Comparative analysis of methods for predicting the toxicity of chemicals (literature review)

The number of registered chemicals has doubled over the past seven years to 200 million compounds. Currently, the development of alternative research methods is becoming increasingly important. The methods of cross-reading and machine learning are of the greatest interest to researchers.

The purpose of the study is to conduct a comparative analysis of read-across and machine learning methods used in predicting the toxicity of chemicals.

A search was conducted for regulatory legal acts on two information and legal portals – ConsultantPlus and Garant.ru. The search for scientific literature was conducted using the PubMed database, the Cyberleninka scientific electronic library and the eLIBRARY electronic library using keywords such as "read-across", "toxicity prediction", "machine learning", and their analogues in Russian. The reports in Russian and English for the last 25 years have been selected, taking into account the inclusion and exclusion criteria. The conducted review showed the multidirectional application of read-across and machine learning in predicting the toxicity of chemicals. Despite the fact that there is a number of limitations to the use of these methods, a number of studies have demonstrated sufficient reliability and accuracy of their use. The combined use of read-across and machine learning will allow more effective predicting of chemical toxicity.

Conclusion. The conducted review showed the multidirectional application of read-across and machine learning in predicting the toxicity of chemicals. Despite the fact that there is a number of limitations to the use of these methods, a number of studies have demonstrated sufficient reliability and accuracy of their use. The combined use of read-across and machine learning will allow more effective predicting the chemical toxicity.

Contribution:
Guseva E.A. – research concept and design, material collection and data processing, text writing;
Nikolayeva N.I. – editing;
Savranets E.V., Zhantlisova D.M. – material collection and data processing;
Onishchenko G.G. – editing.
All authors are responsible for the integrity of all parts of the manuscript and approval of the manuscript final version.

Conflict of interest. The authors declare no conflict of interest.

Funding. The study had no sponsorship.

Received: January 15, 2025 / Accepted: March 26, 2025 / Published: June 27, 2025

1. CAS REGISTRY. Available at: https://cas.org/cas-data/cas-registry

2. Zulkarnaev T.R., Solominova T.S., Tiurina L.A., Novikov S.M. The prediction of the class of hazard of organic compounds by their structure. Gigiena i Sanitaria (Hygiene and Sanitation, Russian journal). 1999; 78(1): 56–9. https://elibrary.ru/yqscbx (in Russian)

3. Kirlan S.A., Sementeyeva L.Sh., Tyurina L.A. Molecular design and prediction of chemical compounds from the complex of properties activity-toxicity. Gigiena i Sanitaria (Hygiene and Sanitation, Russian journal). 2008; 87(3): 77–80. https://elibrary.ru/jsaomb (in Russian)

4. Pravdin N.S. Methods of Minor Toxicology of Industrial Poisons [Metodika maloi toksikologii promyshlennykh yadov]. Moscow: Medgiz; 1947. (in Russian)

5. Kharchevnikova N.V. A system for predicting the toxicity and hazard of chemical substances, based on the joint use of logistic and numerical methods. Gigiena i Sanitaria (Hygiene and Sanitation, Russian journal). 2005; 84(6): 21. https://elibrary.ru/ojnrlz (in Russian)

6. Zaritskaya E.V., Polozova E.V., Bogacheva A.S. Modern alternative toxicological research methods and prospects of their use in practical activities. Gigiena i Sanitaria (Hygiene and Sanitation, Russian journal). 2017; 96(7): 671–4. https://elibrary.ru/zfbyyf (in Russian)

7. Benfenati E., Chaudhry Q., Gini G., Dorne J.L. Integrating in silico models and read-across methods for predicting toxicity of chemicals: A step-wise strategy. Environ. Int. 2019; 131: 105060. https://doi.org/10.1016/j.envint.2019.105060

8. Novikov S.M. Hygienic regulation for jointly acting harmful factors. Gigiena i Sanitaria (Hygiene and Sanitation, Russian journal). 2020; 99(2): 222–3. https://elibrary.ru/zdjwks (in Russian)

9. Australian Industrial Chemicals Introduction Scheme. Working out your hazards using read-across information. Available at: https://industrialchemicals.gov.au/help-and-guides/working-out-your-hazards-using-read-across-information

10. Patlewicz G., Ball N., Booth E.D., Hulzebos E., Zvinavashe E., Hennes C. Use of category approaches, read-across and (Q)SAR: general considerations. Regul. Toxicol. Pharmacol. 2013; 67(1): 1–12. https://doi.org/10.1016/j.yrtph.2013.06.002

11. van Leeuwen K., Schultz T.W., Henry T., Diderich B., Veith G.D. Using chemical categories to fill data gaps in hazard assessment. SAR QSAR Environ. Res. 2009; 20(3–4): 207–20. https://doi.org/10.1080/10629360902949179

12. Kovarich S., Ceriani L., Fuart Gatnik M., Bassan A., Pavan M. Filling data gaps by read-across: a mini review on its application, developments and challenges. Mol. Inform. 2019; 38(8–9): e1800121. https://doi.org/10.1002/minf.201800121

13. Varsou D.D., Afantitis A., Melagraki G., Sarimveis H. Read-across predictions of nanoparticle hazard endpoints: a mathematical optimization approach. Nanoscale Adv. 2019; 1(9): 3485–98. https://doi.org/10.1039/c9na00242a

14. Ball N., Bartels M., Budinsky R., Klapacz J., Hays S., Kirman C., et al. The challenge of using read-across within the EU REACH regulatory framework; how much uncertainty is too much? Dipropylene glycol methyl ether acetate, an exemplary case study. Regul. Toxicol. Pharmacol. 2014; 68(2): 212–21. https://doi.org/10.1016/j.yrtph.2013.12.007

15. Helman G., Shah I., Williams A.J., Edwards J., Dunne J., Patlewicz G. Generalized Read-Across (GenRA): A workflow implemented into the EPA CompTox Chemicals Dashboard. ALTEX. 2019; 36(3): 462–5. https://doi.org/10.14573/altex.1811292

16. Rorije E., Aldenberg T., Peijnenburg W. Read-across estimates of aquatic toxicity for selected fragrances. Altern. Lab. Anim. 2013; 41(1): 77–90. https://doi.org/10.1177/026119291304100109

17. Berggren E., Amcoff P., Benigni R., Blackburn K., Carney E., Cronin M., et al. Chemical safety assessment using read-across: assessing the use of novel testing methods to strengthen the evidence base for decision making. Environ. Health Perspect. 2015; 123(12): 1232–40. https://doi.org/10.1289/ehp.1409342

18. Schultz T.W., Amcoff P., Berggren E., Gautier F., Klaric M., Knight D.J., et al. A strategy for structuring and reporting a read-across prediction of toxicity. Regul. Toxicol. Pharmacol. 2015; 72(3): 586–601. https://doi.org/10.1016/j.yrtph.2015.05.016

19. Schultz T.W., Cronin M.T.D. Lessons learned from read-across case studies for repeated-dose toxicity. Regul. Toxicol. Pharmacol. 2017; 88: 185–91. https://doi.org/10.1016/j.yrtph.2017.06.011

20. Poroikov V.V., Dmitriev A.V., Druzhilovskiy D.S., Ivanov S.M., Lagunin A.A., Pogodin P.V., et al. In silico estimation of the safety of pharmacologically active substances using machine learning methods: a review. Bezopasnost’ i risk farmakoterapii. 2023; 11(4): 372–89. https://doi.org/10.30895/2312-7821-2023-11-4-372-389 https://elibrary.ru/fovdiu (in Russian)

21. Low Y., Sedykh A., Fourches D., Golbraikh A., Whelan M., Rusyn I., et al. Integrative chemical-biological read-across approach for chemical hazard classification. Chem. Res. Toxicol. 2013; 26(8): 1199–208. https://doi.org/10.1021/tx400110f

22. Erturan A.M., Karaduman G., Durmaz H. Machine learning-based approach for efficient prediction of toxicity of chemical gases using feature selection. J. Hazard. Mater. 2023; 455: 131616. https://doi.org/10.1016/j.jhazmat.2023.131616

23. Hu X.M., Hou Y.Y., Teng X.R., Liu Y., Li Y., Li W., et al. Prediction of cytochrome P450-mediated bioactivation using machine learning models and in vitro validation. Arch. Toxicol. 2024; 98(5): 1457–67. https://doi.org/10.1007/s00204-024-03701-w

24. Lin Z., Chou W.C. Machine learning and artificial intelligence in toxicological sciences. Toxicol. Sci. 2022; 189(1): 7–19. https://doi.org/10.1093/toxsci/kfac075

25. Golbamaki A., Benfenati E., Roncaglioni A. In silico methods for carcinogenicity assessment. Methods Mol. Biol. 2022; 2425: 201–15. https://doi.org/10.1007/978-1-0716-1960-5_9

26. Venkatapathy R., Wang C.Y., Bruce R.M., Moudgal C. Development of quantitative structure-activity relationship (QSAR) models to predict the carcinogenic potency of chemicals I. Alternative toxicity measures as an estimator of carcinogenic potency. Toxicol. Appl. Pharmacol. 2009; 234(2): 209–21. https://doi.org/10.1016/j.taap.2008.09.028

27. Viganò E.L., Ballabio D., Roncaglioni A. Artificial intelligence and machine learning methods to evaluate cardiotoxicity following the adverse outcome pathway frameworks. Toxics. 2024; 12(1): 87. https://doi.org/10.3390/toxics12010087

28. Cheng K., Pan Y., Yuan B. Cytotoxicity prediction of nano metal oxides on different lung cells via Nano-QSAR. Environ. Pollut. 2024; 344: 123405. https://doi.org/10.1016/j.envpol.2024.123405

29. Russo D.P., Strickland J., Karmaus A.L., Wang W., Shende S., Hartung T., et al. Nonanimal models for acute toxicity evaluations: applying data-driven profiling and read-across. Environ. Health Perspect. 2019; 127(4): 47001. https://doi.org/10.1289/EHP3614

30. Kutsarova S., Mehmed A., Cherkezova D., Stoeva S., Georgiev M., Petkov T., et al. Automated read-across workflow for predicting acute oral toxicity: I. The decision scheme in the QSAR toolbox. Regul. Toxicol. Pharmacol. 2021; 125: 105015. https://doi.org/10.1016/j.yrtph.2021.105015

31. Cases M., Briggs K., Steger-Hartmann T., Pognan F., Marc P., Kleinöder T., et al. The eTOX data-sharing project to advance in silico drug-induced toxicity prediction. Int. J. Mol. Sci. 2014; 15(11): 21136–54. https://doi.org/10.3390/ijms151121136

32. Cavasotto C.N., Scardino V. Machine learning toxicity prediction: latest advances by toxicity end point. ACS Omega. 2022; 7(51): 47536–46. https://doi.org/10.1021/acsomega.2c05693

33. Hisaki T., Kaneko M.A.N., Hirota M., Matsuoka M., Kouzuki H. Integration of read-across and artificial neural network-based QSAR models for predicting systemic toxicity: A case study for valproic acid. J. Toxicol. Sci. 2020; 45(2): 95–108. https://doi.org/10.2131/jts.45.95