Developmental Toxicity Risk Assessment: A Rough Sets Approach

Abstract:

A rough-sets approach was applied to a data set consisting of animal study results and other compound characteristics to generate local and global (certain/possible) sets of rules for prediction of developmental toxicity in human subjects. A modified version of the rough-sets approach is proposed to allow the construction of an approximate set of rules to use for prediction in a manner similar to that of discriminant analysis. The modified rough-sets approach is superior in predictability to the original form of rough-sets methodology. In comparison to discriminant analysis, modified rough sets (approximate rules) appear to be better in overall classification, sensitivity, positive and negative predictive values. The findings were supported by applying the modified rough sets and discriminant analysis on a test data set generated from the original data set by using a resampling plan.

Key-Words

Risk Assessment - Rough Sets - Rule Learning - Information Systems

• REFERENCES

• 1 Jelovsek F, Mattison D, Chen J. Prediction of risk for human developmental toxicity: How important are animal studies for hazard identification?. Obstet Gynecol 1989; 74: 624-36.
• 2 Enselein K, Lander T, Strange J. Teratogenesis: A statistical structure-activity model. Teratogenesis Carcinog Mutagen 1983; 03: 289-309.
• 3 Mattison D, Jelovsek F. Pharmacokinetics and expert systems as aids for risk assessment in reproductive toxicology. Environ Health Perspect 1988; 76: 107-19.
• 4 Wasserman P. Neural Computing, Theory and Practice. New York: Van Nostrand Reinhold; 1989
• 5 Pawlak Z. Rough-sets. Int J Inform Comput Sci 1982; 11: 341-56.
• 6 Pawlak Z. Rough classification. Int J Man-Machine Studies 1984; 20: 469-83.
• 7 Hashemi R. An automated rule generator (R²) based on the rough sets. In: Gordon J. ed. The 3rd Oklahoma Symposium on Artificial Intelligence. Tulsa OK; 1989: 89-112.
• 8 Michalski R. A theory and methodology of inductive learning. In: Machine Learning. Michalski R, Carbonell J, Michell T. eds. Palo Alto: Tioga Publ Comp; 1983: 83-134.
• 9 Hashemi R, Razzaghi M, Jelovsek F, Talburt J. Conflict resolution in rule learning from examples. In: Berghel H, et al. eds. Proceedings 1992 ACM/SIGAPP Conference on Applied Computing. Kansas City Mo: ACM Press; 1992: 598-602.
• 10 Schardein J, Schwetz B, Kenel M. Species sensitivities and prediction of teratogenic potential. Environ Health Perspect 1985; 61: 55-67.
• 11 Smith M, Kimmel G, Kochar D. et al. A selection of candidate compounds for in vitro teratogenesis test validation. Teratogenesis Carcinog Mutagen 1983; 03: 461-80.
• 12 Fukunaga K. Introduction to Statistical Pattern Recognition. 2nd ed.. New York: Academic Press Inc; 1990: 51-8.
• 13 Efron B. Bootstrap methods: another look at the jacknife. Ann Statist 1979; 07: 1-26.
• 14 Efron B. The jacknife, the bootstrap and other resampling plans. CMS-NSF Monograph 39. Bristol UK: SIAM; 1982
• 15 Efron B, Tibshirani R. Bootstrap methods for standard errors, confidence intervals, and other measures of statistical accuracy. Statist Sci 1985; 01: 54-77.
• 16 Swanepoel JWH. A review of bootstrap methods. South African Statist J 1990; 24: 1-34.