Boston University
Machine learning–guided discovery of brain-targeted lipid nanoparticles through computational modeling and experimental validation
Abstract
dc:description.abstractLipid nanoparticles (LNPs) have emerged as a leading platform for the delivery of nucleic acid therapeutics, enabling the clinical success of mRNA vaccines and advancing the development of RNA-based medicines (Hou et al., 2021; Eygeris et al., 2022). Despite these advances, efficient delivery of therapeutic cargo to the brain remains a significant challenge due to the restrictive nature of the blood–brain barrier (BBB) (Pardridge, 2012; Saraiva et al., 2016). Conventional experimental screening of brain-targeted LNP formulations is labor-intensive, costly, and unable to efficiently explore the vast design space generated by variations in lipid composition and formulation parameters (Mitchell et al., 2021). Machine learning (ML) has emerged as a promising approach for accelerating LNP discovery by identifying relationships between formulation characteristics and biological performance (Witten et al., 2024). This thesis presents an implementation framework for integrating machine learning into brain-targeted lipid nanoparticle discovery. The published Lipid Optimization with Neural Networks (LiON) workflow developed by Witten et al. (2024) was successfully reproduced, and a curated database of publicly available brain-targeted LNP formulations was developed to support future computational analyses. In addition, representative experimental studies, including LNP physicochemical characterization, encapsulation efficiency assessment, and in vitro and in vivo biological evaluation, were performed to establish an experimental workflow for future validation of machine learning-prioritized formulations. The findings demonstrate the feasibility of integrating computational and experimental approaches into a unified discovery pipeline while highlighting current challenges associated with limited brain-specific datasets, data heterogeneity, and model generalizability. Although detailed prediction results are not presented because they form part of ongoing laboratory research, this work provides a foundation for future machine learning-guided brain-targeted LNP discovery and illustrates how standardized computational workflows can be combined with experimental validation to accelerate the development of next-generation drug delivery systems.
Degree
thesis:*- Name thesis:degree_name
- Master of Science
- Level thesis:degree_level
- masters
- Discipline thesis:degree_discipline
- Biology
- Grantor
- Boston University
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Mursalova, Sevinj
- Advisor dc:contributor.advisor
-
- Lin, Jen-Wei
Subjects
dc:subject × 2Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/2144/53645