On Additive Artificial Intelligence Discovery of Nanoparticle-Neurodegenerative Disease Drug Delivery Systems

Submitting author affiliation:
University of Basque Country, Leioa, Spain

Beilstein Arch. 2024, 202410. https://doi.org/10.3762/bxiv.2024.10.v1

Published 15 Feb 2024

cc-by Logo
This preprint has not been peer-reviewed. When a peer-reviewed version is available, this information will be updated.


Neurodegenerative diseases are characterized by slowly progressive neuronal death. Conventional treatment strategies often fail due to poor solubility, lower bioavailability, and the inability to effectively cross the Blood–Brain Barrier (BBB). Therefore, the development of new Neurodegenerative Disease Drugs (NDDs) requires immediate attention. Nanoparticle (NP) systems are increasingly of interest for transporting NDDs to the central nervous system. However, discovering effective Nanoparticle Neuronal Disease Drug Delivery Systems (N2D3S) is challenging due to the vast number of NP and NDDS compound combinations, as well as various assays involved. Artificial Intelligence/Machine Learning (AI/ML) algorithms have the potential to accelerate this process by predicting the most promising NDDS and NP candidates for assay. Nevertheless, the relatively limited amount of reported data on N2D3S activity compared to assayed NDDs makes AI/ML analysis challenging. In this work, the IFPTML technique, which combines Information Fusion (IF), Perturbation Theory (PT), and Machine Learning (ML), was employed to address this challenge. Initially, we conducted fusion into a unified dataset comprising 4403 NDDS assays from ChEMBL and 260 cytotoxicity NP assays from journal articles. Through a resampling process, three new working datasets were generated, each containing 500,000 cases. We utilized Linear Discriminant Analysis (LDA) along with Artificial Neural Networks (ANN) algorithms like Multi-Layer Perceptron (MLP) and Deep Learning Networks (DLN) to construct linear and non-linear IFPTML models, respectively. The IFPTML-LDA models exhibited Sensitivity (Sn) and Specificity (Sp) values in the range of 70% to 73% (>375K training cases) and 70% to 80% (>125K validation cases), respectively. Conversely, the IFPTML-MLP and IFPTML-DLN achieved Sn and Sp values in the range of 85% to 86% for both training and validation series. Additionally, IFPTML-ANN models showed an Area Under the Receiver Operating Curve (AUROC) of approximately 0.93 to 0.95. These results indicate that the IFPTML models could serve as valuable tools in the design of drug delivery systems for neurosciences.

Keywords: Neurodegenerative disease; Nanoparticle; Machine Learning; LDA; ANN

Supporting Information

Format: XLSB Size: 34.1 KB   Download

How to Cite

When a peer-reviewed version of this preprint is available, this information will be updated in the information box above. If no peer-reviewed version is available, please cite this preprint using the following information:

He, S.; Segura Abarrategi, J.; Bediaga, H.; Arrasate, S.; González-Díaz, H. Beilstein Arch. 2024, 202410. doi:10.3762/bxiv.2024.10.v1

Download Citation

Citation data can be downloaded as file using the "Download" button or used for copy/paste from the text window below.
Citation data in RIS format can be imported by all major citation management software, including EndNote, ProCite, RefWorks, and Zotero.