Characterizing the spiking dynamics of subthalamic nucleus neurons in Parkinson's disease using generalized linear models

Uri T. Eden, John T. Gale, Ramin Amirnovin, Emad N. Eskandar

Research output: Contribution to journalArticlepeer-review

7 Scopus citations


Accurately describing the spiking patterns of neurons in the subthalamic nucleus (STN) of patients suffering from Parkinson's disease (PD) is important for understanding the pathogenesis of the disease and for achieving the maximum therapeutic benefit from deep brain stimulation (DBS). We analyze the spiking activity of 24 subthalamic neurons recorded in Parkinson's patients during a directed hand movement task by using a point process generalized linear model (GLM). The model relates each neuron's spiking probability simultaneously to factors associated with movement planning and execution, directional selectivity, refractoriness, bursting, and oscillatory dynamics. The model indicated that while short-term history dependence related to refractoriness and bursting are most informative in predicting spiking activity, nearly all of the neurons analyzed have a structured pattern of long-term history dependence such that the spiking probability was reduced 20-30 ms and then increased 30-60 ms after a previous spike. This suggests that the previously described oscillatory firing of neurons in the STN of Parkinson's patients during volitional movements is composed of a structured pattern of inhibition and excitation. This point process model provides a systematic framework for characterizing the dynamics of neuronal activity in STN.

Original languageEnglish (US)
JournalFrontiers in Integrative Neuroscience
Issue numberJUNE 2012
StatePublished - Jun 20 2012
Externally publishedYes


  • Modeling
  • Oscillations
  • Parkinson's
  • STN
  • Spikes

ASJC Scopus subject areas

  • Sensory Systems
  • Cognitive Neuroscience
  • Cellular and Molecular Neuroscience


Dive into the research topics of 'Characterizing the spiking dynamics of subthalamic nucleus neurons in Parkinson's disease using generalized linear models'. Together they form a unique fingerprint.

Cite this