Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Predicting clinical variables from neuroimages using fused sparse group lasso

Beer, Joanne C (2018) Predicting clinical variables from neuroimages using fused sparse group lasso. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Submitted Version

Download (6MB) | Preview


Predictive models in which neuroimage features serve as predictors and a clinical variable is modeled as the outcome are good candidates for clinical application because (1) they can exploit dependencies between predictor variables and thus potentially explain more variability in the outcome than a mass univariate approach, and (2) they allow inference at the individual level, such that a prediction can be obtained for a new individual whose data was not used to train the model. This dissertation proposes methods for neuroimaging prediction models that not only aim for predictive accuracy, but also seek interpretability and potential insight into the underlying pathophysiology of neuropsychiatric disorders.

In the first part of this dissertation we propose the fused sparse group lasso penalty, which encourages structured, sparse, interpretable solutions by incorporating prior information about spatial and group structure among voxels. We derive optimization steps for fused sparse group lasso penalized regression using the alternating direction method of multipliers algorithm. With simulation studies, we demonstrate conditions under which fusion and group penalties together outperform either of them alone. We then use fused sparse group lasso to predict continuous measures from resting state magnetic resonance imaging data using the Autism Brain Imaging Data Exchange dataset. In the second part of this dissertation we use fused sparse group lasso to predict age from multimodal neuroimaging data in a sample of cognitively normal adults aged 65 and older. In general, we show how the incorporation of prior information via the fused sparse group lasso penalty can enhance the interpretability of neuroimaging predictive models while also yielding good predictive performance.

Public health significance: Psychiatric disorders and neurological diseases such as Alzheimer's present a large public health burden. As of yet, there have been relatively few translations of basic neuroscience findings to clinical applications in psychiatry. Prediction models using neuroimaging data can potentially help clinicians with diagnosis and prediction of prognosis and treatment response. Establishing interpretable neuroimaging-based biomarkers can improve our understanding of the neurobiological mechanisms underlying neuropsychiatric disorders and suggest approaches for prevention and treatment.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Beer, Joanne Cjcb117@pitt.edujcb1170000-0001-8583-8467
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Thesis AdvisorAnderson, Stewartsja@pitt.edusja0000-0001-8948-0650
Thesis AdvisorKrafty, Robertrkrafty@pitt.edurkrafty0000-0003-1478-6430
Committee MemberTudorascu, Danadlt30@pitt.edudlt300000-0003-4675-3692
Committee MemberAizenstein, Howardaizen@pitt.eduaizen0000-0003-4897-6582
Date: 20 September 2018
Date Type: Publication
Defense Date: 17 July 2018
Approval Date: 20 September 2018
Submission Date: 11 July 2018
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 130
Institution: University of Pittsburgh
Schools and Programs: School of Public Health > Biostatistics
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: autism, neuroimaging, penalized regression, predictive model, regularization, structured sparsity
Related URLs:
Date Deposited: 20 Sep 2018 21:45
Last Modified: 20 Sep 2018 21:45


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item