A Quantitative Read-Across Tool for Advanced Nanomaterials Using Ordered Weighted Average Similarity and Inverse Distance Weighting Interpolation

Assessing the safety of advanced nanomaterials following the Safe and Sustainable by Design (SSbD) framework requires efficient, scientifically robust non-testing approaches to reduce both costs for the companies and the use of experimental animals. While read-across is accepted by regulators and widely encouraged under chemical safety frameworks, applying it to nanoforms is hindered by multidimensional physicochemical profiles where small variations drastically alter toxicological outcomes. To address this challenge, this work introduces a novel data-driven read-across methodology that features an innovative methodological shift: instead of standard Euclidean distance, it computes multidimensional similarity using an Ordered Weighted Average (OWA) metric. The framework explicitly penalizes mismatches in critical descriptors through an asymmetric sigmoid transformation, creating an inherently conservative, regulatory-aligned model. Missing data gaps are then filled using an adapted Inverse Distance Weighting (IDW) interpolation based on Shepard’s method. The distance-decay hyperparameter was systematically optimized, and model robustness was validated utilizing Leave-One-Out Cross-Validation (LOO-CV). The methodology was evaluated using an integrated proof-of-concept case study fusing chemical quantitative structure–activity relationship (QSAR) bioactivity descriptors with experimental advanced material data. Hyperparameter tuning achieved a joint minimization of Root Mean Square Error (RMSE) and maximization of Nash-Sutcliffe Efficiency (NSE). Targeted predictions for molecular initiating event bioactivity profiles accurately replicated the continuous descriptor landscape. The model successfully mapped complex gradients, accurately preserving clusters of high-activity nanoforms at specific size scales as well as lower-activity regions for heavier compositions. Ultimately, this OWA-based IDW framework provides an accurate, transparent, and reproducible tool for predicting missing bioactivity and hazard parameters. By embedding precautionary logic directly into its distance metrics, it fulfils strict regulatory expectations for data gap filling, offering a reliable asset for the safety assessment of complex material forms. This work has been carried out under the POTENTIAL project, Grant Agreement No. 101092901.

Authors

Publication Details

Journal
Zenodo (CERN European Organization for Nuclear Research)
Published
2026-11-12
DOI
https://doi.org/10.5281/zenodo.24157852
Primary Topic
Machine Learning in Materials Science
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A Quantitative Read-Across Tool for Advanced Nanomaterials Using Ordered Weighted Average Similarity and Inverse Distance Weighting Interpolation

Lisa Pizzol, Elisa Giubilato, Alex Zabeo, Matteo Carisi et al.
Zenodo (CERN European Organization for Nuclear Research)
Machine Learning in Materials Science
article

A Quantitative Read-Across Tool for Advanced Nanomaterials Using Ordered Weighted Average Similarity and Inverse Distance Weighting Interpolation

Lisa Pizzol, Elisa Giubilato, Alex Zabeo, Matteo Carisi, Fabio Rosada, Danail Hristozov, Diego Passarella
article en

Abstract

Assessing the safety of advanced nanomaterials following the Safe and Sustainable by Design (SSbD) framework requires efficient, scientifically robust non-testing approaches to reduce both costs for the companies and the use of experimental animals. While read-across is accepted by regulators and widely encouraged under chemical safety frameworks, applying it to nanoforms is hindered by multidimensional physicochemical profiles where small variations drastically alter toxicological outcomes. To address this challenge, this work introduces a novel data-driven read-across methodology that features an innovative methodological shift: instead of standard Euclidean distance, it computes multidimensional similarity using an Ordered Weighted Average (OWA) metric. The framework explicitly penalizes mismatches in critical descriptors through an asymmetric sigmoid transformation, creating an inherently conservative, regulatory-aligned model. Missing data gaps are then filled using an adapted Inverse Distance Weighting (IDW) interpolation based on Shepard’s method. The distance-decay hyperparameter was systematically optimized, and model robustness was validated utilizing Leave-One-Out Cross-Validation (LOO-CV). The methodology was evaluated using an integrated proof-of-concept case study fusing chemical quantitative structure–activity relationship (QSAR) bioactivity descriptors with experimental advanced material data. Hyperparameter tuning achieved a joint minimization of Root Mean Square Error (RMSE) and maximization of Nash-Sutcliffe Efficiency (NSE). Targeted predictions for molecular initiating event bioactivity profiles accurately replicated the continuous descriptor landscape. The model successfully mapped complex gradients, accurately preserving clusters of high-activity nanoforms at specific size scales as well as lower-activity regions for heavier compositions. Ultimately, this OWA-based IDW framework provides an accurate, transparent, and reproducible tool for predicting missing bioactivity and hazard parameters. By embedding precautionary logic directly into its distance metrics, it fulfils strict regulatory expectations for data gap filling, offering a reliable asset for the safety assessment of complex material forms. This work has been carried out under the POTENTIAL project, Grant Agreement No. 101092901.

Zenodo (CERN European Organization for Nuclear Research)
Machine Learning in Materials Science
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon