Related Experiment Video
Updated: Jan 11, 2026
08:18
Self-Nanoemulsification of Healthy Oils to Enhance the Solubility of Lipophilic Drugs
Published on: July 27, 2022
1.5K
Predicting the critical micelle concentration of binary surfactant mixtures using machine learning.
Aditya Choudhary1, Saaketh Desai2, Methun Kamruzzaman3
1Sandia National Laboratories, Albuquerque, NM, USA. achoudh@sandia.gov.
Journal of Cheminformatics
|November 13, 2025
Summary
A new machine learning model accurately predicts the critical micelle concentration (CMC) of binary surfactant mixtures, including novel combinations. This framework accelerates surfactant design for various industries by reducing experiments and enabling rational optimization.
Area of Science:
- * Physical Chemistry
- * Materials Science
- * Computational Chemistry
Background:
- * Surfactant mixtures are vital in industries like pharmaceuticals, cosmetics, and energy.
- * Predicting their critical micelle concentration (CMC) is crucial for performance but challenging due to complex interactions.
- * Current methods struggle with the chemical diversity and nonlinear behavior of surfactant mixtures.
Purpose of the Study:
- * To develop a novel artificial neural network (ANN)-based machine learning framework for predicting the CMC of binary surfactant mixtures.
- * To enable accurate CMC predictions for both known and entirely new surfactant combinations.
- * To accelerate data-driven surfactant design and formulation optimization.
Main Methods:
- * Utilized cheminformatics-derived molecular descriptors for individual surfactant components.
- * Employed aggregation strategies (concatenation, arithmetic mean, harmonic mean) to combine descriptors.
- * Trained and validated an ANN model, identifying the arithmetic mean strategy with ANN as optimal.
- * Applied SHAP analysis for model interpretability.
Main Results:
- * The ANN model, using the arithmetic mean descriptor aggregation, achieved high accuracy in predicting CMC values.
- * Demonstrated dual capability: precise interpolation within known mixtures and accurate prediction for novel combinations.
- * SHAP analysis confirmed key chemical features (hydrophobic surface area, electronic descriptors, headgroup basicity) driving predictions.
Conclusions:
- * The developed machine learning framework provides a powerful tool for predicting CMC in binary surfactant mixtures.
- * This approach significantly reduces experimental effort and accelerates the rational design of surfactants.
- * Findings support the development of advanced surfactant formulations for diverse applications, including pharmaceuticals, personal care, and enhanced oil recovery.
