The Mental Health Hack
Can a startup cure eating disorders?
When Megan Jones Bell, a petite brunette, turned 13, she looked at her gregarious friend at her all-girls school in Palo Alto and wanted to be just like her. She tagged along with her to dance class and took note of how she dressed and talked. When her friend put herself on a diet, Jones Bell did, too, eating lunches of only stone wheat crackers and peanut butter. After watching a Lifetime movie about a young woman’s eating disorder, she learned to count calories and began to do situps in her room every night.
Later, in high school, life at Jones Bell’s home deteriorated. Her mother was battling severe depression, which no one talked about, and her parents divorced. Within her household, Jones Bell’s extreme dieting went unnoticed. When she enrolled at the University of California, San Diego, “I fell off a cliff,” she told me. She came home for spring break gaunt, her straight brown hair falling out, afraid to let her friends see her. Her parents insisted she see a doctor. “He told me, ‘You could die any minute,’” she said.
Jones Bell’s mom withdrew her from college and checked her into an outpatient program at El Camino Hospital in Mountain View. She ate her meals in the hospital and was treated with Cognitive Behavioral Therapy, or CBT. Jones Bell describes CBT as “not a deep dive into your psyche. It’s a tweak to your behaviors.” The idea is that by increasing self-awareness, you notice negative patterns and can make small adjustments — training your mind to let harsh self-judgments pass by like clouds, for instance, or listing ways of feeling powerful rather than seeking control by eating only green foods. CBT, she believes, saved her life.
Fifteen years later, Jones Bell was sitting across from me, radiant from her recent yoga session, wearing jeans and a silky white tank top. Brushing her hair from her freckled cheek, she ordered the German white off the menu at Salt House, a restaurant on the ground floor of her office building in San Francisco’s South of Market neighborhood. Today, she is the chief science officer of Lantern, an online behavioral health company focused on treating eating and mood disorders.
Lantern is one of a growing number of tech startups trying to move into the mental health business. Talkspace offers unlimited texting with a therapist for $25 a week. Doctor on Demand, which launched in 2013 as a network of virtual physicians, has added psychiatrists and psychologists. Ginger.io gathers behavioral data from your iPhone — noting when you stop talking to your friends or don’t leave your house — to predict the onset of anxiety and depression. A number of new companies focus on therapeutic peer-to-peer listening. In Lantern’s case, for $49 a month, an app- and web-based program delivers CBT — in the form of guided meditation, visualization, food logs, and chat check-ins — to patients dealing with anxiety, eating disorders, and depression.
According to the U.S. Department of Health and Human Services, more than half of Americans with some form of mental illness are not getting treatment. Roughly 55 percent of U.S. counties have no practicing psychiatrists, psychologists, or social workers. Suicides in the country have hit a 30-year high. “You have a set of conditions that are tremendously preventable affecting 50 million Americans,” said David Ebersman, the former CFO of Genentech and Facebook. His new company, Lyra Health, matches users with therapists and psychiatrists who can videochat and prescribe medication.
More than 300 mental health startups have launched in the past two years alone, and venture capital investment has grown dramatically. The established mental health community has taken notice, too: Last year’s American Psychiatric Association conference featured a pitch-off for integrating mental health and tech. (The winner was a company called NeuroLex that uses speech-analysis software to detect signs of autism, schizophrenia, and depression.) Although many doubt that breathing exercises and text exchanges with a faceless, part-time counselor can replace visits to an experienced psychiatrist, proponents hope these new companies might democratize mental health care, making it cheap, anonymous, and easy to access. “I want the culture to treat mental health like physical health,” Jones Bell said. “You are not embarrassed to say you are eating a salad, or going on a run or to yoga.” Using Lantern, she hopes, will become as routine as hitting the gym.
The notion that artificial intelligence might offer therapeutic benefits was born more than half a century ago. In the 1960s, an MIT professor named Joseph Weizenbaum developed ELIZA, a computer program that could crudely mimic a human in conversation with another human. ELIZA ran many scripts, but the most popular one emulated a psychotherapist: The program would prompt the user with questions and comments like “Tell me more” and “How does it feel?” ELIZA was convincing enough that some users became attached to it. In 2014, a vastly more sophisticated program, Ellie, interviewed 24 Colorado National Guard servicemen before and after they were deployed to Afghanistan as part of a Defense Advanced Research Projects Agency–funded study to better diagnose post-traumatic stress disorder. Rendered like a video game character, Ellie has a pixie haircut, wears a tan cardigan, and, sitting attentively in a large magenta chair, takes about 1,800 measurements per minute of behavioral cues like postures, facial expressions, fidgets, pauses, and tone shifts. When soldiers talked to Ellie, they revealed more symptoms of PTSD than in their standard post-deployment health surveys. “How do you get honest disclosure from people on topics that are stigmatized?” asked Jonathan Gratch, who co-directs the Computational Emotion Group at the University of Southern California’s Institute for Creative Technologies. “People, when they talk to the character, they feel less fear of negative reaction. They feel less judged, disclose more intimate info.”
Eating disorders, which Lantern hopes to tackle, are notoriously difficult to treat. People are often simultaneously anxious or depressed and in denial. Cognitive Behavioral Therapy, the treatment Jones Bell received at El Camino Hospital, is considered the most effective approach.
After a few months in the outpatient program, Jones Bell was released. When she returned to college, she started delivering eating-disorder-prevention workshops to middle school and high school students. She eventually applied to a graduate program at Stanford in hopes of studying under psychiatry professor Craig Barr Taylor. A veteran in the science of eating disorders, Taylor had developed an eight- to 12-week CBT-based online program called StudentBodies. “He wasn’t taking any more grad students, so I literally stalked him,” Jones Bell said.
StudentBodies was a promising tool. Eating disorders frequently hit young women when they leave home for college. Women with a higher body mass index (BMI) and those who already exhibit unhealthy weight-loss behaviors, like using laxatives to shed pounds, are among the most vulnerable. And it was these groups that StudentBodies helped. In a two-year study published by Taylor in 2006, none of the subjects with higher BMIs developed an eating disorder after using StudentBodies, whereas 12 percent of those in the control group did. In a 2011 study conducted at four German universities, women anxious about their weight who used StudentBodies had a 67 percent reduction in binges and an 86 percent reduction in purges relative to others awaiting the service. Jones Bell was energized: StudentBodies was the very first intervention shown to prevent eating disorders in high-risk groups.
In 2011, Jones Bell asked then-Stanford President John Hennessy if she could develop StudentBodies into a program for college students. Four and a half to 6 percent of the nation suffers from some form of eating disorder. But, in 2009, less than 2 percent of Stanford’s students had sought help from Stanford Health Services for eating-related issues. That would suggest that as many as 700 students with some level of eating disorder weren’t getting care. StudentBodies, she promised, could screen the student body, identify those most at risk, and prevent them from developing an eating disorder or slow the progression of symptoms. And the online CBT approach could be modified to treat other mental illnesses, too.
With Hennessy’s support, Jones Bell received a professorship and one year of funding to work on eating disorders at Stanford. She held regular meetings with coaches, residence-hall advisers, and other campus leaders to discuss how to promote healthy body image and healthy eating. By screening university members online, she identified six times more students with some form of eating disorder than had been identified the previous year. But after two years, she wanted to reach people beyond Stanford’s campus. So she began meeting with entrepreneurs. She eventually connected with two men who had cashed out of Trulia, the real estate site, and were shopping around their idea for a goal-setting app called Thrive On. “Mid-meeting, I counterpitched them,” Jones Bell said. Curing eating disorders, one of the deadliest mental illnesses, struck them as a more worthwhile enterprise than goal setting. And a better business, too. With Jones Bell, they could make a research-driven intervention that no one else could. One hitch: They needed Jones Bell to join them full time. That was the only way funders would back them. Shortly after that meeting, she traded her palm-tree-lined campus for an orange and aqua open-plan office in San Francisco’s South of Market neighborhood. Thrive On became Lantern. (“Thriving was way too cheery. Some of these people can’t leave their houses,” said Jones Bell.) The company now has 40 people on staff, $22 million in VC funding, and a deal with Facebook to offer treatment to its employees.
All of Lantern’s eating disorder programs are iterations of StudentBodies. In addition to meal tracking and chat-based personal coaching, there are mindfulness activities, like a six-minute meditation “reset” for panic, and cognitive restructuring exercises, like one that asks users to look in the mirror naked and identify positive attributes. The programs also learn in rudimentary ways. “If it looks like stress is triggering binge-eating at night, we’ll introduce anxiety techniques,” Jones Bell said. For instance, a user will, at her coach’s discretion, be instructed to lie down and complete a muscle-by-muscle relaxation session.
According to a 2012 review of 103 clinical studies, similar online CBT techniques have helped relieve a number of other disorders, including anxiety, depression, obsessive compulsive disorder, panic disorder, and even irritable bowel syndrome. “This is the beginning of a story which will have a very major effect on how psychological problems are treated,” said Oxford University professor Chris Fairburn, head of the Center for Research on Eating Disorders. Jones Bell put it even more strongly: “We are dismantling psychotherapy to the molecular level.”
For three days, at four-hour intervals, my phone has pinged and told me to “Go Start your Body Image Tracking.” I launch the app and consider how I’d rate my body image over the past few hours: highest, lowest, average. It is 9 a.m., and I’m in my office at San Francisco’s The Grotto, a writers’ collective a few blocks from Lantern. Last night, I polished off a movie-sized box of Milk Duds, but this morning I ran hard at a track, so — middling? I distractedly give myself a 5 out of 10 in all categories.
I’m trying out Lantern’s eating disorder program. I need it. Or something like it. I’ve been fighting against my body since I was 13. My mom was a chronic weight watcher, though mostly the roller-coastery, unsuccessful kind. She would skip lunch, snack while cooking and then eat no dinner, refuse dessert but then lean over and eat my dad’s Häagen-Dazs. She tried liquid diets, was obsessed with aerobics (once when we were snowed in at a ski lodge in Utah, she even led classes). Late in my high school career, she and I signed up for a three-month program at Diet Center, where we went together to be weighed every week, eating our soft-boiled eggs and steamed vegetables with lemon, and carrying Wasa crisps around in little baggies.
It was the summer after my freshman year of college that I started eating as little as I could. I was Eurailing around Italy, and for days I ate nothing but cherries (causing such terrible constipation my boyfriend had to give me an enema). Many nights I couldn’t fall asleep because of my hollow stomach, but eventually the hunger pains subsided, and starving myself all day and eating a small dinner became easy. When I started feeling my hip bones jutting out, I was hooked.
I returned to school to much attention, though it was also the loneliest time I can remember. I was always cold and hunched up and strung out. I was preoccupied with exercising twice a day. When my Levis began falling off, I went to buy new ones, and, looking in the store mirror, swimming in a size 25, I suddenly felt scared. The discipline was hard to sustain, and, periodically, I would binge on my roommates’ food: peanut butter and crackers or Pepperidge Farm Milanos. Right afterward, I would lock myself in a single stall somewhere on campus and throw up.
In my 20s, I stopped the harsh restricting and bingeing. I went on antidepressants and got a job. Still, I cared more about being thin than about having fun, or being polite, or being with people. Restaurants were hard since I couldn’t know how much oil went into the food. I always believed that I would outgrow this, accept my genetics, jettison the self-disgust I felt when I slipped up on my eating regimen. Having kids made this obsession seem more shameful. I’m now a model for three other psyches, including my daughters, ages 11 and 14. How could I not screw them up? They live with a mom who survives mainly on black coffee, nonfat Siggi’s yogurt, salad, and popcorn. I never mention weight, but they see how I handle my own.
I have done lots of therapy, but I still hide many of these feelings: It’s mortifying to admit the extent of the mental trap I am in. So the anonymity of therapy by app is extremely appealing to me. My coach is Salena, whom I encountered first as a tiny round head on my screen. We started with an optional 15-minute introductory phone call, during which I admitted that my 14-year-old already has many rules around eating and that last year she’d tried to get herself under 100 pounds to be faster for soccer. Salena and I have since exchanged 42 texts. A handful of our interactions have felt somewhat personal. But more often her responses come across as generic, automated. I wondered whether she was following a script.
“Hi Diana, I was reviewing your meals for yesterday and it does not look like you are eating much up until dinner. Is that normal for you?”
A few days later, after I revealed a little more, Salena responded:
“That sounds difficult to restrict all day and I am not surprised that you end up eating a lot of candy then in the evening. One of the interventions that we know is most effective is consistent eating. We recommend 3 meals and 2-3 snacks a day. This way you are not leaving yourself starving and more likely to binge. What do you think about that?”
She uses way too many exclamation points:
“These are all great reasons to do this program, and I look forward to witnessing your progress! You asked me to check back in with you tomorrow so that we can set a goal for the week!”
In addition to texting with Salena, I am doing daily exercises to learn how to recognize triggers and prevent myself from falling into a loop of negative thinking. The soothing guided audio tells me that self-esteem isn’t fixed. It changes constantly based on how I value the many dimensions of myself — working-me, wife-me, mom-me, body-me. The woman on the audio asks me to write the “recipe of me,” assigning a percentage of importance to each of these parts, totaling 100. “Try not to judge,” the audio voice says. “We’re just brainstorming here.”
I notice that I’m being more cognizant of going long stretches without eating. The logging, however — of meals and of body image — is painstaking, insufferable. And then Salena suddenly quits on me.
“Hi Diana, I wanted to let you know that I am going to be leaving my coaching position at Lantern to pursue another career opportunity. You will be transferred to Megan, a fantastic coach, tomorrow Thursday 26th!”
As I discover, the qualifications — and reliability — of advice-givers is a vulnerability for many startups like Lantern. The peer-to-peer emotional-support site 7 Cups of Tea advertises that it provides support “in 140 languages, across 180 countries, to over 100,000 people each week.” To join its community of “compassionate trained active listeners,” all you need to do is complete its “active listening course” and check yes to declare “I am not homicidal, suicidal or abusing anyone.” On Ginger.io, a mental health platform, coach Lisa warned me upfront that she was not a licensed therapist but said “we can work together and talk about what’s on your mind.” Talkspace sends the neediest users to a list of helplines. Does this even qualify as mental health care? Clinical psychologist Marlene Maheu, the executive director of the TeleMental Health Institute, which trains practitioners in the safe and ethical use of new technologies to deliver mental health care, says no. “If you think about a licensed professional, they have 1,500 to 2,000 hours of training after you get your degree,” she told me.
Many of the features on these sites and apps can feel like trivial or gimmicky responses to a person who is hurting. Tools in Ginger.io’s Calm Down Kit include a soothing beach scene with wave sounds and cute puppy videos. Free eating disorder app Recovery Record includes Pair Up, a feature in which users can send one another affirmations or baby-animal figures for encouragement. “There’s some research that shows that looking at baby animals increases oxytocin, which stimulates positive emotions,” said founder Jenna Tregarthen. “They feel so much shame, and then this baby hedgehog pops up on their phone!”
Critics of online mental health care warn that these shortcomings could have serious consequences. “People might say they are depressed, but it could be that they are taking three medications and what they are feeling is actually lethargy,” said Maheu. Even video involves a problematic distance, she believes. “I can’t really see in that little screen if their eyes are welling up. That’s the level of sensitivity I’m trained to have.” According to Maheu, it’s up to clinicians to overcome technology’s limitations, to ask the questions that will close the gap between them and their patients. Stanford professor of clinical psychiatry Elias Aboujaoude, who researches the intersection of psychology and technology, warns there is much to work out. “What if you have electrolyte imbalances and need medical attention?” he asked. “How is that going to get addressed online? And people with mild depression can move to severe, suicidal depression. What happens then?”
John Torous, head of the American Psychiatric Association’s Smartphone App Evaluation Task Force, fears people will sign on to these new therapies with their bold promises and feel hopeless if they don’t get better. Many companies post disclaimers, warning they aren’t medical providers, Torous said. “But then they post studies proving CBT’s effectiveness,” he said. “You can’t have it both ways.” A 2015 review of eating disorder apps published in the International Journal of Eating Disorders found that in half of 24 apps surveyed, “the advice was poor and in some instances potentially harmful.” Even Ginger.io’s founder and CEO Anmol Madan acknowledged that right now the industry can seem like “lots of shiny toys and snake oil.” Still, he believes the shift toward online interventions is inevitable, especially given the dearth of care for most Americans. “The reality is we have to start using technology,” he said.
One afternoon, I tagged along with Jones Bell to a meeting with her eating disorder research group in an overflow conference room across the street from Lantern’s office. The team consists of academics from around the world (including StudentBodies creator Craig Barr Taylor) who give Lantern input and partner on studies, like one that’s currently testing online CBT as a treatment for eating disorders on 35 college campuses, including Harvard, the University of Chicago, and Washington University in St. Louis, as well as six state universities in Missouri.
They were gathered to check in on these research programs and to discuss new interventions. Jones Bell jumped to the white board. “What are you most excited about as a stakeholder in this endeavor?” she asked. “Let’s go around the room.” One researcher thought Lantern’s suite of offerings for anxiety and depression was bringing more visibility to eating disorders as a significant mental health issue. Another researcher liked that the therapy was geared to not just college students but also professionals. When it was Taylor’s turn, Jones Bell interrupted. “Do you think we have sufficient evidence on programs for men? That could be a research opportunity for Missouri.” Taylor shook his head. “I can tell you right now we haven’t got the slightest idea what to do with males,” he said. “Are you going to let us do cool micro-randomized controlled trials using Lantern’s data?” he asked — Lantern hopes to individualize its treatment, and Taylor wants to measure how users might respond to slight variations in prompt wording and timing. Jones Bell gave him a toothy grin. “I’ve been thinking about that, too.”
During a break, I asked Jones Bell how Lantern’s coaching works. Are the counselors reading a script? Do they schedule texts? Or do they queue up responses to be sent automatically if the program detects me using certain words or phrases? And what are the counselors’ qualifications? Are they clinically trained or more like data-entry workers following protocols?
“We train them,” Jones Bell said. “We don’t expect they know CBT. They are not doing therapy. They are not teaching CBT,” she added. Lantern, she explained, hires social workers, psychologists, and health coaches and provides them with an initial two weeks of training and more thereafter. Each coach works with about 200 patients. “It’s an unanswered question how much the coaches’ training matters,” said Taylor. I asked him whether tech was going to replace conventional counseling. “We’re moving more in the direction of AI than in the direction of heavy coaching,” he said.
“I don’t think that people today are ready to trust an AI alone,” Jones Bell told me. “You lose so much of the secret sauce of this. A lot of what our coaches are doing is maintaining motivation and empathetic support,” she said. “They are there to help the user feel they are not alone in this.” In other words, it’s not my therapeutic rapport with Salena that determines my progress. It makes no difference if it’s Megan, so long as she checks in every day and holds me accountable for taking another small step. Online therapies don’t need to be able to help everyone; many people just need a nudge, whether toward more constructive patterns of thought and behavior or toward more substantial, in-person therapies.
Before the meeting wrapped up, a few Lantern employees showed off some of the program’s new upgrades, including the self-customization of your calming music and better visuals to guide users through the muscle-relaxation exercise. The company is focused on making the program more intuitive and fun, so that people use it more. “We can get 40 to 50 percent symptom reduction by session 30,” Jones Bell said to the group. “We need to get them there.” Jones Bell won’t tell me current program completion rates. In her 2015 study, online CBT had a notable effect on those who completed the program — but 52 percent didn’t finish. It’s a big but. “It’s like Fitbit,” said Adina Safer, a health care consultant who has worked with numerous digital health companies over the past 20 years. “It’s one thing to sell these tools — it’s another to have them be used. This has to be more than a fun experiment.”
When I met with Jones Bell at Salt House last fall, she described Lantern’s ideal user. “It’s somebody who has decided that they want something to change,” she said. “I think we can enhance someone’s motivation to live differently, but there needs to be enough there already.” As I talked with Jones Bell over the following months, I thought about what that meant. A user the company introduced me to fit this description: Jessie Joachim, a 24-year-old Colorado native who had stumbled on Lantern 18 months ago.
In college, Joachim told me, “I was coming home from class and bingeing and purging for like six straight hours. It was like having a second job.” She had been recruited to Northern Arizona University with a full scholarship for golf, but she couldn’t keep up with both the eating disorder and the rigorous training, so she dropped out of golf. “I was losing muscle, and I quit playing so I could keep losing weight and not deal with the consequences,” she said. Eventually, she left school altogether.
“I finally realized my eating disorder was taking away way more than it was giving me. It was too painful to go on this way.” She ended up trying Lantern, which she stuck with for the full three months of the program. “I found the food logging on Lantern kind of triggering,” she said. “I’d feel ashamed if I was reporting a binge-purge.” But her coach persistently asked questions that pushed her to acknowledge that her problem was serious. “She would ask, ‘What is the eating disorder doing for you?’ or ‘Do you think you should consider a higher level of care?’” Joachim said. The generic responses didn’t bother her; it was the act of writing things out that helped.
Most important, she eventually heeded her coach’s advice. She joined an outpatient program that required a strict eating regimen, and she began psychotherapy. And she has integrated some of the mindfulness tools she learned on the app into her life. When she’s overwhelmed, she’ll take ten minutes to focus on her breathing or she’ll lie on the floor and observe how different parts of her body are feeling. She’s improved a great deal, but is she better, I asked? She was silent for a moment, and I could hear a quiet sigh. Then she said something that probably applies to me and most of Lantern’s users: “I’m definitely a work in progress.”