An active learning framework for quantifying the effect of neuromodulation
Loading...
Author(s)
Eslampanah Sendi, Mohammad Sadegh
Advisor(s)
Editor(s)
Collections
Supplementary to:
Permanent Link
Abstract
Open-loop neuromodulation is a neurosurgical treatment that modulates the brain's neural functioning by delivering an electrical signal using predefined stimulation parameters to a specific deep anatomical structure of the central nervous system. Despite the clinical success of neuromodulation in treating different neurological and neuropsychiatric disorders, the success rate is still not satisfying, and the result varies from patient to patient. The first step in developing a patient-specific neuromodulation therapy is finding the effect of the stimulation on the neural dynamics and linking the stimulation parameters and neural response for each individual. Sampling stimulation parameters to make this link based on the grid search is time-consuming, expensive, and even impossible when the parameters scale up. Optimal sampling based on active learning is an intelligent solution for designing an experiment in which human decision-making is less than optimal for the task. In more detail, active learning is a paradigm in which machine learning models can direct the learning process by providing dynamic suggestions/queries for the "next-best experiment." The main contribution of this Ph.D. dissertation includes 1) developing an analytic pipeline to find the effect biomarker of stimulation, 2) developing an active learning framework to design optimal experiment integrated with an in silico model, 3) implementing the active learning framework in a neuromodulation device for an in vivo model.
Sponsor
Date
2022-04-29
Extent
Resource Type
Text
Resource Subtype
Dissertation