A study has found that Facebook posts can help detect mental and physical health conditions like depression, anxiety, and diabetes. Researchers say that they can predict whether or not a Facebook user will have at least one of 21 medical conditions, simply by analysing their profiles and posts.
The study was “particularly effective at predicting diabetes and mental health conditions including anxiety, depression and psychoses”.
Researchers at the University of Pennsylvania School of Medicine and Stony Brook University said that they studied 999 patients’ social media presence on Facebook and could deduce whether or not that patient is likely to present with a medical condition.
Lead Author Dr Raina Merchant said, “… insights gleaned from these posts could be used to better inform patients and providers about their health” as what someone posts online is reflective of their state of mind, personality, and lifestyle.
Patients who volunteered for the study allowed automated data collection tools to scrub their profile for information and for their medical records to be linked to Facebook, as well. The researchers then studied the data in three verticals— only Facebook data, only demographics data, and a combination of the two.
How does this study work?
The researchers explain that they obtained Facebook statuses going back five years and restricted themselves to only those subjects who used a total of 500 words in all status updates combined as it is the threshold word count for language analysis. They also obtained demographics such as age, race, sex, and prior diagnoses.
Instead of analysing words that were most commonly used, the study analysed words that were most significantly related to a certain medical condition. For example, “drunk” or “bottle” were linked to alcoholism. These phrases were then grouped together and viewed as word clouds to visually ascertain which ones stood out.
Also read: Second person to be HIV-free spells hope for the global AIDS fight: All you need to know
The study found that more expletive language was linked to substance abuse, while religious language was linked to diabetes.
Dr. Merchant said that healthcare professionals can use information from a patient’s social media profile to better understand their lifestyles and state of mind and more efficiently advise them medically.
However, the study does not directly address the performative aspect of social media. Scores of public figures and social media influencers may not necessarily be honest and transparent in their thoughts and feelings.
People also use social media for a variety of reasons–– while some use it as an outlet or medium of expression, others use it as business platforms or a source of creative discovery. Those who sell wellness products, for instance, might be more likely to use certain terminology related to physical or mental conditions. The study does not clarify whether or not such profiles will be a hinderance to prediction.
Additionally, social media platforms are often used facetiously and people make comments in jest, sarcasm or irony–– the study does not explicitly mention that it takes literary devices or tonality into account.
The study is also limited to Americans and not intersectional.
This is problematic because other cultures may have different norms of social media usage. Although other countries may also use Facebook in English, non-American phrasing carries different, deep-rooted cultural and socio-economic context that must be taken into consideration especially because the study is so heavily based on linguistic analysis.
Moreover, the study also brings up crucial questions about privacy and the dangers of sharing too much identifiable information online.
Google’s AI can accurately predict lung cancer
In May, Google announced that it was developing a new artificial intelligence (AI) tool that can help diagnosis lung cancer more accurately than currently available methods.
“Using advances in 3D volumetric modeling alongside datasets from our partners (including Northwestern University), we’ve made progress in modeling lung cancer prediction as well as laying the groundwork for future clinical testing,” said Google.
Using 3D modelling, researchers are Google use a patient’s current and previous CT scans to form a malignancy prediction. After testing 45,856 chest scans and corroborating its findings with six different US board-certified radiologists, Google says that its AI tool reduces false positives by more than 11% and already is 5% more accurate than radiologists.
The value of technology in all sectors—especially in healthcare—is swiftly increasing. Technological innovation is redefining the speed of healthcare development and vastly improving its quality. The one hurdle now is to make these state-of-the art technological tools affordable and accessible to the masses.
Rhea Arora is a Staff Writer at Qrius
Stay updated with all the insights.
Navigate news, 1 email day.
Subscribe to Qrius