Mortality research leads to spin-off technology that mines Facebook and Twitter for opinions
A short article on data mining and network analysis:
Verbal autopsy is a technique which assigns a probable cause of death based on interviews with families about the deceased’s prior symptoms, which then is used to estimate the distribution of causes of deaths, known as cause-specific mortality fractions (CSMFs). These are crucial to setting health-system and research priorities, in poorer countries lacking medical certification systems, where more than two-thirds of the world’s population lives.
King was researching a promising approach involving probabilistic models, where instead of trying to assign a single cause of death to each case at all, he instead calculated the probabilities that various disease symptoms are associated with a death, and then aggregated those probabilities across an entire set of cases. “One day I came in, and I realized that the verbal autopsy problem was mathematically the same problem as the presidential one,” recalls King. “I realized I could apply the verbal autopsy method to text, where the symptoms in verbal autopsy are words in text, and death categories are the categories into which you want to put blog posts” to estimate the distribution of blog and Tweet opinions on a given topic.
There is a package developed by the authors free available for R. Further readings on verbal autopsy: