Optimizing Memory via Real-time Monitoring of Neural Data with Adaptive Brain Stimulation Machine Learning Algorithms
Session Number
Project ID: MEDH 25
Advisor(s)
James Kragel, University of Chicago
Discipline
Medical and Health Sciences
Start Date
17-4-2024 10:00 AM
End Date
17-4-2024 10:15 AM
Abstract
Aging and neurological disorders lead to neurodegeneration that decreases memory function. Deep brain stimulation (DBS) has emerged as a potential solution by stimulating neurons with electricity to enhance neuronal activity during memory tasks. This study hypothesizes that DBS improves memory when targeted towards deficient memory encoding states due to state-dependent effects by influencing brain networks related to semantic organization. This study utilized data from 38 patients with implanted electrodes performing free-recall tasks with and without stimulation. The spectral data from recorded electrodes, which were fed into a personalized algorithm, modeled each patient's unique neural data, predicting memory encoding state at an area under the curve (AUC) of 98%. A subsequent DBS success classifier, using pre-stimulus memory encoding state qualities, achieved an 84% AUC for predicting memory outcome following stimulation.
This study established that stimulation targeted at memory networks enhances memory by targeting poor encoding states. In addition to state-dependency, it was discovered that stimulation applied to the left lateral temporal cortex (LTC) enhanced the ability to create semantic clusters and recall performance using natural language processing metrics. While non-targeted stimulation leads to multifaceted results, targeted stimulation based on when and where factors introduce a promising treatment for improved episodic memory.
Optimizing Memory via Real-time Monitoring of Neural Data with Adaptive Brain Stimulation Machine Learning Algorithms
Aging and neurological disorders lead to neurodegeneration that decreases memory function. Deep brain stimulation (DBS) has emerged as a potential solution by stimulating neurons with electricity to enhance neuronal activity during memory tasks. This study hypothesizes that DBS improves memory when targeted towards deficient memory encoding states due to state-dependent effects by influencing brain networks related to semantic organization. This study utilized data from 38 patients with implanted electrodes performing free-recall tasks with and without stimulation. The spectral data from recorded electrodes, which were fed into a personalized algorithm, modeled each patient's unique neural data, predicting memory encoding state at an area under the curve (AUC) of 98%. A subsequent DBS success classifier, using pre-stimulus memory encoding state qualities, achieved an 84% AUC for predicting memory outcome following stimulation.
This study established that stimulation targeted at memory networks enhances memory by targeting poor encoding states. In addition to state-dependency, it was discovered that stimulation applied to the left lateral temporal cortex (LTC) enhanced the ability to create semantic clusters and recall performance using natural language processing metrics. While non-targeted stimulation leads to multifaceted results, targeted stimulation based on when and where factors introduce a promising treatment for improved episodic memory.