The authors present a modular multiscale computational platform intended to support development of mRNA-based therapeutics. The platform integrates two complementary mechanistic models that cover different parts of the pharmacological cascade for mRNA immunotherapies: a systems-level immune response model and a whole‑body pharmacokinetic model. A machine learning component maps IVT‑mRNA nucleotide sequences to kinetic parameters, enabling simulations that are specific to a given product.
This integrated approach is framed as a step toward a quantitative pharmacological framework that explicitly represents the three core components of mRNA therapeutics: the lipid nanoparticle (LNP) carrier, the mRNA payload, and the expressed protein product.
One module is a Quantitative Systems Pharmacology (QSP) model that describes the immunological response to mRNA vaccines. The QSP model spans the cascade from antigen expression within antigen‑presenting cells through downstream processes including B cell activation and the production of circulating antibodies.
The QSP model is positioned to capture dynamic biological processes relevant to vaccine or immunotherapy efficacy, linking intracellular antigen expression to adaptive immune outcomes. In the platform, this model supplies predictions about immune response magnitude and kinetics that can inform design choices.
The second major component is a Physiologically Based Pharmacokinetic (PBPK) model that tracks the whole‑body disposition of antibodies encoded by administered mRNA. This module addresses systemic distribution and clearance of the expressed therapeutic protein, producing tissue and plasma concentration predictions over time.
By coupling PBPK outputs with QSP‑predicted antibody production, the platform can simulate how intracellular expression translates into systemic exposure of the therapeutic protein.
The PBPK model incorporates an explicit molecular layer that resolves mechanistic processes at the cellular and subcellular level. This layer represents LNP uptake, endosomal trafficking and endosomal mRNA escape, and intracellular translation of the mRNA into protein. Including these molecular events permits mechanistic linkage between administered LNP‑formulated mRNA and the rate and amount of protein produced within target cells.
Modeling these molecular steps is intended to capture sources of variability that influence antigen or therapeutic protein expression and thus downstream pharmacodynamics.
A machine learning pipeline is used to map nucleotide sequences of in vitro transcribed (IVT) mRNA directly to kinetic parameters within the mechanistic models. This sequence‑to‑parameter mapping enables the platform to produce product‑specific simulations without requiring full de novo parameterization for each new construct.
The authors present this ML layer as a practical mechanism to translate sequence-level differences into quantitative changes in model parameters that govern uptake, escape, translation, and ultimately antigen or therapeutic protein kinetics.
The platform is proposed as a practical tool to support model‑informed design and development of mRNA therapeutics, expanding quantitative approaches beyond infectious disease vaccines. By integrating QSP and PBPK mechanistic descriptions together with sequence‑informed parameterization, the platform aims to provide a more complete pharmacological framework for the tri‑component nature of mRNA medicines: LNP, mRNA, and expressed protein.
The authors position the platform as a step toward filling a gap they identify: the lack of an integrated quantitative framework that spans molecular, cellular, and whole‑body scales for mRNA therapeutics.
The preprint states that all data produced in the work are contained within the manuscript. The authors report that the study used exclusively publicly available human data obtained from the scientific literature and that original sources are cited in the manuscript. They declare no competing interests.
The authors also confirm adherence to relevant ethical guidelines and indicate that any necessary institutional approvals or trial registrations (where applicable) are addressed; specific IRB or trial identifiers are reported in the manuscript as applicable. The preprint was posted on medRxiv on August 28, 2026, and is available under a CC‑BY‑NC‑ND 4.0 International license.