Phase estimation with autoregressive padding (PEAP): addressing inaccuracies and biases in EEG analysis

A team of researchers including Miriam Kirchhoff and six colleagues has published a new paper introducing Phase Estimation with Autoregressive Padding (PEAP), a method designed to solve critical accuracy problems in electroencephalogram (EEG) analysis. Their research evaluated four established phase estimation methods—Phastimate, SSPE, ETP, and PhastPadding—and found they all suffer from systematic biases and induced phase shifts, particularly at data segment edges where accurate estimation is crucial for applications like EEG-TMS (transcranial magnetic stimulation). These biases compromise the validity and comparability of phase-dependent neuroscience findings.

PEAP addresses these limitations by preventing strong bandpass filtering-induced artifacts that plague existing methods. The new approach uses autoregressive padding to handle edge effects more effectively, resulting in 3.2% to 9.2% improved accuracy for continuous phase estimation without showing the significant biases of established methods. Importantly, the researchers demonstrated that method differences don't vary between clinical populations (chronic stroke patients) and healthy controls, supporting PEAP's translatability to real-world medical applications where reliable brain signal analysis is essential for diagnosis and treatment monitoring.