Thalamocortical dysrhythmia detected by machine learning.

Summary

Thalamocortical dysrhythmia (TCD) is characterised by resting-state alpha being replaced by cross-frequency coupling of low- and high-frequency oscillations. Taking a data-driven approach, the authors applied support-vector-machine learning to resting-state EEG from patients with Parkinson's disease, neuropathic pain, tinnitus and depression. They found a spectrally equivalent but spatially distinct form of TCD depending on the disorder, while also identifying brain areas common across all four, turning TCD from a theoretical construct into a measurable, disorder-specific signature.

Key findings

• Applied support-vector-machine learning to resting-state EEG across four disorders including tinnitus.
• Found a spectrally equivalent but spatially DISTINCT TCD pattern per disorder.
• Also identified brain areas common to Parkinson's, pain, tinnitus and depression.
• Moves TCD from theory to a detectable, data-driven signature in humans.