Objective:

To evaluate the efficacy of biomarker-guided brain stimulation in improving memory performance and to introduce a novel neurostimulation platform for real-time, AI-driven memory enhancement.

Background:

Closed-loop brain stimulation is emerging as a promising method for modulating neural networks responsible for cognitive and affective disorders.

Design/Methods:

Patients undergoing invasive monitoring for epilepsy surgery took part in memory experiments, and we trained machine-learning classifiers to predict momentary lapses in memory performance. Using these classifiers to control brain stimulation in subsequent closed-loop sessions, we evaluated stimulation’s acute mnemonic effects in a sham-controlled, double-blind design. As a first step towards evaluating this therapy in a chronic setting we prototyped a 64-channel fully-implantable system and report preliminary data in an ovine model.

Results:

In acute studies involving 47 patients with implanted electrodes we demonstrated reliable mnemonic benefits of closed-loop stimulation near white matter tracts of the lateral temporal cortex. Stimulation also improved biomarkers of memory encoding. In a preclinical ovine study we also demonstrated the capabilities of a 64-channel system capable of delivering this therapy in clinical trials for patients with memory loss.

Conclusions:

Closed-loop brain stimulation can improve memory performance by selectively modulating networks when they shift into less functional states. This approach shows promise in both epilepsy and TBI populations, with future clinical applications supported by novel, AI-driven neuromodulation technologies.