Utilizing changes in human brain connectivity to establish a dose-response relationship involved in the therapeutic actions of prefrontal brain stimulation on depression symptoms

项目总结/文摘

PROJECT SUMMARY/ABSTRACT This is a Diversity Supplement Proposal for Azeezat K. Azeez, Ph.D., entitled “Machine Learning for Predictive Clinical Outcomes to Neuromodulation Therapy for Treatment-Resistant Depression”. It is a Supplement to the Parent R01, held by Nolan Williams, MD titled “5R01MH122754-02: Utilizing changes in human brain connectivity to establish a dose-response relationship involved in the therapeutic actions of prefrontal brain stimulation on depression symptoms”. The goal of the Parent Grant is to (1) test changes in resting-state functional connectivity (rsFC) using functional magnetic resonance imaging (fMRI) scans daily and (2) examine how rsFC changes relate to clinical improvement due to a novel and effective neuromodulation intervention, Stanford Accelerated Intelligent Neuromodulation Therapy (SAINT). This will improve our understanding of the underlying mechanism of Major Depressive Disorder (MDD), particularly Treatment-Resistant Depression. Notwithstanding the high efficacy of SAINT, relative to existing therapeutics a substantial number of participants do fail to respond. Failure to respond, particularly in TRD, leads to detrimental health and economic effects on the participant as well as on the health system. Our lack of ability to predict who will respond to treatment constitutes a major translational gap in the SAINT technology. Therefore, the goal of the current diversity supplement is to employ machine learning algorithms on neuroimaging data to predict who is most likely to respond to treatment. Data science, neuroimaging, and neurostimulation are converging at an exciting junction, the intersection of these disciplines is where the Diversity Supplement lies. A combination of Machine Learning classifier models (supervised and unsupervised) and selection of appropriate imaging features, trained on training data, then tested, and validated will yield a model with high accuracy for predicting clinical outcomes. A combination of these parts will allow us the highest probability of developing a successful algorithm that can be packaged into software to accompany neuromodulation intervention. The current Supplement aims to 1) classify Treatment Response between cohorts; Active, Sham, and Neurotypical Control, and 2) accurately predict remission and response outcomes in Treatment Severity classes. The Diversity Supplement would allow Dr.Azeez to gain proficiency in 1) Machine Learning Techniques, 2) Clinical Assessments, and 3) Professional Development while under the 2-year funding period. Training and research for the project will be conducted at Stanford University which offers excellent intellectual and physical resources to complete the proposed work. The research proposed in the Supplement will help to launch Dr. Azeez’s career in developing Computational Aids for Clinicians in Psychiatric Medicine. This is a major goal of the supplement application and one that will prepare the candidate, Dr. Azeez, for the short-term goal of preparing a competitive NIH K- Award, and the long-term goal for a career as an independent academic researcher. This proposed work has the potential to improve the lives of patients suffering with depression.