Toxishield: Computational Analyser for Food Additives and their Health Implications

Background: The increasing use of food additives in food products highlights the need for scalable computational approaches for population-level health risk assessment. The study presents Toxishield, a proposed algorithm that models a hierarchical pathway linking food products to their constituent additives, additive-level risk rankings, disease associations, and categorical disease risk levels. Methods: Large-scale ingredient datasets were integrated with curated chemical–disease knowledge bases to construct a unified food–additive–disease association matrix, supported by standardized preprocessing to enable systematic additive extraction. Statistical association analysis was employed to identify meaningful computational risk signals, including strong positive additive–disease correlations quantified using the Pearson correlation coefficient (r ≥ 0.7). The Toxishield algorithm was then applied to model hierarchical relationships between food products, additives, and diseases, generating additive risk rankings, disease linkage scores, and category-level risk classifications through recurrence-based aggregation. Results: Based on this analysis, additives and associated disease outcomes were classified into four distinct risk categories: very low risk, low risk, moderate risk, and high risk. Toxishield achieved 98.42% accuracy, precision, and F1-score, with a 1.58% error rate, and consistently outperformed baseline models in mapping food products to additives and associated human health risks. Conclusion: Overall, Toxishield enables scalable and interpretable computational food safety surveillance aligned with Sustainable Development Goal 3: Good Health and Well-Being. Trial registration Not applicable.