Facebook: A new source for medical diagnostics?
A recent study showed that Facebook analysis can reliably predict conditions such as diabetes, pregnancy and mental health conditions.
A new study suggests that words used in Facebook posts could help identify medical conditions such as diabetes and depression.
The study, by the University of Pennsylvania’s school of medicine and Stony Brook University, used an automated data collection technique to analyze Facebook posts of nearly 1,000 patients. The study used AI language processing to analyze 949,530 Facebook status updates containing 20,248,122 words, taken from posts that contained at least 500 words. The patients had agreed to have their electronic medical records linked to their profiles on Facebook.
The researchers looked at the predictive powers of three models: one that analyzed Facebook posts; one that looked at demographics such as age and sex; and one that combined both.
Related: Can Google search history predict ER visits?
From a list of 21 medical conditions, the researchers found that all 21 could be predicted through Facebook language—and that this set of data was more predictive than just using demographic data alone.
Both “Drink” and “God” show links to medical conditions
One of the more remarkable findings of the study was that not only were there some obvious Facebook language links to conditions— “drink” and “bottle” were predictive of alcohol abuse—but that less obvious links were found as well.
For example, people who used religious language such as “God” and “pray” were 15 times more likely to have diabetes than people who used these terms the least. The study also found that hostile language, such as “dumb” and some expletives, were predictive of drug abuse and psychoses.
“Our digital language captures powerful aspects of our lives that are likely quite different from what is captured through traditional medical data,” said the study’s senior author Andrew Schwartz, PhD, a visiting assistant professor at Penn in Computer and Information Science, and an assistant professor of Computer Science at Stony Brook University. “Many studies have now shown a link between language patterns and specific disease, such as language predictive of depression or language that gives insights into whether someone is living with cancer. However, by looking across many medical conditions, we get a view of how conditions relate to each other, which can enable new applications of AI for medicine.”
The study noted that the conditions where the Facebook analysis showed the most predictive ability were diabetes; pregnancy; and mental health conditions such as anxiety, depression, and psychosis.
Questions and opportunities
The study’s authors acknowledged that there are some concerns about the implications of collecting health data from Facebook posts. “Like genomic data banking, the power of social media language to predict diagnoses raises parallel questions about privacy, informed consent, and data ownership,” the study said. “The extra ease with which social media access can be obtained creates extra obligations to ensure that consent for this kind of use is understood and intended.”
In addition, the researchers noted that more study is needed in this area, including increasing the sample size to more accurately reflect the general population (the study used patients from a specific academic health center.)
The study also notes that using social media to predict health status could be a way to predict health issues among people who might slip through traditional screening methods. “In revealing what people think, feel, and do, social media patterns capture emotional, cognitive, behavioral, and environmental markers that have substantial predictive validity and are otherwise fairly elusive to researchers and clinician,” the study notes.
“People’s personality, mental state, and health behaviors are all reflected in their social media and all have tremendous impact on health,” the researchers conclude. “This is the first study to show that language on Facebook can predict diagnoses within people’s health record, revealing new opportunities to personalize care and understand how patients’ ordinary daily lives relate to their health.”
Read more: