Prediction of experimental data for an independent variable using the experimental data collected for other independent variables in a study of skin cancer caused by exposure to UV radiation.

In this study, two algorithms (ONE and TWO) are introduced to determine the position of the t-distribution of variable V(i) (with 95% confidence) in the treated group in reference to the t-distribution of variable V(i) (with 95% confidence) in the control group of an experimental study involving UV radiation exposure of a group of rodents. The outcome of applying the two algorithms is two discretized files. A reduct of each file is generated using the rough sets methodology and then the measurements for one independent variable are predicted using the measurements of the other independent variables in the same reduct. The rough sets methodology and the fuzzy-rough classifier are used for this prediction. The results reveal that (1) algorithm TWO is the best, (2) the values for non-core variables are predicted with minimum accuracy of 87%, and (3) the prediction of values for core variables is not successful.

The prediction of methylmercury elimination half-life in humans using animal data: a neural network/rough sets analysis.

Artificial neural networks and Rough Sets methodology have been utilized to predict human pharmacokinetic elimination half-life data based on animal data training sets. Methylmercury (Hg) pharmacokinetic data was obtained from 37 literature references, which provided data on species, gender, age, weight, route of administration, dose, dose frequency, and elimination half-life based on either whole-body Hg analysis or blood Hg analysis. Data were categorized into various formats for analysis comparisons. Rough Sets methodology was utilized to identify and remove redundant independent variables. Artificial neural networks were used to produce models based on the animal data, which were in turn used to predict and compare to the human elimination half-life values. These neural network predictions were compared to allometric graphical plots of the same data. The best artificial neural network prediction was based on a "thermometer" categorical representation of the data.

Building an organ-specific carcinogenic database for SAR analyses.

FDA reviewers need a means to rapidly predict organ-specific carcinogenicity to aid in evaluating new chemicals submitted for approval. This research addressed the building of a database to use in developing a predictive model for such an application based on structure-activity relationships (SAR). The Internet availability of the Carcinogenic Potency Database (CPDB) provided a solid foundation on which to base such a model. The addition of molecular structures to the CPDB provided the extra ingredient necessary for SAR analyses. However, the CPDB had to be compressed from a multirecord to a single record per chemical database; multiple records representing each gender, species, route of administration, and organ-specific toxicity had to be summarized into a single record for each study. Multiple studies on a single chemical had to be further reduced based on a hierarchical scheme. Structural cleanup involved removal of all chemicals that would impede the accurate generation of SAR type descriptors from commercial software programs; that is, inorganic chemicals, mixtures, and organometallics were removed. Counterions such as Na, K, sulfates, hydrates, and salts were also removed for structural consistency. Structural modification sometimes resulted in duplicate records that also had to be reduced to a single record based on the hierarchical scheme. The modified database containing 999 chemicals was evaluated for liver-specific carcinogenicity using a variety of analysis techniques. These preliminary analyses all yielded approximately the same results with an overall predictability of about 63%, which was comprised of a sensitivity of about 30% and a specificity of about 77%.