Text Analytics & Drug Discovery — Dr Ian Dix

The discovery of a novel pharmaceutical agent, that is both safe and effective for a given indication, can take over 10 years and in excess of $800m.  A priority for the industry is to reduce these figures without impacting product quality.  Attrition is a key factor, with only a very small fraction of the Lead compounds identified leading to a marketable drug: Decision making at the various milestones in the drug discovery cycle needs to be improved. Apart from having the right experimental data, a key component in decision-making is the prior knowledge or contextual information relevant to the decision.  The quandary we face is that an estimated 80% of this drug-discovery relevant knowledge is locked in unstructured textual documents so how do we access it in a timely and cost effective manner.  Can technologies such as Text Mining and Knowledge Representation deliver the right information, to the right people, at the right time, at the right cost and in the right format to improve decision-making?  This is a complex problem with skill, technology, data ownership, and sociological dimensions. 

Here I will discuss some of the experiences of, and challenges facing, AstraZeneca Discovery in this area.