Test-retest reliability of spectral parameterization by 1/f characterization using SpecParam

Daniel McKeown*, Anna Finley, Nickolas Kelley, James Cavanagh, Hannah Keage, Oliver Baumann, Victor R. Schinazi, Ahmed Abdelhaim Moustafa, Doug Angus

*Corresponding author for this work

Research output: Other contributionDiscipline Preprint RepositoryResearch

51 Downloads (Pure)


SpecParam (formally known as FOOOF) allows for the refined measurements of electroencephalography periodic and aperiodic activity, and potentially provides a non-invasive measurement of excitation:inhibition balance. However, little is known about the psychometric properties of this technique. This is integral for understanding the usefulness of SpecParam as a tool to determine differences in measurements of cognitive function, and electroencephalography activity. We used intraclass correlation coefficients (ICC) to examine the test-retest reliability of parameterized activity across three sessions (90 minutes apart and 30 days later) in 49 healthy young adults at rest with eyes open (EO), eyes closed (EC), and during three EC cognitive tasks including subtraction (Math), music recall (Music), and episodic memory (Memory). ICCs were good for the aperiodic exponent and offset (ICCs > .70) and parameterized periodic activity (ICCs > .66 for alpha and beta power, central frequency, and bandwidth) across conditions. Across all three sessions, SpecParam performed poorly in EO (40% of participants had poor fits over non-central sites) and had poor test-retest reliability for parameterized periodic activity. SpecParam mostly provides reliable metrics of individual differences in parameterized neural activity. More work is needed to understand the suitability of EO resting data for parameterization using SpecParam.
Original languageEnglish
Media of outputArticle
PublisherbioRxiv - the preprint server for Biology
Number of pages37
Publication statusSubmitted - 23 Sept 2023


Dive into the research topics of 'Test-retest reliability of spectral parameterization by 1/f characterization using SpecParam'. Together they form a unique fingerprint.

Cite this