When drug-company representatives visit Richard Friedman, a professor of clinical psychiatry and the director of the psychopharmacology clinic of the Weill Cornell Medical College in New York, and ask him how he picks the right antidepressant for his patients, “I would take a quarter out of my pocket, flip the coin and say I’d let chance decide because their drug was no better or worse than their competitors,’” he wrote in The New York Times’ “Well” blog.

A common frustration among people with depression, ADHD, and other disorders is that their doctors must rely on a coin flip (or a die-roll, or another chance-metaphor of choice) to choose a medication to start their treatment. For example, a doctor might ask a patient to start taking a common antidepressant for a few weeks and see if it works. If it doesn’t, they start over with a different medication. The fraught process can mean weeks of suffering and false hope for the patient.

“The frustrating part about medication for me had been its trial-and-error style,” one depression patient wrote on Reddit a few months ago. “But there's no other way to do it yet.”

The reason, Friedman told me, is because two people who seem depressed might actually have different illnesses—one might have a problem with her serotonin system, and the other with his glutamate system. There’s no way for doctors to know the true culprit: “We can’t do what we do for someone with a sore throat and culture it,” he said. “And there is a huge downside: It’s called suffering, time, and money.”

The situation is similar for people with anxiety, ADHD, and other mental-health ailments.

Several companies are trying to change that by applying reams of data and, in some cases, genetic testing, to mental-health conditions. The goal is for psychiatrists to be able to test patients and find medications that are more likely to work well right away.

“Psychiatry remains the only discipline of medicine that has no test to predict treatment response,” said Evian Gordon, the founder of one such company, Brain Resource. “This is providing, for the first time, an objective step as to which drug might be responsive.”

These start-ups represent the medical world’s increased interest in so-called “personalized medicine.” Many doctors and health-policy experts are eager to find treatments that will work best for specific patients rather than rely on what’s worked on the average patient in the past. Genetic tests already help guide treatment for some cancers and rarer conditions like cystic fibrosis. The White House recently launched a precision-medicine initiative that will aim to uncover more specific cures, in part through gene testing.

But some experts say that genetic testing for psychiatric conditions still has a long way to go.

A Tennessee-based company called Harmonyx, for example, introduced a test this year that examines mutations on four different genes that, it claims, determine the body’s compatibility with several popular ADHD medications.

When a patient presents a pharmacy with a prescription for a given ADHD drug, the pharmacist could order a Harmonyx $89 cheek-swab test that, in as little as 24 hours, would suggest the patient will respond best to one of several popular ADHD medications, including Adderall, Vyvanse, Ritalin, Strattera, and Dexedrine. (Genetic tests like Harmonyx are already allowed by law).

But specialists I interviewed said that while both Brain Resource and Harmonyx are promising, they don’t necessarily represent an improvement over the existing process for choosing depression and ADHD drugs.

James McGough, a child psychiatrist at UCLA medical center, says that while there is some guesswork involved in choosing the right ADHD medication, doctors are also limited by unrelated factors, like insurance coverage and whether the child is old enough to swallow pills. What’s more, it only takes about two weeks to determine whether a given ADHD drug will work at a given dose—much less time than for antidepressants. “You have to weigh that against the price of the test. Is it really going to make you so much better off?” he said.

According to McGough, it’s a stretch to say that any mental-health condition, or responses to psychiatric medications, are controlled only by genes—especially just four of them. “It’s not like having blue eyes or brown eyes,” he said.  “Psychiatric disorders are likely caused by a whole array of genes in combination with environmental and other factors.”

McGough says he’s seen other companies attempt to storm the area of gene testing for psychiatric conditions, but there is still no scientific evidence to suggest these tests have a role to play in managing ADHD. “There’s no area to break into yet. This is all Star Trek.”

Mark Stein, director of the ADHD and Related Disorders program at Seattle Children's Hospital, agreed, saying that while he liked the idea of a cheap genetic test, an ADHD gene test like Harmonyx still needs many controlled follow-up studies before it’s released to consumers. Harmonyx claims that one of the genes it tests predicts whether the body will struggle to metabolize certain drugs, such as Strattera—and the thinking is that those individuals either shouldn’t take the drug or should take a lower dose. But Stein says that some research suggests that slow metabolizers might actually do better on Strattera.

Harmonyx has also developed genetic tests that aim to show a patient’s suitability for various cholesterol and blood-clotting drugs, and those technologies might be farther along than the ADHD test, Stein told me.

Bob Bean, the CEO of Harmonyx’ parent company, Transnetyx, responded that genetic testing to determine drug response is different from genetic testing to diagnose ADHD, and that the former is supported by research. With Strattera, “the FDA recommends poor metabolizers receive a dose that is less than half that prescribed to those who are extensive metabolizers in order to avoid adverse events,” Bean said, and a test like Harmonyx can uncover that.

Brain Resource, meanwhile, gives depression patients an online test that gauges memory, self-regulation, and emotion. Their responses are then compared against a database that contains the results of 1,000 individuals who took the same test. The thinking is that patients who score similarly will be helped by similar drugs. Currently, Brain Resource evaluates a patient’s likely response to three common depression drugs, which are sold under the brand names Lexapro, Zoloft, and Effexor. The online test takes 40 minutes, and according to Brain Resource’s founders, the results are available within two minutes. (The test is currently awaiting FDA approval.)

Friedman says the initial, 1,000-person study the test is based upon still needs to be replicated. As it is, the algorithm is not necessarily better than chance, he said—and chance is largely how psychiatrists are already determining treatment. “If you do a lot of tests on data, just by chance alone, you’re going to get significant findings,” he told me. “It’s interesting, but it’s not ready for prime time yet.”

Gordon, the company’s founder, counters that Brain Resource is based only on results that were at least 10 percent better than chance. “This increases the likelihood of the result being robust,” he said. “But the ultimate test is indeed replication, and that is currently underway.”

Friedman said that if depression patients want to reduce the amount of conjecture in their treatment process, they can start by keeping careful logs—whether in smartphone apps or on paper—of every medication they’ve already tried.

“I see this all the time, where I’ll ask a patient, ‘what kind of treatment have you had?’ ‘Oh, I can’t really member,’” Friedman said. “Things are tried over and over again.”

For now at least, personalized medicine in mental health might still be just that—up to each person to manage.