Development of a Predictive Classification Model for Surfactant-Induced Skin Irritation

This study investigates the chemical properties of surfactants that significantly influence skin irritability using a predictive classification approach based on multiple linear regression and conditional inference trees. A data set comprising irritation values (Zein number, ZN) for 20 commercial surfactants and their binary mixtures was generated using an in vitro zein test. Key variables (hydrophilic-lipophilic balance (HLB), surfactant concentration, and ionic character) were evaluated to build robust statistical models. The multiple regression model explained 80% of the variability in skin irritation (adjusted <i>R</i> ² = 0.801), while the classification tree achieved an overall accuracy of 72%, with precision and recall values of 0.70 and 0.68, respectively. The results highlight the hierarchical influence of surfactant properties, with HLB emerging as the most significant predictor, followed by concentration and ionic character. Notably, mixtures of anionic and nonionic surfactants showed reduced irritation potential compared to individual anionic surfactants. These findings offer valuable insights for the formulation of safer and more effective surfactant-based products.