The (Social Media) Doctor Is In: Twitter Can Be Used to Monitor Health

“I hate Donald Trump!” “I’m exhausted and my boss doesn’t care,” “I'm jealous of my sister's new car.”

If this is what your tweets look like, then you might want to reconsider your words: research shows you may be at risk for heart disease. Most people don’t realize it, but the language they use in social media posts can be used to predict their well-being, like their risk for heart disease or other serious conditions. How can social media be used to monitor people’s health and overall well-being?

In one study, researchers evaluated 100 million tweets from 1,300 counties in the United States. The language in these tweets was analyzed and categorized as having either negative or positive sentiment. Then, the authors separated the words of each tweet into word clouds that reflected “risky” language (e.g., despise, hate, jealous, tired) or “protective” language (e.g., opportunity, strength, hope, great).

Using machine learning methods, the authors created algorithms that compared the sentiment of the tweets from each county to CDC data on causes of death. Their findings were dramatic: counties whose tweets expressed more negative emotion (e.g., tweets filled with words such as “hate”) had more heart disease–related deaths compared to counties that featured tweets with more protective language.

When looking further into the data, the methods worked very well at predicting death from hardening of the arteries (atherosclerosis), which is the leading cause of death in the United States. It’s interesting that the people who were tweeting were not the people whose deaths were measured. Instead, the overall tone of the tweets — which were from people too young to be suffering from heart problems — appeared “to have captured a snapshot of the psychology of the community at large.” The authors had therefore discovered a similar result to one that our own team found in an earlier study focusing on HIV.

The authors claim that their methods predicted death from heart disease more accurately than risk factors such as obesity, smoking, and diabetes. Moreover, the prediction accuracy remained strong even after they considered classic predictors of heart disease such as education as poverty. These claims might make you raise your eyebrows, but decades of research has shown that the words people use shows a lot about their psychology. In this case, the algorithms created by the researchers were able to predict personality traits. In fact, the authors claim the algorithms they developed predicted personality traits as well as or better than friends who filled out personality surveys about the participants!

So, how does this research impact your life? For one, an entirely new field of research, known as “digital epidemiology,” has sprung up around social media. Now that social media is widely used (65% of American adults visit social media sites regularly, and 90% of young adults use at least one service), doctors and researchers have an entirely new tool to predict communitywide well-being. Healthcare providers already monitor Google searches to forecast disease outbreaks (e.g., flu, malaria, STDs), which helps determine where resources should be allocated.

For individuals, the hope is that as more and more people use social media, health predictions will become even better. For example, future studies may ask patients to provide access to their social media accounts when they go to an emergency room or visit their primary care doctor. Language in postings and status updates could provide clues to the risk for depression, which is a major risk factor in the recurrence of many diseases. And existing studies have already helped identify post-partum depression among new mothers.

If trends are any indication, social media is going to become even more popular and integrated into tools to solve real-world problems. I’m hopeful that we’ll find ways to use social media to predict other leading causes of death in the near future, and that we’ll find an effective way to protect personal privacy at the same time.

Vaccines and Evidence-Based Medicine: A Q&A with Sean Young, PhD

Researchers have discredited studies that link autism to vaccines, yet it remains a topic of concern for some parents, particularly among the affluent. What’s your take on why this remains an enduring issue in the news?

It’s easy to find associations between things. Sometimes those associations are true, sometimes they aren’t. For example, a classic example is the association between eating ice cream and death by drowning. Someone could look at that link and say that eating ice cream causes drowning, but that would obviously be false. The real reason for the link is that people eat ice cream on hot days, and they also go swimming on hot days. When it’s hot, you find more people eating ice cream and drowning, but it’s not that ice cream causes drowning, it’s that both happen together on hot days. A common expression you’ll hear in cases like this is “correlation does not imply causation.” That is, having two things happen together doesn’t mean one thing necessarily caused the other.

How does this relate to autism? People have seen an increase in autism and they are scared, and they’ve also seen an increase in vaccinations during roughly the same time period in the late 20th century. Some people see these associations and start making claims that vaccines cause autism, but science doesn’t back up those claims. But it’s a compelling argument because it’s built a base among educated, affluent people who are scared their kids will get autism. When people are fearful, it’s hard to use science or facts to convince them that their fears are unfounded – people’s fear, rather than science, takes precedence. Problems that elicit fear and other strong emotions make good news because people will pay attention.

Have you used social media to help people understand the benefits of adhering to a vaccination schedule or general medication schedule?

We haven’t done anything around using social media to change people’s perceptions of the link between vaccines and autism or other health problems. I have, however, talked in a previous post about how data can be used to understand and predict events. Because people readily share their views about vaccines, we could apply similar methods to mine social media data about vaccines and use that to predict whether people support vaccines and how this support would affect vaccination rates and disease outbreaks.

Despite advances in understanding both HIV and the human immune system, a fully successful vaccine to treat HIV is still not available. Do you have hope that one might still be developed, or do you feel other preventative therapies such as PrEP are as far as we’ll go?

I have hope, but I don’t think anyone knows the answer to this question. One current approach is to use genetics/genomics approaches to change genes and target HIV susceptible and infected cells.

Do you agree with doctors who have implemented a policy of only treating children who have been immunized according to the American Academy of Pediatrics schedule? And more generally, what do you see as the core behaviors/belief systems that might cause people to not have their infant vaccinated?

As a scientist, I trust that science is the current best approach, whether it’s science recommended by the American Academy of Pediatrics or other organizations. All that matters to me is whether it’s good science, as opposed to poor science like the correlation relationship I described above. Science might not always prove to be right, as new results emerge and testing methods change, but at any given time I trust that scientific approaches are best. Rather than focus on what causes people to not have their infants vaccinated, I prefer to focus on what is working and leverage that science. We know that social norms have a tremendous affect on what people do, and that extends to vaccines. By creating a social norm that encourages people to vaccinate their kids, such as by using the HOPE social media model, I think we can make major changes in vaccination rates and reduce the spread of diseases caused by lack of vaccination.

From Houston to Hanukkah: The Psychological Benefits of New Experiences

Last week, after finishing a presentation at the National HIV Prevention Conference, I took a cross-country flight from Atlanta to Los Angeles (via Houston). After boarding the plane, I found my seat next to a middle-aged woman. To be courteous, I introduced myself to her. In a distinct Southern drawl, she told me her name was Laura and that she was flying home to Houston to spend Christmas with her family.

I nodded and began to arrange my carry-on items. I started a mental review of what had transpired at the conference: who I’d met, whether my presentation was successful, and what I had to do when I arrived home.

"Do you live in Houston?” Laura asked.

“No,” I said, welcoming the break in silence to learn about her life. I explained that I was returning from a meeting and was anxious to get home after a busy schedule of traveling the past few weeks.

“I understand,” she said. “I’m looking forward to the holidays to relax with my family. I planned to use this flight to do some Christmas shopping. Have you finished your Christmas shopping?”

 “Well, I’m Jewish. We celebrate Hanukkah,” I said. “So luckily, I’m already done with most of my shopping.”

“Oh,” she said. She opened her laptop, paused, and said, “That’s great. I know someone who’s Jewish.”

I laughed. “On behalf of our people, I hope he or she didn’t disappoint you,” I joked.     

Despite our apparent differences, we wound up talking throughout the flight—about her transition from an accountant to an event planner, my work as a behavioral scientist, and about life in Los Angeles vs. Houston. I realized that by the end of the two-hour flight we knew a lot about each other’s lives and beliefs. “You have to see the rodeo in the spring,” she said as we touched down in Houston. Before heading out, she handed me a piece of paper with her email address and phone number. “Come visit during March. My husband and I would love to show you a real Texas rodeo,” she said, with a wink and genuine warmth.

On the second leg of my trip, I thought a lot about Laura. Before meeting her, I would have thought we’d have little to talk about, no common ground. Her views and daily life were way out of alignment with my own, yet getting to know her turned out to be one of the highlights of my short trip.

Research has shown that we prefer to associate with people who think like we do. This tendency, known as confirmation bias, is the behavior of seeking or interpreting ideas in a way that favors personal beliefs. Finding ways to understand confirmation bias is a major feature of the work of Jonathan Haidt, author of The Righteous Mind. Dr. Haidt focuses on the world of politics, but his underlying theme is that relationships shouldn’t simply be about trying to sway or inform people. Rather, every relationship offers the opportunity to learn a new perspective—something I always try to keep in mind.

Being open to a conversation with a stranger on a plane (or on your local street corner) won’t cure the world’s ills, but it’s a start at uniting people from different backgrounds and cultures, and it might lead to a new friendship—or even the opportunity to attend a rodeo.

Sean Young, PhD, Reports Back from CDC HIV Conference on New Year's Resolutions

1. As a psychologist and researcher, what are the hardest New Years Resolutions for you to stick to?  Do you find that professional goals are easier to achieve than personal goals?

I wouldn't say the difficulty is broken down by professional goals vs personal ones. The great thing about psychology is that it doesn't care about domain. It doesn't care whether people are making resolutions to change business goals, health goals, relationship goals, or any other goals. What matters is the context of those goals, within the person, their surrounding, and their experience. That's a bit vague so I'll be more concrete.

The hardest New Years Resolutions to stick to are ones that require the biggest change in lifestyle to keep. If something is really tough to change, it will be tough to keep. If it's easy to change, it will be easier to keep. Take dieting as a resolution. My undergrad professor, Traci Mann, has done a lot of research and shown that diets don't work. Aside from biological reasons in people's genetics, most diets fail because they're diets, or big immediate changes in people's behaviors. They have been eating unhealthy food for a long time and decide that because it's the New Year, they'll have the ability to instantly change the way they eat. Most New Years Resolutions fail for the same reason. People want to instantly change something that has been part of their lifestyle for weeks, months, or years.

So the bad news is, New Years Resolutions need to be kept in perspective with how people have been living. If a person walks 50 steps a day, making a resolution to walk 10,000 steps a day won't last. The good news is, that there are ways to keep resolutions. People just need to keep them in perspective and make resolutions that are manageable. There are a lot of other ways to help keep on track based on our research. Some of these I mentioned in last week's Q and A, like the science of social. 

2. After attending the CDC's recent HIV prevention conference in Atlanta, do you find yourself shifting your own priorities to align with the research community as a whole?

I've realized I've been a researcher my whole life. It started long before my research assistant days at UCLA or doctoral work at Stanford. It started as a child as I loved learning about things and how they work. One of the most important things that I keep learning is that I need to always keep an open mind. I need to always listen to other people's ideas and perspective, no matter how crazy people might think they are, because I learn from them and it helps to guide my research. That's a broad answer to your question. The straight answer is, definitely. I'm constantly rethinking studies, research, and my own assumptions based on what I learn from the research community as well as everyone else. I learned a lot about people's perceptions of PREP at the HIV prevention conference in Atlanta and have been thinking about how technologies can be incorporated into Prep education and behavior change.

3. What are the main takeaways that you got from the CDC conference?  Where will HIV prevention be at this point in 2016?

The main takeaways is that although there is still a lot of work to do to reduce the spread of HIV, we've been getting some answers. Really importantly, we've been having support for controversial approaches from top officials, like the NIH director support use of Prep. For me, as a technology researcher wanting to find ways to predict, prevent, and change HIV risk, the main takeaways is that there is so much opportunity for tools to be used in this space. Researchers are very open to these tools but don't have the time to be aware of them. Because innovation and tech tools seem to always be at the forefront of how HIV is spread, we need to use that innovation to prevent and stop the spread of HIV. I'm excited that our team has the ability to do that and we're getting a great response from people all over the world who want to work with us and apply our research.

4. What steps can clinicians, families and societies take to remove the stigma from both HIV prevention pills and HIV testing?

Stigma is the belief that a person or thing is unwanted, disgraced, or or shameful. It can lead to a lot of negative consequences. When people are stigmatized they feel badly about themselves, they can lose their friends and family, their jobs, and can have worse health. When things are stigmatized, like getting an HIV test, it makes people to not want to do them. We've done a lot of studies on how stigma works and how it stops people from taking care of their health. (One of those studies you might like involved telling students they were at risk for a disease, and learning they they more or less convinced themselves they couldn't have contracted a disease if it was stigmatized).

Stigma is caused by lack of knowledge, lack of discussion, and lack of normalcy. The way to reduce or get rid of stigma is to educate people, make them aware of how stigma works, and make them see the stigmatized person or thing is common. For example, HIV testing is stigmatized. Just showing up to an HIV testing site could make people stigmatized. They could be judged by others in the clinic, by their doctors, by people seeing them getting tested. They could be judged as being "the type of people" who have HIV. To reduce this stigma, we can do things like talking about testing more, getting people to test more, and making testing more public so that people can see how many people test for HIV and that people from all ages, races/ethnic groups, and educational statuses test for HIV. That's great that so many people test, and it needs to be made more public. We found that stigma could be reduced by making the stigmatized thing (for example, testing) required. We also found that offering it in traditional settings like in vending machines may reduce stigma and get more people to test for HIV. 

5. What steps can Grindr, Tinder, Match.com and other online dating websites take to help prevent HIV?

These are dating/hook-up businesses and so they're less interested in preventing HIV than in their business, so I wouldn't expect them to make any major changes to help prevent HIV. Some of them are concerned about losing users if they try to promote HIV testing as they don't want to be branded as a public health site or place that is trying to get people to do anything other than find dating or hook-up partners. That being said, there are a few things they can do that could help prevent the spread of HIV and shouldn't negatively impact their business. First, they can be open to HIV researchers. Second, they can offer a feature that allows people to say if they have gotten an HIV test. Third, these sites and researchers can begin sharing data with each other to mutually find how to make their users safer and healthier.

FAQ with Sean Young, PhD

1. What was the "aha" moment where you decided that behavior prediction technology was the path you would dedicate yourself to?  Was there something you read or witnessed or experienced?

To clarify — I’m answering your question about my interest in behavior “prediction” technology, but my general interest in psychology and technologies came before this when I was at Stanford and working at NASA)

It was actually a multi-year 'aha' moment. I think all of my ‘aha’ moments come from my interest in doing things that most people think is crazy, weird, or not possible. If my pursuit leads to something valuable, it sparks an ‘aha’ moment. For example, my friend Gopal was working at Cisco and pulled me to work with him on a project called "mind map, or using technologies to map how humans think.

It was an ambitious project and so we started small looking at whether we could use data to predict anything about people. This was back in 2006 or 2007 when Facebook was growing and we could use Facebook book data for the study. I started keeping spreadsheets of my friends on Facebook to learn when they changed their profiles and explore why. I would document, by hand, when they changed anything, like when they added where they were going to school, removed an interest in football from their profile and replaced it by saying were interested in baseball instead, or changed their relationship status from single to "it's complicated." I'd call up my friends and ask them why they changed these  things. People had mixed reactions when I asked them. Most of my close friends understood me and chalked it up to me being inquisitive and weird. Some of my friends thought it was funny that I was doing this, and others thought it was creepy. (Little did they know that within a few years this would be commonplace-- there would be companies creating robots to monitor almost every action they took in life. At the time, behavioral targeting (or the ads you see targeted towards you when you’re online were terrible—guys with lives that revolved around playing fantasy sports would be shown ads for getting maternal eggs harvested. Ads are now a lot more accurate because they use these types of data to understand people) From this work, we ultimately published a pretty interesting paper on how to read behind the lines of what it really means when people change their relationship status. This was one of the first papers I know using social data to predict things.

At the same time, I was studying how to use technologies to prevent the spread of HIV. We were building online communities based on behavior change science, through an approach we called Harnessing Online Peer Education (HOPE). HOPE had people at risk for HIV join our online communities. (Since those initial studies we found that the HOPE approach has been pretty successful. In multiple studies in different groups across the world, we’ve seen HOPE get people engaged in long-term behavior change that sticks.) But we found something interesting happening. People were invited to private HOPE online community groups but immediately were sharing really personal information to strangers in the group, like what is was like being a married man who has sex with other men without telling his wife, or types of illegal drugs people were using. If people were instantly sharing such personal information with strangers through our HOPE communities, then perhaps they were also doing it on sites that were more public, like Twitter. The advantage to searching here was that Twitter provides access to a ton of their data, so we would be able to study “big data.” As I dug into the data, I was shocked at what people shared. There were a lot of people who had thought that social media was a pointless fad, a way for people and businesses to self-promote, and that it had no potential for social good. As I looked through the data, I thought I was finding a gold mine for advancing research in a ton of areas. Not to mention I was really excited because I wouldn’t be able to do 1/10th of the work on my own. I’d need to rejoin with my engineering friends I missed from graduate school and get their help. And I’d be able to learn a lot along the way about how people in different fields do research.

2. I can see from your research work on HIV that Twitter has played a critical role in your ability to predict the spread of disease.  Can this be applied to influenza and less extreme cases, or is there something specific about HIV on social media that made those results possible?

Twitter and other social media can definitely be applied to influenza and other areas. In fact, it’s easier to apply it to those areas. As long as people feel comfortable and free using social technologies, then theoretically we should be able to use data from these technologies to predict almost anything. The limiting factor is whether we have what is called “gold standard data,” or data on actual events occurring (like cases where people have contracted HIV or influenza), as well as the frequency of getting those data. For example, influenza data are provided frequently, every week I think. That makes it much easier to create models to predict influenza compared to something like HIV, where national data on HIV cases are released about 2 years after they occur. That means that there’s huge value if we can predict HIV. It’s 2015 right now and we won’t know until well into 2016 or 2017 how many HIV cases occurred. If we can use social data to predict that number a little earlier it could have a huge impact on people’s health, it could reduce disease, save money, and prolong people’s lives.

3. Is there some fundamental truth about social media that lends itself to public health?  For example, 30 years ago, if I dedicated myself to hooking up with an anonymous stranger for unprotected sex, I might scribble my phone number on a bathroom stall at a rest stop gas station.  If Sean Young, PhD existed as an HIV prevention advocate in the early 1980's, would you be calling up gas station owners for frequent reports on the latest bathroom stall graffiti?

That’s an interesting question. I wouldn’t say there is a fundamental truth about social media or social data. As a psychologist, I would say that everything that people do leaves a trace of their psychology. The trick is knowing how to interpret what they do and read between the lines to know what it means. Not everyone would write on a bathroom wall, and of the people that would, they would only do it in certain times and contexts. (I’ll save people from my bathroom humor by not describing when some of those times would be). Social media is no different. There isn’t anything unique about us being able to get data on people’s behaviors from social media that we couldn’t get from other places. The point is that there is sooooo much data—for example 500 million tweets a day --- that it’s easy to analyze the data. Researchers want as much data as possible to confirm their hypotheses. Because there’s so much social data it gives us the ability to test and refine hypotheses about why people do things and how we can use this information to predict and solve real-world problems.

4. If you could have a face to face with the CTO of Twitter and request some changes to their technology that would improve health outcomes and prevent disasters, what would they be?

Ha ha. You’re touching on some things I used to think about from a social entrepreneurial perspective back in the day, like building a Twitter designed to get people working together to solve important global issues. Twitter’s most recent earnings call shows they’re having some major business problems with an uncertain future so I’d let Jack off the hook and let him figure out what to do with Twitter first. After that, I’d ask him whether he’d be willing to help with an effort to bring together companies with large amounts of data like Facebook, Twitter, Snapchat, and Google, to have them provide access to public health researchers. Then he’d probably laugh and never talk to me again. 

5. What clues do human beings leave on social media that they are about to hurt themselves or others?  Are there some big flashing signs that society is missing and should be educated about?

There’s no quick answer for this. Psychologists know you have to observe people over time to understand their patterns. Clinicians can look for certain signs like people not making consistent eye contact to know something might be wrong. There are ways of translating this to social media. But the clearest example would be the college student in October 2014 who talked on Twitter about his fear and disgust about his own and other people’s lives. A month later he killed 3 students and then took his own life.

6. Looking at what's trending on social media now, including my own, is there a glaring prediction for some coming cataclysmic event that you can share with me?

A big drop in the stock market.

7. Has there ever been a moment, working with predictive technology, that you have felt compelled to have someone committed to psychiatric care or call the police to warn of a crime?

No. So far we’ve been working at a population level, or looking at large groups of people to learn about their patterns. More recently, we’re studying individuals and so it’s likely I’d get those moments in the future. There’s still an interesting ethical question about whether and when people should act on those insights and actually intervene.

8. You work in the Family Medicine Department at the David Geffen School of Medicine at UCLA.  How can physicians and therapists use social media to gain greater insight into their patients?  Do you foresee Twitter feeds becoming a diagnostic data type formally integrated into electronic health records?

While these questions are focused primarily on Twitter, there’s a lot more than Twitter that can be used to provide predictive information to help in areas like public health and medicine. Other forms of social media, like Instagram, as well as wearable device data and search data all provides insights about behavior too. The department and the health system as a whole is already interested in applying some of our research. For example, the health system has asked me to be involved in integrating social data with medical records data to learn how we can have a more complete picture of patient health. Also, we’ve built a technology platform used to change patient behavior that is being tested on UCLA patients to improve their health. We’ll be able to use the data from studies and technologies like that to better understand patient needs, increase patient engagement/retention, and improve delivery of care. 

9. From a theoretical perspective, is there anything in the work of Sigmund Freud, Alfred Adler or the other pioneers of psychology that prepared you for your research?  Is there anything that you find yourself returning to philosophically?

I remember an early psychology course I took in grad school at Stanford. The professor began the class by saying, this is a psychology class and so many of you might think of Freud. Almost everything that we’ll study has already been mentioned in some way by Freud. But there’s no basis for most of ideas. He wasn’t a scientist. That’s why there are barely any psychologists who still practice Freudian psychoanalysis or teach his theories. We use science to study our ideas and so this will be the last mention you’ll hear of Freud in this class.

As a social and behavioral psychologist, most of my theories come from classic psychologists like Kurt Lewin, Leon Festinger, Stanley Milgram, and Daniel Kahneman/Amos Tversky; my graduate school mentors like Lee Ross, Benoit Monin, Albert Bandura, Claude Steele, and Bob Zajonc; and friends and colleagues like Danny Oppenheimer, Dave Nussbaum, Jonah Berger, Chris Bryan, and Hal Hirschfield. 

10. Are you planning future studies that incorporate Snapchat, Instagram, Whatsapp or other social media technologies?

We’re in conversations like that right now and so I’m not sure what the companies would allow me to divulge at this point, but the short answer is yes.