A new study of military veterans and active-duty service members at risk for suicide suggests that brief smartphone check-ins between therapy sessions may predict worsening suicidal thoughts more accurately than relying on standard weekly clinical assessments alone. The findings point to a promising new approach for monitoring suicide risk, although researchers say additional studies are needed to confirm the results in broader patient populations.

The study, led by LDI Senior Fellow Lily Brown, PhD, Director of the Center for the Treatment and Study of Anxiety (CTSA) at the University of Pennsylvania’s Perelman School of Medicine, and Tao Lin, PhD, Instructor at the CTSA, evaluated whether ecological momentary assessment (EMA) could improve clinicians’ ability to detect worsening suicide risk between therapy sessions. The paper, “Does Ecological Momentary Assessment Improve Prediction of Suicide Risk Among Military Personnel and Veterans, and How Frequent Should it be?,” was published in the November, 2026 online edition of the Journal of Affective Disorders.

Dramatic Fluctuations

For years, suicide researchers have recognized that weekly therapy sessions provide only brief snapshots of a patient’s mental state. A person’s suicide risk can fluctuate dramatically over the course of days—or even hours—making it difficult for clinicians to recognize dangerous changes before the next appointment. EMA is designed to help fill those gaps.

EMA is a method used in psychiatric research that collects real-time information about a person’s thoughts, emotions, behaviors, and surroundings throughout the day, usually through a smartphone app. Instead of asking, “How have you felt over the past two weeks?” as done in standard weekly in-person therapy sessions, the EMA asks, “How do you feel right now?” or “How are you feeling at this moment?”

Despite its name, “ecological” refers not to the natural environment but to studying people in their everyday lives rather than in clinics or laboratories. “Momentary” means experiences are recorded as they occur or shortly afterward.

Brown said safety and liability concerns are factors in why EMA hasn’t yet become routine clinical practice.

“Clinicians are overwhelmed and understandably concerned about changing their practice in ways that they perceive could increase liability—like assessing suicide risk between sessions, when they might not be available to respond to an emergency,” Brown said.

Minimum Accessment Frequency

“Clinicians always wonder, ‘what technology should I use and how much will it cost? How am I going to have time to interpret this information in a meaningful way? How many prompts will overwhelm patients? And, what is the bare minimum assessment frequency that improves precision?’ Our article addresses the last question: every other day seems just about right. The other pieces remain the next frontier in overcoming logistical barriers,” Brown explained.

The issue of finding more accurate ways to monitor current and former military members at risk for suicide is a pressing national issue. According to data from the U.S. Department of Veterans Affairs (VA) National Veteran Suicide Prevention Annual Reports, the number of veteran suicide deaths has been more 6,000 each year since 2008—that’s more than 17 suicides daily across the country.

The researchers reported that after adjusting for age differences, female veterans had suicide rates about twice those of female non-veterans, while male veterans had rates nearly 50% higher than male non-veterans.

Real Time Data Insight

One of the biggest challenges in preventing suicide is that a person’s risk can rise and fall quickly—sometimes within hours. People are often at greatest risk outside normal clinic hours, with one previous study finding that suicidal feelings frequently peak in the middle of the night when mental health professionals are typically unavailable. Yet clinicians typically rely on patients remembering and describing how they felt over the previous week during therapy appointments. Because memory is imperfect, these retrospective reports can miss brief but potentially life-threatening spikes in suicidal thoughts that occur between visits, leaving warning signs unnoticed.

This study involved 62 current and former military members recruited from Marine Corps Base Camp Lejeune in North Carolina and using a nationwide recruitment campaign. The paper combines participants from an RCT and a digital phenotyping study whose observation windows overlapped. Participants received four brief EMA surveys each day for 28 days. The suicide-monitoring portion consisted of a single question asking participants to rate the intensity of their suicidal urge on a scale ranging from “none” to “extreme.”

Rather than testing different texting schedules in separate groups of patients, the researchers used computer simulations to estimate how the system would have performed if fewer surveys had been completed. Those analyses suggested that collecting responses about once every other day retained nearly all of the predictive information while reducing participant burden. Because this was a simulation rather than a randomized comparison, the authors say future studies are needed to confirm the optimal monitoring schedule.

Stronger Predictor

One of the study’s most important findings was that a patient’s peak level of suicidal urge between therapy sessions proved to be an even stronger predictor of future suicidal thoughts than the person’s average level of distress. The result suggests that brief, intense surges in suicidal thinking—episodes that patients may later forget or fail to mention during weekly appointments—carry valuable information about worsening suicide risk.

The researchers caution, however, that the study involved only 62 military service members and veterans receiving outpatient psychotherapy. Because the findings have not yet been validated in an independent patient population, they should be viewed as an encouraging first step rather than definitive evidence that EMA should replace existing monitoring approaches. Larger studies will be needed to determine how well the method performs in other groups of patients and to identify the most effective prompting schedule.

---