Karim, Helmet
(2017)
Early Prediction Of Late-Life Depression Remission: Multi-Factor Kernel-Based Machine Learning Utilizing Single Dose Pharmacological Functional Magnetic Resonance Imaging.
Doctoral Dissertation, University of Pittsburgh.
(Unpublished)
Abstract
Treatment of major depressive disorder (MDD) currently relies on a prolonged trial and error process to identify the best pharmacological regimen. This process is further prolonged in older adults with major depressive disorder (Late-Life Depression or LLD), where it is associated with a host of negative outcomes, including suicide, worsening medical comorbidity, and poor quality of life. Functional magnetic resonance imaging (fMRI) brain changes have been associated with depression severity and treatment outcomes. Previous studies have shown that recovery from depression can be predicted using both pre-treatment neuroimaging as well as follow-up scans from the early treatment period. Pharmacological functional magnetic resonance imaging (phMRI) is an approach that utilizes multiple fMRI scans to investigate changes in functional neuroimaging following acute doses of pharmacotherapy. It has been demonstrated that antidepressants have a fast uptake period, effecting resting state networks as well as functional brain activation after only a single dose. We aimed to evaluate the efficacy of phMRI to identify these very early (single dose) functional changes, and use these to predict remission. Data was collected from an open-label pharmacologic treatment study of LLD (N=51). Multi-modal MRI, including phMRI, were acquired at 5 time-points. Results showed accurate prediction of depression remission from pre-treatment, as well as phMRI after only a single dose of pharmacotherapy. The trajectory of the neuroimaging changes across the treatment trial suggest an initial engagement of large scale resting networks, followed by engagement of implicit emotion control networks, and later changes in explicit emotion regulation. Utilizing kernel-based (multi-factor principal components) machine learning, we found that leveraging both pharmacological neuroimaging and clinical data improved prediction efficacy of remission. In this body of work, we have integrated multiple imaging modalities to explain the long delay in clinical response to antidepressants, and to identify early markers of response.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
26 September 2017 |
| Date Type: |
Publication |
| Defense Date: |
26 May 2017 |
| Approval Date: |
26 September 2017 |
| Submission Date: |
8 June 2017 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
163 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
Swanson School of Engineering > Bioengineering |
| Degree: |
PhD - Doctor of Philosophy |
| Thesis Type: |
Doctoral Dissertation |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
Depression; Late-Life; Machine Learning; PCA; Multi-factor Analysis; Acute Pharmacological Effect |
| Date Deposited: |
26 Sep 2017 16:50 |
| Last Modified: |
26 Sep 2017 16:50 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/32413 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}