The Medicine Machine

**This is a summary and response to an article entitled "The Body Synthetic", which is published in this month's edition of Wired Magazine**


Would you trust a medical treatment regimen developed from a computer? David Eddy, renowned physician and mathematician, hopes you would. Over his career, Eddy has had a remarkable impact in the healthcare field, obsessively promoting his "evidence-based medicine" as to what should be the field's standard of practice. In doing so, Eddy has been a game-changer in a field often ruled by tradition, but his medical recommendations have not come without controversy.

In 1980 while working for the American Cancer Society, Eddy crafted new guidelines that recommended women be screened for cervical cancer every three years rather than annually; the switch in policy would save approximately $1 billion in medical costs per year. In the late 1980s, now consulting for Duke University and Blue Cross Blue Shield, Eddy argued against the insurance company from covering a breast cancer treatment that involved high-dose chemotherapy followed by bone marrow transplant. The reason - there was no data that indicated the treatment saved lives. Several years later, after a national firestorm of debate, and even hate mail, clinical trials proved all along that Eddy was right - the treatment had no effect on patient survival.

Fast forward to 2009, where Eddy's obsession and acuity for expansive data trees has translated into Archimedes, Inc., a company formed around a computer clinical trial model system of the same name. Developed and tweaked over the 15 years, Archimedes struck gold when it correctly predicted the results of the Cards trial, a seven year trial that ran from 1997 to 2004 which demonstrated the medical benefits of statin drugs. This favorable result was on top of a remarkable 99% correlation rate after testing 74 previous clinical trials. Eddy boasts of this result, stating that equations "really do seem to represent what Mother Nature is doing."

But despite the seemingly convincing results from Archimedes, some clinicians remain skeptical about the reliability of a mathematical model to correctly emulate the results of a would-be clinical trial. Zeke Emmanuel, health policy adviser for the Obama administration, calls Archimedes "sophisticated but speculative." Additionally, Archimedes wrongly predicted the effects of a new cholesterol drug named Torcetrapib. Eddy doesn't deny this unfavorable result from his program and shifts the focus, saying that Archimedes could positively impact the health industry when it makes wrong predictions by causing physicians to reevaluate the conventional wisdom regarding the involved disease physiologies.

My main concern with using Archimedes or other similar computer model systems is echoed in the article. The problem is several layers deep. The first layer can be inferred from Eddy's own statements. Just because Archimedes can represent what Mother Nature is doing doesn't mean it knows how. Next, the fact that Archimedes needs to be tweaked to represent the latest knowledge of a certain disease or physiological pathway itself isn't problematic. In fact, as with the less technological canons of medicine, this tweaking is a positive action to impart better treatment. The problem in doing this for a complex computer model system with equations linked to each other is that changing one variable can affect a myriad of equations, potentially throwing off estimations for another part of the model.

The last layer is linked to the second. The output from Archimedes is the result of the apex of data crunching, but no one (save for Eddy himself) can follow the highly advanced equations used to derive the results. As the director of the Diabetes Center at Massachusetts General Hospital David Nathan says, "All the calculations happen inside a black box."

From this quote, I am reminded of my high school AP Calculus class. Other than for practice questions from the test itself that required the use of a calculator, our class wasn't allowed to use them. The reason? Our math teacher insisted on us knowing how to do by hand what the calculator - or as he called, "the magic box" - did behind the scenes. That way, we had a better appreciation for what the calculator was doing, plus it allowed us to double check our work. Relating this anecdote to the computer model, how do we know that the calculations being churned by Archimedes are really modeling what happens in our body?

A final note. I am intrigued that the program is called Archimedes. We all know of the famous ancient mathematician, and those who receive emails from me are quite familiar with one of his most famous quotations: "Give me a lever long enough... and I shall move the world." The concept of actually moving the Earth is quite overwhelming, but the math doesn't lie - a long and strong enough lever will indeed be able to move this planet of ours. I include that quotation in my email signature because it reminds me of human innovation and the abilities of science to affect our world; to take the most daunting problems head on and say, "I can solve you." And while I'm not sure if I would subject my health solely to the results of a computer model, I am, too, aware of the power of science impacting our lives. I don't want to be an immovable blockade against medicinal progress, but knowing how difficult it can be to model molecular pathways from my education as a scientist, I just want us to make sure we have every base covered before we start replacing human clinical trials with computer models.

What do you think? Please share your thoughts below.

What You See Is What You Get

So I haven't blogged in ages. I could use the convenient and all-too-common "I'm busy" excuse, but honestly, that's a cop out. I haven't been too busy for wasting time browsing Facebook or Texts From Last Night or other completely mindless things. True, I have been busy (not so much now), but the main reason my posts have ceased up to this point has been based on an internal conflict about the mission of this blog. I enjoy my little bully pulpit to rant about my opinions on science and non-science news items, but the main reason I created this blog was so that my readers - my family, friends, and peers - could connect with a more personal side of me.

"Penn To Paper" is my online diary, if you will. It seems likely that most individuals keeping their innermost thoughts and reflections under the protection of a little padlock and key actually want to share them. I certainly do. Not necessarily because I always have interesting things to say, but because perhaps my readers can individually connect with the things I write.

This connection, however, is diminished when I'm hesitant about how much I give up about myself. I'm not saying that I intend to reveal every single detail about my life, but I have always excessively monitored my behavior and to some extent, my personality depending on whom I'm with at a certain moment. In social psychology terms of Self Monitoring, that makes me a high self-monitor. It's not to say I'm fake, but I like to analyze each social situation and adjust accordingly. And to be honest, being a high self-monitor has some fairly significant drawbacks.

I finally took the time to figure out why I constantly tweak my behavior and personality depending on the situation I'm in. All throughout high school and college, I've felt the need to be liked by all individuals I come into contact with in order to cope with my self-consciousness in several areas. And I'm still doing this in graduate school, but now as I take some time to reflect on it, I have to ask myself why? It's not worth it to change who I am in order to gain some false approval just so I can feel better about myself. I think I'd feel a lot better about myself if I just were the same person all the time.

A big part of doing the above is being more open about myself, so I'm going to mention some things below - some humorous and snarky, some much more serious. In doing so, I hope you can gather a more complete picture of who I am.

-I have extreme Rutgers spirit, to the point where all my Penn friends almost surely find it annoying (and have probably done so for the past month).
-I enjoy watching trashy TV, and I don't plan on stopping anytime soon.
-During elementary school, I can count how many times I had friends over my house on one hand (due to the fact that my parents were divorced and my mom had to work two jobs).
-I tend to rationalize everything (see above item).
-I take all forms of rejection as an opinion of myself, rather than as a reflection of the individual circumstance.
-I have a habit of posting lyrics as my Facebook status; Oftentimes, they are directed towards certain people.
-I feel the only people who really understand me are my mom, dad, and twin brother.

That's a decent sized list detailing some of my thoughts and ideals. Overall, this post has very cathartic and has helped me refocus. Take what you want from it, but I hope this has shown you that there is much more to me than just the energetic redhead you see on a daily basis.