As our health needs and systems become more complex, it seems fitting that solutions to our biggest health care questions will be equally sophisticated. Using an immense amount of search data, Columbia University, Stanford, and Microsoft have found unreported side effects for prescription drugs.
Jon Markoff of The New York Times reports that in a study published in Journal of the American Medical Informatics Association, two drugs, namely an antidepressant, paroxetine, and drug for cholesterol control, pravastatin, were shown to cause high blood sugar, which was not previously reported by the FDA.
The implications for the study and its results could be tremendous as researchers expand their analysis to other drugs and even include larger multiples of drug interactions, as Markoff explains in a video interview. The potential to identify adverse reactions to prescriptions in advance of the FDA, often criticized for its review process, is a cause for celebration to many in the health care community. Delving into such research could lead to new guidelines for drug interactions, increased surveillance of approved drugs, and — dare I say it? — a more efficient health care system.
MIT has realized its potential in a program and conference called Hacking Medicine, whose founders are dedicated to teaching skills in order to “launch disruptive health care businesses.” Using big data, the founders hope to accelerate data analysis and health system innovation, increasing cost savings and quality in the process. However, instead of the usual route of employing physicians, scientists, and policy makers in a committee, Hacking Medicine and others in the start up world are involving everyone from entrepreneurs to students to health care professionals in a collaborative, yet competitive, bi-annual contest.
It’s a novel idea in the history of medicine: Everyone has a stake in health, so let’s utilize our collective knowledge to better it.
Western medicine, as a profession, began in guilds in early Greek times and flourished as elite organizations that protected their organization and education through oaths and secrecy. A colleague of mine once said that the difference between a profession and a trade is that a profession retains an air of mystery. If medicine is one of the last professions, it might be because it still has the greatest amount of mystery surrounding it — in its medical decisions, its specialized body of knowledge, and today, its billing practices and the vast amount of interactions and risks that play a part in everyone’s health.
Markoff points out that the problem with small clinical trials is that they often can’t be applied on a population-size scale, which makes extrapolation difficult even for skilled scientists. From time to time, this results in a drug being withdrawn from the market, leading to criticism of the approval process by the general public, and much hand-wringing and apparent flip-flopping on issues from the FDA.
Forget about the government; the new mantra is a health system of the people, by the people, and for the people. But, before you hoist your flag of red and black, it’s good to proceed with caution, as an unbridled faith (as in most things) in the all-knowing power of search engines it seems could just as easily lead to mass hysteria.
Google has data tracking flu-related searches dating back to 2004. And, every year, the flu prediction data has closely mirrored that of the CDC, which collected data by traditional epidemiological methods. The aggregate data is quite impressive, and Google published an article in 2009 reporting its findings. For some time, it seemed people could not stop talking about how great a thing Google’s predictions were.
For the 2012-2013 flu season, David Wagner of The Atlantic Wire reported that Google’s search predictions for this year’s flu doubled the accurate figures of the CDC, casting doubt onto what many hoped would be a revolutionary analysis process for medicine and public health. Unlike data from clinical trials and lab tests, media hype can alter social behavior, leading to a skewing of the results and, apparently, hysterical Googling of flu symptoms.
The FDA is attempting to correct for this flaw in search analysis by financing the Sentinel Initiative, a program that Markoff describes as a way for the FDA to monitor drug use on a massive scale by collecting information from health care providers. While not perfect, it’s apparent that big data and medicine could learn a thing or two from each other, and we will all benefit if they work together.
You can go grab that flag now.