This year’s TEDMED was held Nov. 18-20 in Palm Springs, Calif. Stanford Medicine is a medical research institution partner of TEDMED, and a group of MD and PhD students who represented Stanford at the conference will be sharing their experiences here.
Picture this: you go to the doctor and find out that your cholesterol is high. Your doctor prescribes you a medication taken by millions of Americans for lowering cholesterol – Pravastatin. A few months later, you see your doctor again because of persistent depression, and again, you are given a commonly prescribed medication – Paxil.
Russ Altman, MD, PhD, opened his 2015 TEDMED talk with this seemingly innocuous scenario. But through the course of his talk, Altman demonstrated how his lab leveraged big data to reveal the adverse side effects of supposedly benign pharmacological interventions.
When choosing medications for my patients during my clinical rotations, I would often cite evidence from randomized controlled trials about the clinical benefits versus the risks of that particular drug. However, this evidence-based medicine has one major limitation: In clinical studies, patients are usually only on one drug.
My patients, on the other hand, would often come in with bags full of prescription bottles in order to show me which drugs they took, since there were too many medication names to memorize. Often, I found myself wondering quietly, “Is there any way to know if combining these drugs could lead to an adverse event?”
As a graduate student in Altman's lab, Nicholas Tatonetti, PhD, sought to address this difficult problem. Using post-market surveillance data on adverse events from the FDA and machine learning, he created a database of predicted adverse events that arose from drug combinations.
Surprisingly, two commonly prescribed drugs, Paxil and Pravastatin, were associated with higher serum glucose, and in turn, could increase the risk of diabetes. Tatonetti, now at Columbia University, validated his findings using data from the electronic medical records of multiple clinical sites. With this data, he demonstrated that patients who were initially on either Paxil or Pravastatin originally, then took the other medication, experienced a clinically significant increase in serum glucose.
As Altman concluded his talk, my classmates and I whispered to one another with excitement and hope. All of us were thinking about our patients, whose long list of drugs had occasionally provoked anxiety about concerns for unforeseen side effects. Now, analytical tools and massive amounts of data were allowing those concerns to be investigated.
I had always felt that I would belong to the generation of doctors whose practice would be revolutionized by big data. And on the TEDMED stage, I had just witnessed firsthand the beginning of that revolution.
Roxana Daneshjou is a MD/Ph.D. student at Stanford University in her last year of clinical training. She is a Paul and Daisy Soros Fellow and she tweets @RoxanaDaneshjou.
Photo by Lawrence Cai