Obesity research

We can predict patterns of weight gain and loss on a population scale!  That's exciting new stuff, and it's made possible by a convergence in our understanding of social networks and powerful computer models called cellular automata.  In cooperation with Ray Browning at CSU and Holly Wyatt and Jim Hill at the CU Health Sciences Center, we have constructed a social-network model that predicts weight distributions in the general population.  Simulations with the model have suggested several new interventions that may help dial back the obesity epidemic.


Some of the surprising conclusions

(1) The popular strategy of dieting with friends is not an effective way to lose weight.  Using computer models that simulate the interaction of millions of individuals, our study shows that people inevitably cluster into social groups or social networks of similar weight.  That's cool, though perhaps not entirely unexpected.  So what's really scary?  In a scenario worthy of Michael Corleone in the Godfather III (“Just when I thought I was out... they pull me back in.”), someone can try to leave their overweight social network by dieting or exercise, but the inertia of the group will pull them back in.  Dieting with friends only postpones the inevitable. 

(2) Beside traditional dieting, the study shows that there are many possible alternative solutions to the obesity epidemic.  The key is to use social networks to our advantage.  For example, in computer simulations where a random 1% of the population was maintained at a lower weight (for example, by public health interventions), a domino-like effect caused large segments of the remaining populace to also lose weight.  Combined with carrots and sticks such as end-of-year bonuses for weight loss and health insurance penalties, the weight loss was particularly dramatic.


Want some visuals?

This is an example of a hypothetical social network showing who is friends with whom. An individual's weight (overweight blue, normal weight green) changes with time.

Here's a network with 10,000 people. Color indicates weight (or actually, body mass index, which adjusts for variations in height). Red individuals are targeted for intervention.


Make your own movies of the obesity epidemic!

Check out this QuickTime movie of the clustering in action! Green is normal weight, blue is overweight, and dark blue is obese. Although the weights in this movie start pretty random, everyone quickly adjusts and the weights cluster together. After a while, the obese clusters start to dominate and the normal weight clusters shrink dramatically.

And try playing with the model to make your own movies! The model is user friendly (more or less), runs on any PC, and is free. The technique is still very new, so there is plenty of room for both amateurs and professionals to make new discoveries. An ambitious high school science fair project could simulate trends in obesity and easily discover a new solution to the obesity epidemic. Go for it!


More details and the official publication

So how's the model work?  Recent empirical studies show that you are far more likely to be obese if your friends and family are also obese. Why is that?  It's because you look at your friends, colleagues, and family and use their weight as a gauge for what your own weight should be. This process of looking at neighbors to "update" your own weight is the hallmark of cellular automata, a type of system studied extensively in both computer science and physics.  Our model places millions of people on a network and then lets them interact by looking at their neighbors to gauge what their own weight should be.  The details of the interactions are surprisingly irrelevant.  Any reasonable social interactions result in the clustering of weight described above.

A paper detailing our work has been published by the journal Obesity, Exploiting Social Networks to Mitigate the Obesity Epidemic, 2009.  For the online version, see the January 15, 2009 edition at http://www.nature.com/oby/journal/vaop/ncurrent/index.html. The really curious can also download our PowerPoint presentation from the Obesity Society's October 2008 meeting in Phoenix.



Back to sabbatical page

Back to main page


Please contact me directly with any questions, comments, revisions, corrections, rants, or raves.

Contact:  David Bahr at dbahr at regis.edu


Hit Counter