An approach to intelligent drug design support

An exploration-based design model and an integrated artificial intelligence architecture, consisting mainly of an assumption-based truth maintenance system and a blackboard inferencing control system, have been used in a collaborative project called Castlemaine to develop an AI-based support system in the domain of small molecule drug design. The authors present this support system and discuss the AI techniques used in the Castlemaine drug design process model, knowledge representation, and intelligent design support for indirect drug design. On the basis of an initial evaluation of the Castlemaine prototype, some research directions in which the capability of the Castlemaine system can be further enhanced are presented. In particular, the issue of embodying an incremental learning process into the design model and the application of induction techniques to the task of generalization of the pharmacophore model are discussed.<<ETX>>