AdaptDelivery: a polymer informatics platform for predicting structural properties of polymers via machine learning
Abstract
Polymers are highly versatile and cost-effective materials with a wide range of tunable physicochemical properties, making them indispensable across numerous technological and biomedical applications. However, their intrinsic chemical heterogeneity and complex structural organization pose significant challenges for the rational design of advanced systems, particularly in gene and pharmaceutical delivery. To address these limitations, the emerging field of polymer informatics leverages data-driven and computational approaches to enable efficient prediction of material properties and accelerate design processes through machine learning techniques. A data-driven polymer informatics framework is developed by integrating molecular dynamics simulations with machine learning models to enable rapid and accurate prediction of key structural descriptors, including the radius of gyration and solvent-accessible surface area. The models establish quantitative relationships between polymer chain length and the corresponding structural descriptors across a diverse range of polymer architectures. Recognizing that different properties may require different model configurations, property-specific neural network architectures are employed to enhance predictive performance. The resulting models are deployed within an accessible online platform, , designed to streamline polymer design workflows through real-time prediction and visualization. Ongoing developments focus on expanding the chemical space and incorporating additional properties, thereby increasing the platform’s applicability across scientific and technological domains.
Related
- Polymer Informatics Atlas: Property-Specific Representations, Chemistry-Aware Generalization, and Reliability-Aware Machine Learning Across Five Polymer Properties
- A machine learning-based platform for predicting the potential energy of small molecules
- Predicting sensory texture properties of cosmetic emulsions by physical measurements
- Less can be more for predicting properties with large language models.
- Predicting Thermophysical Properties of Halide Salts Using On-the-Fly Machine-Learned Interatomic Potentials Across Multiple Exchange-Correlation Functionals
BibTeX
@article{parajdi2026adaptdelivery,
title = {{AdaptDelivery: a polymer informatics platform for predicting structural properties of polymers via machine learning}},
author = {Lorand Gabriel Parajdi and Istvan Toth and Alex-Adrian Farcas and Zsuzsanna Balint and Alexandru Kiraly and Alexandra Farcas},
journal = {Scientific Reports},
year = {2026},
month = oct,
doi = {10.1038/s41598-026-70802-0},
url = {https://www.nature.com/articles/s41598-026-70802-0},
note = {CC BY 4.0},
}
RIS
TY - JOUR
TI - AdaptDelivery: a polymer informatics platform for predicting structural properties of polymers via machine learning
AU - Lorand Gabriel Parajdi
AU - Istvan Toth
AU - Alex-Adrian Farcas
AU - Zsuzsanna Balint
AU - Alexandru Kiraly
AU - Alexandra Farcas
PY - 2026
DA - 2026/10/04
JO - Scientific Reports
DO - 10.1038/s41598-026-70802-0
UR - https://www.nature.com/articles/s41598-026-70802-0
AB - Polymers are highly versatile and cost-effective materials with a wide range of tunable physicochemical properties, making them indispensable across numerous technological and biomedical applications. However, their intrinsic chemical heterogeneity and complex structural organization pose significant challenges for the rational design of advanced systems, particularly in gene and pharmaceutical delivery. To address these limitations, the emerging field of polymer informatics leverages data-driven and computational approaches to enable efficient prediction of material properties and accelerate design processes through machine learning techniques. A data-driven polymer informatics framework is developed by integrating molecular dynamics simulations with machine learning models to enable rapid and accurate prediction of key structural descriptors, including the radius of gyration and solvent-accessible surface area. The models establish quantitative relationships between polymer chain length and the corresponding structural descriptors across a diverse range of polymer architectures. Recognizing that different properties may require different model configurations, property-specific neural network architectures are employed to enhance predictive performance. The resulting models are deployed within an accessible online platform, , designed to streamline polymer design workflows through real-time prediction and visualization. Ongoing developments focus on expanding the chemical space and incorporating additional properties, thereby increasing the platform’s applicability across scientific and technological domains.
SN - 2045-2322
ER -