Multiple tests based on a gaussian approximation of the unitary events method with delayed coincidence count

Area of Science:

• Computational Neuroscience
• Statistical Signal Processing
• Neuroscience

Background:

• The unitary events (UE) method is widely used for detecting coincident neural spike activity.
• Traditional binned coincidence counts suffer from synchrony detection loss.
• The multiple shift coincidence count improved detection but lacked statistical investigation.

Discussion:

• This work introduces a delayed coincidence count, generalizing the multiple shift count for both discretized and non-discretized point processes.
• A novel Gaussian approximation for the coincidence count is proposed, incorporating a plug-in step for unknown parameters.
• The Benjamini and Hochberg approach controls the false discovery rate when multiple tests are performed.

Key Insights:

• The new method, MTGAUE (multiple tests based on a Gaussian approximation of the unitary events), extends the validity of previous UE methods.
• MTGAUE can detect both an excess and a lack of coincidences relative to independence.
• The method demonstrates robustness to underlying model changes and is validated on real neural data.

Outlook:

• Further statistical investigation into the properties of the delayed coincidence count is warranted.
• Exploring the application of MTGAUE to other complex time-series analysis problems in science.
• Expanding the use of MTGAUE for analyzing large-scale neural recordings and understanding neural network dynamics.