Network‐based Prediction of Adverse Drug Events and Interactions

when: Thursday, March 20th, 2014 3:00PM-4:00PM
where: room B21 (bât. B, ave Halley)
speaker: Aurel Cami
title: Network‐based Prediction of Adverse Drug Events and Interactions
abstract:
Adverse drug events (ADEs) and drug‐drug interactions (DDIs) are a serious concern for Public Health and the pharmaceutical industry. Every year, millions of patients are affected by drug adverse effects and interactions and billions of dollars are spent to treat the affected patients. Likewise, drug toxicity and adverse effects are lead causes of the high attrition rates and the decreased productivity in the pharmaceutical industry. We have recently developed integrative, network‐based predictive approaches for ADEs and DDIs. Rather than waiting for sufficient post‐marketing evidence to accumulate for a specific ADE or DDI, these predictive approaches rely on leveraging contextual information from previously known drug‐safety relationships, and thus have the potential to predict certain candidate effects and interactions earlier than they can be detected by existing pharmacovigilance methods. We will discuss the major facets of these new approaches, including the data types used to develop the predictive models, feature computation, training, scoring, simulated prospective validation, and the impact of safety data choice on the prediction performance. We will conclude with a discussion of the connections between our work and some recently developed system pharmacology predictive approaches for ADEs.