Real-time measurement instead of prediction enables high-precision phase-specific stimulation of the human brain

Phase-specific brain stimulation provides a tool to probe the functional relevance of ongoing neural oscillations. Current closed-loop stimulation approaches often use predictive algorithms to compensate for processing and stimulation delays, but prediction error can limit phase-targeting accuracy, particularly at higher frequencies. A real-time approach that minimizes latency without relying on phase prediction may improve the precision and interpretability of phase-specific stimulation.

In this technical proof-of-concept study, we developed and validated a prototype real-time closed-loop system for electroencephalography-triggered transcranial magnetic stimulation (EEG-TMS). The system performs signal acquisition, phase estimation, and stimulation triggering within a 1 ms closed-loop update cycle and does not require phase prediction. Empirical testing was performed in human participants receiving EEG-triggered TMS across eight target phases and ten target frequencies ranging from 4 to 40 Hz.

Across all target phases and frequencies, the system achieved single-digit phase accuracy without phase prediction. On average, stimulation was 5.6° away from the intended target phase. The typical single-pulse error was 3.7°, and trial-to-trial variability was 7.8° These results indicate stable phase-targeting performance across the tested phase and frequency range.

This real-time EEG-TMS system demonstrates the technical feasibility of prediction-free phase-specific stimulation across physiologically relevant frequencies. By reducing dependence on predictive phase extrapolation, the approach provides a platform for studying phase-dependent brain physiology and for developing closed-loop neuromodulation protocols with improved temporal precision.