From nodal- to network-centric, state-based neuromodulation for neuropsychiatric disorders

Abstract Deep brain stimulation (DBS) has progressed from modulation of single-site nuclei into an adopted therapy for distributed network neuromodulation, with expanding indications spanning movement disorders, psychiatry, pain, epilepsy, and cognition. However, dominant DBS paradigms remain anchored to single-site assumptions, fixed stimulation parameters, and unimodal biomarker control, which increasingly limit selectivity and scalability as indications grow more dimensionally complex. This review proposes that the next critical advance for DBS is network-based, state-based control, in which stimulation is designed around modulating network dynamics as they evolve over time rather than targeting functionally isolated anatomical nodes. We trace the progression from nucleus-centric targeting toward connectivity-informed strategies, identifying three emerging architectural paradigms: single nodal hubs, tract and intersecting bundle targets, and multi-nodal configurations. We then examine how control policies, including continuous, adaptive, and patterned stimulation, must be matched to the temporal and functional properties of the targeted networks to achieve effective modulation. Finally, we argue that advancing beyond single-biomarker adaptive DBS toward multimodal, multi-feature state estimation is essential for capturing the heterogeneous symptom profiles characteristic of multi-network disorders. This requires integrating connectomic structural mapping, electrophysiological decoding from cortical and subcortical recordings, and behavioral kinematics to estimate symptom vectors rather than scalar severity measures. Together, the requirements of structural network mapping, network-aligned control policies, and robust state decoding constitute a principled framework for scalable, personalized neuromodulation capable of addressing circuit dysfunction across neurological and psychiatric conditions.