Integrating mechanistic insights and models with machine learning: Applications in addressing neurological disorders
Saboo, Krishnakant V.
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/122094
Description
Title
Integrating mechanistic insights and models with machine learning: Applications in addressing neurological disorders
Author(s)
Saboo, Krishnakant V.
Issue Date
2023-09-08
Director of Research (if dissertation) or Advisor (if thesis)
Iyer, Ravishankar K
Doctoral Committee Chair(s)
Iyer, Ravishankar K
Committee Member(s)
Srikant, Rayadurgam
Koyejo, Oluwasanmi
Worrell, Gregory A
Department of Study
Electrical & Computer Eng
Discipline
Electrical & Computer Engr
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Machine learning
reinforcement learning
mechanistic models
mechanistic insights
brain disorders
Alzheimer's disease
epilepsy
disease progression modeling
cognition prediction
seizure cluster prediction
brain stimulation
Abstract
Neurological disorders currently affect over one billion people and are the leading cause of disability worldwide. This thesis develops a novel framework that integrates mechanistic models and insights with machine learning (ML) to address neurological disorders by diagnosing them early and accurately, developing treatments, and improving our understanding of their underlying biology. Mechanistic insights and models express known disease-related biological processes either implicitly or explicitly as mathematical relationships amongst pertinent variables. These implicit/explicit relationships mitigate the noise in multi-modal data and help ML models efficiently extract unknown relationships related to the disease. Thus, our approach provides a holistic view of dynamically evolving, multi-scale, complex diseases in the face of unknown biological relationships by overcoming data-related challenges such as partial observability, high dimensionality, noise, and limited data. Our approach led to novel ML-based techniques that (i) work well with limited data, (ii) express and learn complex relationships, and (iii) capture causal relationships, which enabled diagnosis and in silico exploration of treatments. This thesis exemplifies the proposed framework by addressing clinically important problems in (i) Alzheimer’s disease (AD) and (ii) epilepsy. (i) Integrating mechanistic models with reinforcement learning (RL) enabled early diagnosis of individuals at risk of future cognitive decline by predicting personalized 10-year AD progression. (ii) We developed a unique mechanistic state-space model of epilepsy that could forecast seizures several days in advance and demonstrated its utility for designing RL-based adaptive brain stimulation to alleviate seizures. Our approach also led to new insights on short-term memory, localization of epileptogenic tissue, individualized prediction of seizure clusters, and the discovery of brain structures that mitigate the effect of AD on cognition. The proposed framework can be extended to improve the diagnosis, treatment, and understanding of other diseases of the body.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
