Graduate Seminar Series: Clinical Stream – Srdjan Sumarac

Graduate Seminar Series: Clinical Stream
Graduate Seminar Series for the Institute of Biomedical Engineering (BME). This day is for clinical stream presenters.
If you would like to invite your Principal Investigator, please add their email via the ‘Add Guest’ button and they will also be notified of your presentation.
Presentation Title: Machine learning approach for predicting severity of symptoms in Parkinson’s disease from intraoperative single-neuron recordings
Abstract: We hypothesized that pathophysiologically-relevant information about the symptoms of Parkinson’s disease may be directly encoded within neurophysiological features of individual neurons in the basal ganglia. Using an extensive database of intracranial recordings acquired during awake deep brain stimulation surgery (n=225 patients), we calculated spiketrain features (e.g., firing rate, burst index, and spiketrain oscillations in different frequency bands) from 1614 high-quality single-neuron segments of the subthalamic nucleus, substantia nigra pars reticulata, or globus pallidus internus. Corresponding preoperative Unified Parkinson’s Disease Rating Scale Part III (UPDRSIII) motor scores were also amalgamated for each patient. To account for multiple observations per patient, we used linear mixed models (LMM) to study the relationships between spiketrain features and UPDRSIII scores. We also trained a light gradient boosting machine (LightGBM) regressor combined with mixed effects random forests (MERF) to predict the severity of patients’ parkinsonian symptoms using a combination of features. Our machine learning framework could accurately predict patients’ total UPDRSIII scores across various symptoms and disease subtypes. Our feature importance analysis also revealed that synchronization coding generally had a higher impact on model prediction than rate coding. Overall, our findings suggest that the activity of individual neurons contains valuable information about parkinsonian symptom severity.
Supervisor Name: Luka Milosevic, Milos R Popovic
Year of Study: 2
Program of Study: PhD
Zoom link: https://us02web.zoom.us/j/89610372821?pwd=azd4SCtYVWtreVovaGNPV1c2NGY2Zz09
Meeting ID: 896 1037 2821
Password: 483329
Powered by Calendly.com