Discovering New Drug-Drug Interactions By Text-Mining the ...719 603 603 492 225 206 178 178 178 178 178 170 155 116 115 115 115 115 115 115 Fig. 6. The top 20 most likely interacting

Discovering New Drug-Drug InteractionsBy Text-Mining the Biomedical Literature

Beth PerchaCS 224NFinal ProjectMarch 17, 2011

Background

●Americans are on a lot of drugs.

●Drug interactions are bad.

●We don't test for them in clinical trials.

●We need ways to find likely interactions so we can watch out for them.

Methods (I)

Start with indexed version of Medline:17 million abstracts88 million sentences

Two lexicons:Genes (43 VIP genes)Drugs (602; at least one interaction)

AND

Methods (II)

Results (I)

Results (II)

Most likely interactions... Least likely interactions...