Understanding and improving language performance in post-stroke aphasia with machine learning, brain networks, and transcranial magnetic stimulation Shreya Parchure

Format:

Book

Thesis/Dissertation

Author/Creator:

Parchure, Shreya, author.

Contributor:

University of Pennsylvania. Bioengineering., degree granting institution.

Language:

English

Subjects (All):

Bioengineering.

Neurosciences.

Medical imaging.

0202.

0317.

0574.

0800.

Local Subjects:

Bioengineering.

Neurosciences.

Medical imaging.

0202.

0317.

0574.

0800.

Genre:

Academic theses

Physical Description:

1 online resource (196 pages)

Contained In:

Dissertations Abstracts International 87-12B

Place of Publication:

Ann Arbor : ProQuest Dissertations and Theses, 2026

Language Note:

English

Summary:

Aphasia, the loss of language ability associated with injury to left-hemisphere brain structures, is the most common focal cognitive deficit after a stroke. There are no neural focused treatments yet available for this condition affecting up to 2 million people in the US. Targeted brain stimulation therapies like repetitive transcranial magnetic stimulation (rTMS) are promising potential treatments for aphasia. However their clinical translation is hampered by a lack of guiding principles about which brain network regions control healthy language cognition, recovery from aphasia, and responses to neuromodulation. We address these gaps in the first half of this thesis by identifying links between structural brain networks and the specific language behaviors they support, across healthy adults and individuals with aphasia. The latter half of studies modulate these identified connections using rTMS, to establish causal brain-behavior relationships and understand individual responses to neuromodulation for optimizing language performance. Chapter 1 introduces the computational and experimental techniques integrated in this thesis: network science, machine learning, and rTMS. Chapter 2 identifies key predictors of speech in aphasia from structural brain networks, linguistic task difficulties, and clinical measures of language performance. This explainable machine learning approach generalizes to predict word-by-word accuracy on any unseen language task and new individuals. Chapter 3 uncovers healthy language network structures associated with task-specific speech production performance. Then we transiently perturb these networks using rTMS and study task-specific language changes to establish causal structure-function links. Chapter 4 investigates heterogeneous responses to rTMS, a common issue that has hampered its advancement to clinical use in aphasia. We develop a neuroplasticity-based biomarker that may help optimize group-level response rates. Chapter 5 integrates previous developments in digital twin models of brain-behavior links, to evaluate individualized aphasia recovery over the course of a clinical trial of rTMS and speech therapy. Chapter 6 synthesizes these network-informed language system neuromodulation findings to posit that brain-behavior links are promising rTMS targets to optimize associated behaviors. The findings and methods contributed by this thesis together serve as a generalizable framework to identify novel neuromodulation targets and stratify personalized treatment responses in aphasia and other complex cognitive-behavioral brain network disorders

Notes:

Source: Dissertations Abstracts International, Volume: 87-12, Section: B.

Advisors: Hamilton, Roy H.; Cohen, Yale Committee members: Meaney, David; Davis, Kathryn; Medaglia, John

Ph.D. University of Pennsylvania 2026

Vendor supplied data

Local Notes:

School code: 0175

ISBN:

9798247973461

Access Restriction:

Restricted for use by site license