Did you tell your friends on Facebook about the fever you are having or the stomachache you had last night?
If so, researchers suggest that you can be diagnosed months earlier for the health problems you are facing – based on your social media posts.
In a study conducted by Penn Medicine and Stony Brook University, researchers claim that they are able to provide a diagnosis based on the user’s social media posts. They also revealed that the language a user uses on social media could indicate many health conditions including depression, anxiety, and diabetes.
For their research, the researchers analyzed 949,530 posts by 999 participants for 20,248,122 words. Through natural language processing system, they looked for markers of 21 medical conditions and found that most were easy to predict through a combination of demographics and the Facebook language used.
For example, alcohol abuse was marked with the words ‘drink,’ ‘drunk,’ and ‘bottle.’ Alternatively, the users who utilized words like ‘god, ‘family,’ and ‘pray’ were strong indicators of diabetes. Additionally, ‘stomach,' ’head,’ and ’hurt’ were linked with anxiety while ‘pain,’ ‘crying,’ and ‘tears’ were for emotional distress.
Of course, these are just predictions and in many cases, the users mentioning these words were not diagnosed with a particular condition.
This being said, Facebook is already eavesdropping. In fact, the platform admitted to using machine learning to look for words or phrases that may indicate suicidal thoughts or intentions.
But the researchers by Penn are not expanding Facebook and their AI functionality. Instead, it aims to create an opt-in system for patients who agree to have their social media posts analyzed to assist them in refining their medical care.
In fact, the director of Penn Medicine’s Center for Digital Health reports that her team recently worked on a study that would enable a diagnosis of depression up to three months earlier than a clinical diagnosis. She also said that it might be too tough to predict how widespread an opt-in social media post analysis system might be, but it could be very useful for the frequent social media users facing health troubles.
We all know social media posts can be obtained by third-party services. However, efforts are required to ensure users understand how their data is being used. Additionally, they need to understand the extent to which their health is revealed through their online activities.
The main problem is that we, as social media users don’t worry about the little snippets we share on the World Wide Web. We miss the fact that these little chunks of information of our lives can accumulate to a very large picture.
Nevertheless, the data can also lead to superficial assumptions. As the researchers also agreed that simply using a few sets of words or phrases does not make anyone an alcoholic / depressive / diabetic / pregnant.
Let’s wait and watch how the research unfolds into something very useful in the future.
Read next: Reduced Inclination of Users Towards Facebook after Privacy Scandals Surface in a New Report
If so, researchers suggest that you can be diagnosed months earlier for the health problems you are facing – based on your social media posts.
In a study conducted by Penn Medicine and Stony Brook University, researchers claim that they are able to provide a diagnosis based on the user’s social media posts. They also revealed that the language a user uses on social media could indicate many health conditions including depression, anxiety, and diabetes.
For their research, the researchers analyzed 949,530 posts by 999 participants for 20,248,122 words. Through natural language processing system, they looked for markers of 21 medical conditions and found that most were easy to predict through a combination of demographics and the Facebook language used.
For example, alcohol abuse was marked with the words ‘drink,’ ‘drunk,’ and ‘bottle.’ Alternatively, the users who utilized words like ‘god, ‘family,’ and ‘pray’ were strong indicators of diabetes. Additionally, ‘stomach,' ’head,’ and ’hurt’ were linked with anxiety while ‘pain,’ ‘crying,’ and ‘tears’ were for emotional distress.
Of course, these are just predictions and in many cases, the users mentioning these words were not diagnosed with a particular condition.
Is Facebook e-stalking for my doctor?
The researchers claim that social media can provide an effective platform to treat patients who opt for analysis through their social media updates. However, this does not mean the panel of doctor will be eavesdropping on all the posts.This being said, Facebook is already eavesdropping. In fact, the platform admitted to using machine learning to look for words or phrases that may indicate suicidal thoughts or intentions.
But the researchers by Penn are not expanding Facebook and their AI functionality. Instead, it aims to create an opt-in system for patients who agree to have their social media posts analyzed to assist them in refining their medical care.
In fact, the director of Penn Medicine’s Center for Digital Health reports that her team recently worked on a study that would enable a diagnosis of depression up to three months earlier than a clinical diagnosis. She also said that it might be too tough to predict how widespread an opt-in social media post analysis system might be, but it could be very useful for the frequent social media users facing health troubles.
Privacy, consent, and data ownership
If the researchers are able to make a diagnosis simply by judging user’s social media posts, the question regarding privacy, informed consent, and data ownership can also arise in the future.We all know social media posts can be obtained by third-party services. However, efforts are required to ensure users understand how their data is being used. Additionally, they need to understand the extent to which their health is revealed through their online activities.
The main problem is that we, as social media users don’t worry about the little snippets we share on the World Wide Web. We miss the fact that these little chunks of information of our lives can accumulate to a very large picture.
Nevertheless, the data can also lead to superficial assumptions. As the researchers also agreed that simply using a few sets of words or phrases does not make anyone an alcoholic / depressive / diabetic / pregnant.
Let’s wait and watch how the research unfolds into something very useful in the future.
Read next: Reduced Inclination of Users Towards Facebook after Privacy Scandals Surface in a New Report
