Video games built for prevention can reduce teen depression

 Few parents or teachers are likely to suggest that teenagers could benefit from playing more video games. But a new study suggests that a growing class of educational games may be a highly effective tool for treating and preventing depression in adolescents.

Dartmouth researchers recruited 532 high school students for one of the largest clinical trials of a video game designed to raise teens' awareness of mental health issues and substance abuse, according to a report in JAMA Network Open.

The team tracked 269 students who engaged with PlaySmart, a digital behavioral health game developed in the play2PREVENT Lab led by Lynn Fiellin, the study's senior author and a professor of biomedical data science and medicine in Dartmouth's Geisel School of Medicine. 

These students played the game for about an hour after school once a week for up to six weeks. The other 263 students in the trial played commercial role-playing games with no therapeutic content. The researchers followed up with both groups for one year.

One year after the trial, teenagers in the PlaySmart group reported reductions in symptoms of depression that matched or doubled the documented effects of general school-based programs, the study finds. They also reported having a more positive attitude toward seeking psychological care after only six weeks of gameplay.

"One of the biggest challenges in adolescent mental health is that many young people never receive care because of barriers including stigma, limited access to providers, or just not recognizing they could benefit from help," says Caroline Barry, the study's first author. 

"Because behavioral health games like PlaySmart are digital and self-guided, they have tremendous potential to reach adolescents across a wide range of settings who might not otherwise receive care, including communities where access to mental health services is very limited," says Barry, who is a postdoctoral research fellow in the play2PREVENT Lab and the Center for Technology and Behavioral Health (CTBH) at Dartmouth.

Better choices, better outcomes

In PlaySmart, players are the main character of a story in which they navigate scenarios based on experiences that the Fiellin Lab gathered from real teenagers and other stakeholders when they were first developing the game. Behavioral health experts in the lab documented instances of healthy and unhealthy decision-making and identified blind spots in teens' health knowledge.

The game presents students with a series of tasks and choices, such as identifying signs of depression in a friend's room or deciding to turn to a trusted adult for help with a difficult situation. Better choices lead to more positive outcomes for the player and other characters in the game, while bad choices lead to more negative outcomes. 

Players also can work backward through the game to analyze their decision-making, learn recommended approaches to identifying and managing depression, and explore what might have happened had they made different choices.

"One of the biggest challenges in adolescent mental health is that many young people never receive care."

The PlaySmart trial was conducted from October 2021 to February 2024, with data analysis completed in November 2025. Students in the study were between the ages of 16 and 19 and recruited from school-based health programs across Connecticut, where Fiellin first established her lab when she was a faculty member at Yale University. 

They were eligible to participate if they had reported vaping or consuming alcohol, cannabis, or nonopioid drugs within the past 30 days, or if they reported heightened symptoms of depression. Anyone with prior opioid misuse was excluded from the study.

"Serious games for prevention and mental health promotion are a rapidly growing area, but the evidence base is still emerging," Fiellin says. "Few clinical trials in this space combine a randomized design with commercial games as an active control and a 12-month follow-up. The rigorous design of our study alone sets a high bar for demonstrating impact."

About 53% of the 532 students were male, the researchers report. Around 45% were Black and 38% were Hispanic, with both groups having reported among the nation's highest rates of sadness and hopelessness. 

Yet, exploratory analyses of the study data show no significant differences in PlaySmart's effect based on sex, race, grade level, food insecurity, or family history of substance use, Barry says. "The benefit held across all subgroups, which matters for an intervention meant to be delivered schoolwide or in communities broadly," she says.

Fiellin and Barry co-authored the study with Tyra Boomer, deputy director of the play2PREVENT Lab and a research project director in CTBH; Katie Haile, the Fiellin Lab's project manager and a research project director in CTBH; and Haiyi Xie, a Dartmouth professor of biomedical data science and community and family medicine who conducted statistical and data analysis. 

'Fully self-guided by students'

PlaySmart originated as one of 10 projects nationwide selected by an initiative from the National Institute on Drug Abuse and the National Institutes of Health called HEAL, or Helping to End Addiction Longterm, to design and test new interventions for opioid misuse.

The study in JAMA Network Open is the second in a planned series of papers evaluating various effects of PlaySmart that are all based on the same clinical trial. A study led by Boomer and published in the January issue of Nature Health reports that students in the PlaySmart group retained a greater awareness of the harms of opioid abuse following the conclusion of gameplay. 

The Dartmouth team's latest findings are significant when compared to what is normally achieved by universal preventative programs, such as those in a school, that target large populations before symptoms escalate. These can include classroom lessons or small, organized group conversations with teacher advisers where students focus on recognizing and addressing depression in themselves and in peers.

"These programs are delivered broadly, and adolescent mental health is influenced by many factors outside of any single intervention. That's why the prevention literature generally reports modest effect sizes," she says.

A pair of studies from 2017 and 2021 found that depression prevention programs result in small changes in how people score on the eight-point Patient Health Questionnaire, or PHQ-8, a self-reporting tool used in surveys and by clinicians to detect and monitor major depression. Scores range from 0 to 24, with the higher numbers indicating more severe symptoms. 

Teenagers in the PlaySmart group had PHQ-8 depression scores that were 1.12 points lower one year later than the scores of students who did not play the game. This difference is typical of universal and targeted prevention programs and at the upper end of what has been reported in existing studies.

"PlaySmart sits right at the top of the range and was achieved with an in-school digital behavioral health game that's fully self-guided by the students and requires no onsite clinician," Fiellin says. 

A comprehensive 2024 study of short-term school-based intervention programs also finds that marked reductions in PHQ-8 scores rarely last more than six months, with an average decrease of only 0.1 points a year later. "PlaySmart's effect at one year was more than double that benchmark and in a format schools can easily and consistently deliver," Fiellin says. 

PlaySmart is one of five games developed by Fiellin's group that have been implemented in several school districts and other youth-facing organizations. The games are provided through her spin-off company, Playbl, which is independent from her lab at Dartmouth. 

"Right now, PlaySmart is deployed to all students as a universal intervention geared toward prevention," Fiellin says. "However, our new findings suggest this game may have a critical role in increasing the access that young people identified as having more significant struggles with depressive symptoms have to an intervention."

---

Journal

JAMA Network Open

DOI

10.1001/jamanetworkopen.2026.29173 

Children’s and parental screen use and children’s language development

This study examined associations between children’s and parental screen use and children’s language development. Across latent growth models, primary parents’ smartphone screen time was associated with slower growth in children’s receptive vocabulary, although this pattern was not observed for receptive grammar. In addition, children’s ownership of digital devices was associated with lower baseline scores in both receptive vocabulary and receptive grammar. These findings were robust to adjustment for parental education and household income.

Complete study

How colleges shape marital matching and the intergenerational transmission of educational advantage

 The authors of this study use newly digitized records for 1.2 million students at 60 co-educational U.S. colleges, 1915–1943, linked to enrollment records for their children through 1963, to study how colleges shape marital matching and the intergenerational transmission of educational advantage. About 6% of men and 9% of women marry someone from their own college. This homogamy is largely causal: within colleges, more different-sex classmates and longer campus overlap both raise same-college marriage. Same-college spouses are more educated, and this advantage persists across generations. Children of same-college couples more often attend selective colleges, with spousal education explaining a third of the gap.

Teen Suicide and the Outcomes of Surviving Peers

 By age 17, a quarter of U.S. students have experienced a peer suicide. Using linked administrative data from South Carolina and a matched difference-in-differences design, the authors of this study find that exposure to a peer's self-harm death increases the probability of a self-harm diagnosis by nearly 50% and both the incidence and frequency of mental health visits. Effects on care use and criminal behavior are concentrated among white boys, and responses diverge by prior mental health history: students without a prior diagnosis increase felony offending rather than care-seeking. Deaths from assault and transportation accidents produce no comparable rise in self-harm, consistent with contagion.

Effect of labor market downturns on medical school admissions and medical practices

 This paper examines how selection into medicine in the United States influences physician practice styles.  Labor market downturns in other prestigious fields increase applications to medical school and shift the composition of matriculating cohorts toward students from non–pre-medical backgrounds. 

Strikingly, these low–outside-option cohorts practice medicine differently decades later. Conditional on experience, specialty, and outcome year, recession-cohort physicians bill Medicare more per beneficiary despite seeing healthier patients—particularly on discretionary, revenue-generating services—and sort into practice settings with higher-powered financial incentives. The additional spending does not produce detectable improvements in patient health. 

Taken together, the findings point to stronger financial motivation among those drawn into medicine when outside options are weak and suggest that medical school admissions screen imperfectly on the attributes that ultimately shape how physicians practice.

Variation in School-level Child Maltreatment Reporting

Involvement with the child welfare system varies dramatically across places. While some of this variation may reflect differences in underlying maltreatment risk, it may also result from differences in institutional responses to similar circumstances. To disentangle these explanations, this study uses Michigan public school data linked to Child Protective Services reports from 2017–2024. 

School-level reporting rates range from virtually none to more than half of students reported between kindergarten and fifth grade. The authors estimate school-specific reporting tendencies using value-added methods and document substantial variation in reporting behavior after accounting for differences in student risk. Using student movers, they show that these estimates are forecast-unbiased. These estimates imply that school-specific reporting practices account for roughly 40 percent of the variation in reporting rates across schools. 

 Differences in skill explain more than twice as much variation as differences in preferences, and higher-reporting schools tend to be more skilled at identifying high-risk cases. Counterfactual simulations imply that improving detection skill would yield significant welfare gains, while recent policy proposals aimed at reducing or equalizing reporting rates generally lower welfare by increasing the number of high-risk cases that go unreported.

AI Middle School Math Tutoring : Minimal Sudent Engagement

 Generative AI has been promoted as the technology that could transform education by providing every student a personal tutor. This study provides some of the first large-scale experimental evidence, from a two-year cluster randomized trial in 18 Tennessee middleschools in which randomly assigned students used Khan Academy with its AI tutor, Khanmigo, configured to coach rather than give answers, during existing daily remedial mathematics sessions. 

Assignment raises math achievement by 1.3 national percentile ranks per term, or about 0.06 to 0.08 standard deviations over a school year; the implied effect of a full year of active participation reaches 0.14 standard deviations. These gains resemble those from Khan Academy practice without AI assistance. 

One explanation is that students used the tutor infrequently and, when they did, rarely engaged it in substantive mathematical dialogue: 96 percent of students tried Khanmigo at least once, but the median student messaged it on only a third of the days they practiced, and in only 17 percent of the exercise sessions in which they made a mistake. Messages that students did send were mostly bare answers or clicks on suggested prompts. 

The binding constraint appears to be engagement: realizing the promise of AI tutoring will require getting students to use it, not just giving them access.