Logistic regression trees for drug analysis

摘要

An overall system and method for filtering out compounds with potentially poor pharmacokinetic properties, such as unfavorable metabolism or low solubility. The filtering is effected without experimental testing but from consideration as drug candidates or validation or rejection of compounds as drug candidates. This filtering is achieved by comparison of the subject compounds with statistically significant features in a database of already determined acceptable and non-acceptable compounds, by means of a logistic regression tree analysis. A logistic regression tree model and inquiry system is adapted to determined statistically common characteristics (referred to as candidate predictor variables) to structurally compare candidate compounds with the requisite characteristics such as ADME of acceptable compounds.