Swimming superstar Michael Phelps serves on the board of Medibio, which is developing a quick, cheap, and objective way to diagnose an array of mental illnesses.ROB GRIFFITH/AP
A
small Australian biotech has drawn big-name backers — including swimming superstar Michael Phelps — to its audacious goal: to develop a quick, cheap, and objective way to diagnose an array of mental illnesses.
The tool would be a stunning breakthrough in the field of mental health — if it works.
And there’s the rub. Researchers have been trying for decades to find reliable biomarkers for mental illness — that is, tangible biological clues that conclusively indicate whether a person has a particular psychiatric disease. Effort after effort has failed, leaving doctors to diagnose such conditions mostly on the basis of screening checklists and conversations with their patients.
Now, Medibio — a public company that trades on the Australian stock exchange at less than 30 U.S. cents per share — says it’s developed an algorithm based on hundreds of biomarkers related to a patient’s sleep, heart rate, and other functions that can reliably identify illnesses such as schizophrenia, anxiety, and post-traumatic stress disorder. The tests will sell for less than $20 apiece, Medibio CEO Jack Cosentino told STAT.
Other startups are also rushing into the field, seeking to assess mental health by observing behavior such as how people use their smartphones or how their speech patterns shift over time.
Medibio has been a fairly obscure player in the field, but the company has recently attracted some high-profile board members: Olympian Phelps, who has struggled with substance abuse, ADHD, and depression, signed on in June after the company approached him. He calls Medibio’s approach a “game changer.” Former Rep. Patrick Kennedy, a longtime mental health advocate, joined soon after; he told STAT that he’s especially excited by the prospect of doctors using the algorithms to monitor a patient’s progress over time. (Board members make about $50,000 in U.S. dollars a year.)
READ MORE
5 lessons from my decades of struggle with depression and anxiety
Medibio also recently signed a three-year agreement with the prestigious Mayo Clinic, which will provide clinical experts to review its diagnostic tools. And it’s working with insurers to try to get its test covered. The company plans to apply for regulatory approval in Europe later this year, and to submit a medical device application with the Food and Drug Administration next year. It plans to launch with a diagnostic test for major depressive disorder, then expand to other conditions. A risk of false positives and muddled diagnoses
It just takes a blood test to diagnose diabetes; add in tools like imaging and biopsy, and even wildly complex diseases like cancer can be detected objectively.
But no such test exists for mental illness. In the 1980s, there was a flurry of interest in the dexamethasone suppression test. Researchers believed that a deficiency in the neurotransmitter norepinephrine might increase depression. But testing for levels of an associated hormone didn’t prove a reliable diagnostic.
For its modern-day diagnostic, Medibio said it has identified about 100 sleep biomarkers — based on things like heart rate, circadian cycling, posture — that can be recorded with a device like the Apple Watch, which tracks how a patient falls asleep, transitions through sleep, and wakes up. Yashar Behzadi, chief product officer of Medibio, said the algorithms that link certain patterns to specific mental illnesses are based on data from thousands of people.
It’s a well-known fact that mental illness causes sleep disturbances. Many folks with anxiety, for instance, have difficulty falling asleep because they’re worrying about the events of the next day. So an elevated heart rate before sleep might indicate anxiety, Behzadi said. Similarly, he said that people with schizophrenia lack a regular sleep pattern, with chaotic fits of sleep and wakefulness day and night.
But Medibio has published just one small study to back up its claims. And some experts are skeptical.
“The current evidence base would suggest that a lot of these biomarkers we see in sleep are transdiagnostic,” said Jeffrey Ciesla, an associate professor at Kent State University who studies sleep biomarkers in depression. “That’s a fancy way of saying that these sleep biomarkers aren’t specific: You see similar sleep abnormalities in PTSD, eating disorders, and other things.” Sign up for our Morning Rounds newsletter
Indeed, many mental illnesses have comorbidities: A person with PTSD may also have major depressive disorder or anxiety, so it’s often difficult to differentiate the diagnoses, he said. That might cloud a diagnostic test based on sleep patterns.
Ciesla has also found in his own research that people who aren’t depressed often show sleep abnormalities that are consistent with depression — so there’s risk for a high number of false positives.
Another hitch: disease heterogeneity. Schizophrenia, for instance, isn’t just one disease. It’s actually an umbrella term for dozens of conditions that look similar clinically — but have different biological roots. Mental illnesses are stunningly complex, with symptoms from one disease overlapping easily with another. An intriguing study in patients with PTSD
Medibio published a small but intriguing study this year about the efficacy of its test in diagnosing PTSD.
The company funded a study at Emory University that tracked the heart rates of 23 patients known to have PTSD and 25 control subjects with no history of the disease, over the course of 24 hours. It found that the Medibio algorithm could distinguish patients with PTSD from those without the disease about 80 percent of the time. (One caveat: The study used sophisticated cardiac monitoring equipment, not readily available consumer wearables like the Apple Watch.)
Behzadi, the company’s chief product officer, calls that 80 percent figure “a remarkable number.” Two highly trained psychiatrists might agree with each other on a PTSD diagnosis about 70 percent of the time, he said.
That may be because not all patients choose to reveal their true symptoms in a doctor’s visit, said Rachel Yehuda, director of the traumatic stress studies division at Mount Sinai School of Medicine in New York. If patients actively seek treatment and willingly disclose the relevant information, then the standard screening tests should suffice, she said.
“But when you might need a diagnostic aid is when there are a lot of contradictory symptoms,” said Yehuda, who also studies sleep biomarkers in PTSD, but is not involved with Medibio. “Everyone might agree there’s a problem, but maybe aren’t sure about what the nature of the problem is.”
READ MORE
The ‘Gravity blanket’ raised $3 million online with a claim to treat anxiety. Then that promise was deleted
Of course, some people with PTSD actually have no sleep disturbances whatsoever, so it’d be a shame “if someone with the disorder is told they don’t [have it] because of a sleep algorithm,” she said. But overall, Yehuda is encouraged by the company’s direction, and believes that there’s enough evidence of biomarkers related to sleep and autonomic functions like heart rate to warrant a diagnostic test.
“It’s really a very positive development that people are harnessing the knowledge gathered over the last few decades, and are thinking about easy, practical applications,” she said.
The company also sponsored a small study of just 26 people at Johns Hopkins University last year. It also used the ECG devices, as well as an actigraphy watch, to monitor 15 controls and 11 patients with major depressive disorder. Medibio said its diagnostic test showed an accuracy rate of 81 percent, though the data have not been published in a peer-reviewed journal.
To achieve FDA approval, Medibio must test its diagnostic in at least 50 people, CEO Cosentino said. A hot field for startups
There are, of course, potential privacy concerns — these diagnostics rely on companies tracking and analyzing intimate data about patients. But the appeal of digital tools to diagnose and treat mental illness is growing, with a number of startups and academic researchers jumping into the field.
Take Mindstrong Health, which aims to gauge a person’s mental health by analyzing how he or she uses a smartphone. The artificial intelligence company recently raised $14 million — and attracted some high-profile talent. Dr. Tom Insel, the former director of the National Institute of Mental Heath, left his helmat Google spinout Verily to join Mindstrong just this past May.
READ MORE
Genes help explain why anxiety strikes during adolescence
Another mental health biomarker that could be tracked digitally? Speech. Companies like Atlanta startup NeuroLex Laboratories are developing an artificial intelligence platform that aims to catch verbal patterns associated with different mental illnesses. Machine learning tools pick up on vocal cues like tone, cadence, word choice, and sentence length that can help differentiate a person’s psychiatric diagnosis.
Though the field has stirred excitement, some veteran researchers aren’t holding their breath.
“I’ve grown skeptical over the years about objective diagnostic tests in depression,” Ciesla said. “We’ve been down this road before: This is by far not the first time someone’s attempted an objective test in mental illness.”
