Skip to Main Content
 

Global Search Box

 
 
 
 

Files

File List


HeenaManglani_Dissertation.pdf (3.44 MB)

ETD Abstract Container

Abstract Header

Leveraging multimodal neuroimaging and machine learning to predict processing speed in multiple sclerosis

Manglani, Heena Ramesh
http://orcid.org/0000-0001-8078-6389
http://rave.ohiolink.edu/etdc/view?acc_num=osu1655550909505355

Abstract Details

2022, Doctor of Philosophy, Ohio State University, Psychology.
Multiple sclerosis (MS) is a chronic neurological disease of the central nervous system (CNS) characterized by widespread inflammation, neurodegeneration, and reparation failures. Amongst its sequelae, slowed processing speed remains the earliest predictor of disease burden. MS causes heterogeneous and often subtle changes to functional and structural connections in the brain, even before symptoms manifest. Harnessing neuroimaging-based biomarkers to predict individual prognosis may facilitate patient-centered preventative care before cognitive decline becomes life-limiting. Through leveraging machine learning approaches within a cross-validation framework, we can build models from high dimensional functional and structural whole-brain connectivity to predict individual-level cognition. The present study used neuroimaging data from 64 people with relapsing-remitting MS to construct a multimodal structure-function connectome. We used a data-driven iterative pipeline to train and test models to make continuous predictions of processing speed and quantified model performance through prediction accuracy. Behaviorally, processing speed was significantly correlated with both disease severity and depression scores, confirming shared variance between cognitive and clinical function. However, the multimodal connectome did not yield significant predictions of processing speed in the current sample, and predicted processing speed did not correlate significantly with observed disease severity and depression scores. Separate functional and structural connectomes also did not explain meaningful variance in processing speed. This is the first study to apply machine learning regression techniques in a systematic way across two brain parcellations and both multimodal and unimodal connectomes to make individual-level predictions of cognition in people with MS. Although this study fused structural and functional connectivity using one method, alternative data-driven approaches for building multimodal connectomes implemented in larger samples may capitalize on complementary information across modalities to reveal robust cognitive neuromarkers. This study lays the groundwork for future machine learning and connectomic research to make personalized cognitive predictions in MS.
Ruchika Prakash (Advisor)
Charles Emery (Committee Member)
David Osher (Committee Member)
132 p.
Clinical Psychology
multiple sclerosis; information processing speed; multimodal MRI; functional connectivity; structural connectivity; machine learning, disease severity, depression

Recommended Citations

Citations

  • Manglani, H. R. (2022). Leveraging multimodal neuroimaging and machine learning to predict processing speed in multiple sclerosis [Doctoral dissertation, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1655550909505355

    APA Style (7th edition)

  • Manglani, Heena. Leveraging multimodal neuroimaging and machine learning to predict processing speed in multiple sclerosis. 2022. Ohio State University, Doctoral dissertation. OhioLINK Electronic Theses and Dissertations Center, http://rave.ohiolink.edu/etdc/view?acc_num=osu1655550909505355.

    MLA Style (8th edition)

  • Manglani, Heena. "Leveraging multimodal neuroimaging and machine learning to predict processing speed in multiple sclerosis." Doctoral dissertation, Ohio State University, 2022. http://rave.ohiolink.edu/etdc/view?acc_num=osu1655550909505355

    Chicago Manual of Style (17th edition)

osu1655550909505355
79
© 2022, some rights reserved.Creative Commons License Leveraging multimodal neuroimaging and machine learning to predict processing speed in multiple sclerosis by Heena Ramesh Manglani is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License. Based on a work at etd.ohiolink.edu.
This open access ETD is published by The Ohio State University and OhioLINK.