And the #1 Predictor of Heart Disease Is...
by Matt Lampert
Updated: May 16, 2018
Matt: Matt Lampert here at the Socionomics Institute, great to be with you. Hey, here's a question. If you wanted to forecast levels of heart disease in a community, what sorts of variables would you look at? Well, a team of researchers led by Johannes Eichstaedt at the University of Pennsylvania recently tackled this question, and they looked at some variables that you would probably look at, too. They looked at some demographic factors. They looked at levels of hypertension and obesity, diabetes, smoking. They looked at levels of income and education. And if they combined all these things together, they realized that they had an indicator with a pretty strong correlation to observed levels of community heart disease. But they discovered that there was one other factor that was better than all of these combined, and that was Twitter, the emotional content of tweets. Communities whose members tweeted about things that were generally positive in tone, upbeat, optimistic, those communities tended to have lower levels of heart disease than ones where the members tweeted about things that were more downtrod and pessimistic and depressed. In fact, if you add Twitter to all these other factors, your combined predictability is not that different from what you can do with Twitter alone. Now why is that? Well it turns out that the same social mood that impels people to bid stock prices up and down, also regulates important factors related to our social health and wellbeing. In fact, a research group just posted a paper on the Social Science Research Network, where we dig into these linkages between mood, markets, epidemics, and social health. The upshot of it is, if you have the stock market, that serves as a powerful indicator for the future of aggregate levels of health and wellbeing. It's on the Social Science Research Network. You can read it for free. 48 pages long. Lots of great material to dig into. You can check it out in the link below.
No datasource selected or available.