Smarter than the average pixel?

What I understand of this TEDMED presentation by Dr. Eric Schadt is that it is not adequate to try to understand complex systems using simple linear thought (unless, of course, you are running for high political office). The part I am struggling to understand regards the idea that the causation of disease cannot be determined by studying populations of people and using statistical methods of analysis. Isn't epidemiology based on the notion that aggregate research designs can lead to insights into the causations of diseases in populations? And if an independent variable is important in the aggregate explanation for a disease or condition, isn't it likely to be important in the understanding of specific cases? It might be helpful to me to hear Dr. Schadt engage in a conversation with an epidemiologist about patterns of medical causations in individuals and populations.



It is the metaphor of the movie and the "average pixel" that I have not yet understood. Yes, there is no perfectly "average" patient. A specific instance of a disease or condition may be unique in causal origin. But I want to believe that understanding the health of populations sheds light on understanding the likely causes of instances of diseases/conditions. Patterns in complex systems are usually fractal in nature, meaning that the same patterns are evident at multiple scales. Perhaps I am trying to think too deeply about this or am simply missing some essential insight. Reader, I would welcome your comment that could shed some light.

Big Data and Digital Epidemiology

In the following TEDMED video Nathan Wolfe, director of the Global Viral Forecasting Initiative goes beyond talking about the role of viruses in human history to suggest the implications of connectedness and information exchange.

Supercourse: Epidemiology, the Internet and Global Health

Here is the description of "Supercourse" from the home page at the following URL.

http://www.pitt.edu/~super1/

"Supercourse is a repository of lectures on global health and prevention designed to improve the teaching of prevention. Supercourse has a network of over 56000 scientists in 174 countries who are sharing for free a library of 4855 lectures in 31 languages. The Supercourse has been produced at the WHO Collaborating Center University of Pittsburgh, with core developers Ronald LaPorte, Ph.D., Faina Linkov, Ph.D., Mita Lovalekar, M.D., Ph.D. and Eugene Shubnikov M.D.. Please contact us at super1@pitt.edu"