Twitter can also predict heart disease
Official 21st century debate forum, worldwide trend setter, up-to-the-minute information source… Twitter has many uses and they are all diverse, unexpected and surprisingly revealing. Its most recent use? Its capacity to predict the heart disease-related mortality rate of a specific county, discovered by a group of researchers from the University of Pennsylvania, Northwestern University and Melbourne University.
The study published in Psychological Science is focused on the analysis of emotions expressed by nearly 148 million messages published on the social network between 2009 and 2010 in over 1,300 US counties, inhabited by 88 percent of the country’s population. A compilation of feelings in 140 characters with which researchers have been able to draw a map of risks according to the feelings expressed by the different users, proving that those areas characterized by tweets filled with rage, stress or anger actually registered greater Atherosclerotic Heart Disease (AHD), even considering other risk factors (income, education level, etc…), than those counties in which users were more optimistic and happy on Twitter.
In this way, and by preparing a type of “emotional dictionary”, developed as of the compilation of common expressions relating to different frames of mind or words such as “hate” or “wonderful”, the team of researchers was able to measure the state of mind of different communities, establishing a relation between such communities and their heart disease-related mortality rate. “The frame of mind has always been a risk factor for coronary disease”, explained Margaret Kern, one of the researchers participating in the study. “For example, feelings such as hostility or depression are linked to individual-level heart disease through their biological effects. But negative emotions can trigger different social behavior or responses; being more prone to drinking, eating poorly or removing themselves from their surroundings, which can indirectly lead to coronary disease”.
Another conclusion in the study, focused on a more community-level than individual-level analysis, has been verifying –according to the developers—that this model developed solely with a language analysis used on Twitter as the main source, is capable of predicting the AHD mortality level more effectively, for example, than those prepared through a combination of common factors such as socioeconomic or health levels (diabetes, hypertension or obesity). “The relation between language and mortality is particularly surprising”, H. Andrew Schwartz, one of those in charge of carrying out this work confessed to Penn News. “Though people who express feelings considered to be negative are not necessarily those who die from coronary disease, it does show that if many of your neighbors are angry, you are more likely to die from a coronary disease.