Computational Psychiatry Is Changing How Physicians Predict and Treat Bipolar Disorder

The most powerful tool for preventing a depressive or manic episode in bipolar disorder might not be a new drug. It might be a wearable device.

Bipolar disorder affects millions of people and remains one of the most diagnostically complex and undertreated conditions in psychiatry. Unlike normal mood changes, which we all have, mood changes in patients diagnosed with bipolar disorder persist for weeks or months.

Thanks to wearable sensors, combined with nonlinear modeling techniques adapted from physics, patterns are emerging to reveal measurable changes in sleep and activity that precede a mood episode – sometimes by more than a week.

Abby Ortiz, M.D., an Associate Professor of Psychiatry, leads the Computational Psychiatry Program at UT Southwestern, which combines wearable technology and computational modeling to identify early warning signs of mood episodes.

Looking at the Movie, Not the Picture

Distinguishing bipolar disorder from related conditions takes more than a snapshot of a single visit, Dr. Ortiz said. She asks about family history and the fluctuations that surround a patient’s cycles, tracks daily changes in energy and anxiety levels for at least four weeks, and invites family members into the conversation. Layering in the patterns and insights from wearable devices reveals the whole story.

“I look at the movie instead of the picture,” she said. “It’s difficult to predict how someone will respond to treatment without understanding the full pattern, not just the moment they walk into my office.”

That philosophy shaped her research, too. Dr. Ortiz built her program around a landmark longitudinal study she led in Canada, one of the largest digital phenotyping studies of mood disorders to date. Over two years, participants wore biometric devices and completed daily mood questionnaires, generating more than seven million data points, a foundation she brought with her to UT Southwestern.

The Early Warning Window

Dr. Ortiz has always thought episodes do not arise overnight. She conceptualizes episodes as a “cascade of symptoms.” Her team recently published two papers examining what changes in the days and weeks before a mood episode. Depressive episodes, they found, are preceded by measurable declines in activity, with patients moving and walking less, compared to their normal routine, up to seven days before they report feeling low. Hypomanic and manic episodes move even faster. The earliest signal, a shortening of sleep, is detectable two to three days before patients report they feel keyed up.

“For someone with depression, a week is enough time for me to intervene,” Dr. Ortiz said. “For mania, the window is shorter, but if I can see a change in sleep three days out, I can make a medication change before things escalate.”

Validating Before Alerting

Rather than rushing to deploy real-time alerts, Dr. Ortiz is taking a deliberate approach, validating her algorithms across multiple wearable platforms, including Fitbit, Apple, and Oura, before building a system that would notify patients or clinicians directly.

“I don’t want to scare the patient, and I don’t want to scare the clinician,” she said. “I want really good validity first. A lot of groups are doing wearable research, but you cannot make a prediction based on 20 participants followed for a week. We need the method to be right before we send an alert.”

What Patients Are Learning About Themselves

Across the two-year study, participants maintained a 78% compliance rate, completing daily mood questionnaires and wearing their devices consistently. Counter to expectations, the patients who had been hospitalized or had a history of suicide attempts were among the most engaged, not the least.

For some patients, the data itself prompted positive change. One younger participant noticed his mood and anxiety scores worsened every time he drank and cut back on alcohol after seeing the pattern. Another, a high-functioning executive, realized through her data that her depressive episodes clustered on weekends, when isolation replaced the structure and social contact of her workweek. Once she recognized the pattern, she began scheduling walks and time with friends to close the gap.

“It’s like having a bridge between what the doctor tells you is good for your health and what it actually feels like to do it in real time,” Dr. Ortiz said. “Seeing and understanding your own data (e.g., how your mood changes with activity; or how it is affected by sleep) helps close that gap.”

When It Is Not Bipolar

Dr. Ortiz also maintains an active clinical practice in complex mood disorders, where many referrals involve patients labeled treatment resistant after cycling through multiple medications without improvement. In her experience, the diagnosis is rarely wrong, just incomplete. She has traced apparent hypomania to causes as varied as undiagnosed substance use, including a patient consuming 10 to 12 cups of coffee a day, and underlying autoimmune or liver disease presenting as mood instability.

“There are many conditions that can mimic bipolar disorder,” she said. “My work goes into how someone’s mood, anxiety, and energy change over time to pinpoint what treatment will actually work.”

The Bigger Picture

Dr. Ortiz hopes future studies will extend well beyond the typical six-month follow-up, since rare but impairing episodes such as mixed states can otherwise go undetected. She also wants the field to move past observations already obvious to an experienced clinician, toward the harder question of what happens between office visits.

That ambition extends past psychiatry. Dr. Ortiz has published research showing that patients with a history of suicide attempts exhibit distinct breathing patterns tied to an overactive sympathetic nervous system, and she is now studying sleep apnea with UT Southwestern’s Sleep Disorders Clinic, a condition linked to suicide rates roughly five times higher than average in Danish research. “Depression doesn’t happen only in your head,” she said. “It affects your whole body, every system, particularly the cardiovascular and respiratory systems.”

For Dr. Ortiz, that framework rests on an idea borrowed from yoga, where holding a position takes small, constant adjustments, not rigidity. “If you want to hold your stability, you shake a tiny bit, you adjust,” she said. “It is the same for mood. You need those constant adjustments in sleep and activity for your mood to be steady but flexible – ready to respond to whatever the environment throws at you; not stuck. Your mood changes a tiny bit every day, and that is healthy. It’s like bamboo bending in the wind.”

These efforts point toward a future where long-term stability in bipolar disorder is not just a goal, but within reach.