From group-level neurophysiological biomarkers to individual-level phenotyping to identify mechanisms of neurodegenerative disease

In resting-state MEG of 74 healthy controls, 45 bvFTD, and 55 PSP participants, clinical scores classify individuals strongly (controls-vs-PSP accuracy 0.94; bvFTD-vs-PSP accuracy 0.83).

• Why it matters: Group-level physiological biomarkers may not translate to individuals when distributions overlap and disease mechanisms are heterogeneous, limiting precision medicine and mechanistic inference.
• What they did: The study combined clinical phenotype, power spectral density features, and latent network mechanisms from dynamic causal modelling on resting-state MEG, then tested individual assignment via classification.
• The result: MEG spectra supported only weak individual classification (accuracy ~0.6) despite group re-organization, while inferred biophysical parameters gave intermediate performance (PSP-vs-control accuracy 0.70; bvFTD-vs-PSP accuracy 0.74) with mechanistic interpretability.