity—neurotransmitter imbalance, neuroinflammation, and metabolic failure—has long defied effective intervention. Conventional open-loop neuromodulation offers fixed stimulation, but fatigue fluctuates. Closed-loop neuromodulation, adjusting stimulation in real time, holds real promise. Yet its delivery hinges on a single, unresolved bottleneck: the sensing–decision chain. In this review, we deconstruct closed-loop systems into sensing, decision, and intervention, and argue that the true challenge is not technological but informational—how to integrate fast electrophysiological signals with slow molecular biomarkers (inflammatory cytokines, neurotrophic factors, adenosine) into a unified control framework. We show that a “fast-slow variable” architecture offers a practical path forward: fast signals guide immediate responses, slow variables set baselines and thresholds, and their integration enables predictive, pre-emptive intervention. We also examine the molecular correlates of neuromodulation—synaptic plasticity, anti-inflammatory signaling, and neurotrophic regulation—as the mechanistic foundation for therapeutic effect. Finally, we confront the translational triad of causality, inter-individual variability, and the information–energy–time trade-off. We conclude that realizing effective closed-loop neuromodulation for brain fatigue will require parallel advances in both stimulation hardware—improving spatial targeting, dose precision, and modality versatility—and individualized sensing–decision algorithms capable of reliably translating multi-modal biomarkers into timely, safe interventions. Both areas remain active fronts of development, and neither can be neglected in clinical translation.