Headed by Dr. John Pearson, the Duke Pearson Lab conducts research in computational neuroscience ranging from behavioral modeling using deep neural networks, to human intraoperative, multi-electrode recordings and analyses. My work with the Pearson Lab provided generous experience in computationally tackling big data, a regular chance for patient interaction, and the opportunity to take ownership over the full research process, from the initial experimental design, through the data collection, to the subsequent data analysis.
Over the past year I collected all human-subject data accrued through the lab’s two unique patient populations at the Duke University Hospital. Parkinson’s patients undergoing implantation of Deep Brain Stimulation (DBS) participate in intraoperative tasks aimed at gaining a better understanding of the involvement of the sub thalamic nucleus (STN) in impulsivity, conflict, and motivation. Epilepsy patients being considered for surgical resection of their epileptic focus are implanted with electrocorticography (ECog) grids to characterize and localize seizure activity, allowing our lab to conduct neurological research paradigms (including a theory of mind battery and a competitive game task), while collecting multi-electrode recordings of neural activity.
Additionally, I gained proficiency in coding in Python, a skill which I apply to both experimental design and data analysis. In our most recent project, a colleague and myself designed a dot-discrimination task, to be completed intraoperatively by Parkinson’s patients; after being primed with visual cues corresponding to the task’s varying levels of difficulty, patients are asked to discriminate directionality between several different coherences of dots on a computer screen. The study intends to expound the potential cognitive thresholding effects of STN by assessing how cued differential expectation to accurately discern an ensuing stimuli effects STN dynamics and behavior.
I also generated a data processing pipeline for ECog data collected from our epilepsy patient population, and applied Python, MNE, Jupyter Notebooks, and a variety of other computational tools to clean the data, and to employ complex analyses including the calculation of event related potentials, average power analyses, and time-frequency spectral analyses.