Evolutionary optimization of interval rules for drug design

An active area of current research is the discovery of novel drugs for diseases for which no satisfactory drugs have yet been found. To save experimental costs, this knowledge discovery task is assisted greatly by computational experiments because these can reduce the amount of actual experimentation required. This work demonstrates how an adaptive neurofuzzy computation is able to separate molecular data that is bioactive from that which is not bioactive. The resulting classification rules may prove useful in "virtual screening" by means of runtime, effectiveness, and explanatory power. The rules are further improved by using an additional evolutionary strategy to offer a more enriched selection of bioactive molecules.