Events

Talk Summary
Observation is the starting point of discovery. Based on observations, scientists form
hypotheses that are then tested and evaluated. In the information-age, trillions of observations
are being made and recorded every day – from online social interactions to emergency room
visits. With so much data readily available, generating hypotheses using a single scientist’s
mind is no longer sufficient. Instead, we must turn to computational algorithms to “mine” for new
hypotheses and relationships. Data mining is an emerging field dedicated to training algorithms
to recognize patterns in enormous sets of data to automatically identify new hypotheses. In this
talk, I will discuss how we use data mining algorithms to identify unexpected effects of drugs
used singly and in combination with other drugs. Drug-drug interactions (DDIs) are an important
and understudied public health concern. Drug-drug interactions are difficult and expensive to
study because of the complex combinatorial nature to their investigation. I developed new
methods for mining clinical data, and then discovered and validated two previously unknown
novel drug-drug interactions. In the first method, published in 2011, I found that paroxetine
(selective serotonin reuptake inhibitor) and pravastatin (HMG-CoA reductase inhibitor) together
cause hyperglycemia. In the second method, published in 2016, I found that ceftriaxone
(cephalosporin antibiotic) and lansoprazole (proton-pump inhibitor) are associated with
prolonged QT syndrome (LQTS). In both cases, I used a combination of data mining and
laboratory experiments to discover and validate the new DDI. These studies are the first to use
big patient data to discover a drug interaction and then use prospective experiments to validate
the findings. Using integrative informatics methods, we are able to discover drug-drug
interactions that no one considered possible before. In many cases these experiments can be
executed in high-throughput and by robotic systems, with the ultimate goal of automating the
scientific method.

Speaker: Dr. Nicholas Tatonetti

Presentation Access

Meeting ID: 503-779-5102