A Virtual Reality Breathing Game in Forensic Psychiatry

zahra · November 2, 2026, 4:22pm

Can a Virtual Reality Breathing Game Help Forensic Psychiatric Patients Manage Stress and Anger?

Reducing aggression is an important goal in forensic mental health care. However, treatment often relies on talking and reflecting on behavior, which can be challenging for patients with low literacy or limited motivation for treatment.

DEEP takes a different approach. It is an immersive virtual reality (VR) game in which players navigate an underwater world by controlling their breathing. An abdominal belt measures diaphragmatic breathing and provides real-time feedback in the virtual environment.

The DEEP study examined how six men receiving inpatient forensic psychiatric care in the Netherlands responded to the intervention. Researchers repeatedly measured participants’ stress, emotional tension, and anger before, during, and between DEEP sessions. Participants answered three short questions in the Avicenna app (then Ethica) three times a day for 20 days. The team also interviewed participants and their health care professionals (HCPs) to understand their experiences.

The study was led by researchers at the Centre for eHealth and Wellbeing Research at the University of Twente and the Department of Research at Transfore, a Dutch forensic mental health organization. Data collection took place between May and October 2022.

Repeated measurements can reveal individual responses to an intervention, but missing data can limit what researchers can conclude.

Study at a Glance

Study feature	Detail
Study design	Mixed-methods, alternating-treatment single-case experimental design (SCED)
Population	Men receiving inpatient forensic psychiatric care for aggression regulation problems
Participants	6 inpatients aged 23–45 and 5 HCPs
Data collection	May–October 2022
Monitoring period	20 days per participant
Intervention	Four planned DEEP sessions per participant, using an Oculus Rift headset and an abdominal breathing belt
Experience sampling	Three prompts per day at 10 a.m., 2 p.m., and 9 p.m.; up to 60 responses per participant
Measures	Self-reported emotional tension, stress, and anger, rated on a 5-point scale
Interviews	Separate interviews with participants and HCPs about one week after the monitoring period
Tools	Avicenna (then Ethica), DEEP, R, and the MultiSCED web application
Ethics approval	Ethics Committee of the University of Twente, BMS, no. 220190

How the Study Worked

Enrollment and App Setup

A clinician first asked each patient for permission to be contacted by the research team. Researchers then met with interested patients without an HCP present to help reduce the possibility of perceived pressure to participate.

After obtaining informed consent, researchers helped participants install the Avicenna app and explained how to answer the questions. When needed, they read the questions aloud to participants. Data collection began one week later.

Twenty Days of Emotional Check-Ins

Participants received three prompts a day for 20 days. Each prompt asked them to rate how they felt at that moment using three statements:

“At this moment I feel emotionally tense.”
“At this moment I feel stressed.”
“At this moment I feel angry.”

Participants responded on a 5-point scale, from 1 (“not the case at all”) to 5 (“very much the case”).

These brief, repeated assessments allowed researchers to track changes in participants’ reported feelings over time, rather than relying only on a single assessment before and after the intervention.

Alternating DEEP and Non-DEEP Periods

The study used an alternating-treatment single-case experimental design. Participants completed an initial baseline period without DEEP, followed by alternating periods with and without the intervention.

Each participant was scheduled to complete four DEEP sessions on separate days, with days between sessions. A researcher set up the VR headset and breathing belt and guided the participant through the first session. Participants completed the remaining three sessions independently, while the researcher observed.

Three of the six participants did not complete all four planned sessions. One session was missed because of a conflict with another inpatient, another because the researcher was unexpectedly unavailable, and a third because of a technical problem with the VR equipment.

The researchers compared self-reported ratings across the different periods to examine whether changes in stress, emotional tension, or anger were associated with DEEP sessions.

Interviews to Understand Participants’ Experiences

About one week after the 20-day monitoring period, researchers interviewed each participant and their HCP separately. The interviews explored how participants experienced DEEP and whether they noticed changes in their emotional state or behavior.

Combining repeated self-reports with interviews allowed the researchers to examine both measured changes and participants’ experiences of the intervention.

What the Study Found

The experience-sampling results varied across participants.

Short-term responses: One participant showed a temporary reduction in stress after his second DEEP session. Another experienced increased emotional tension after his first session, but the increase did not persist.
Changes over time: One participant showed declines in stress, emotional tension, and anger across the study period. Another showed a decline in stress. These improvements occurred over time but could not be directly attributed to DEEP sessions.
Differences between measurements and interviews: Most participants found DEEP enjoyable and relaxing, and some said it helped them learn or focus on deep breathing. One HCP reported that a participant was less angry and less prone to explosive reactions toward staff during the study.

However, positive experiences did not always correspond to changes in the quantitative measures. Some participants said DEEP helped them relax during a session but found it difficult to use deep breathing during a conflict, when they were already angry or overwhelmed.

The authors concluded that DEEP may be more useful as part of ongoing treatment than as a stand-alone intervention. They also suggested that completing short emotional check-ins three times a day may itself have encouraged participants to reflect on their feelings. This is one possible explanation for improvements over time that were not directly associated with DEEP sessions.

The small sample and variation in individual responses limit conclusions about the intervention’s effectiveness. The study did not establish that DEEP reduced aggression.

What the Data Collection Revealed

Single-case experimental designs rely on repeated measurements across different study phases. In this study, missing data limited the researchers’ ability to calculate SCED-specific effect sizes.

Across the six participants, 70 of 360 scheduled prompts went unanswered, about 19%. Participants missed between 1 and 20 prompts. In addition, three participants missed one of the four planned DEEP sessions because of a scheduling conflict, the researcher’s unavailability, or technical problems.

Three participants did not have the five data points in the intervention phase needed for the planned SCED-specific effect-size calculations. The researchers therefore used visual analysis and linear mixed models to analyze the data.

The authors identified several considerations for future studies:

Collect enough data points in each phase. The authors recommend at least five data points per phase to support SCED-specific analyses.
Integrate the intervention into regular treatment. Extending the DEEP phase and embedding sessions in treatment as usual could help researchers collect more data and examine longer-term changes.
Consider more sensitive response scales. The authors discussed whether a visual analog scale ranging from 0 to 10 or 0 to 100 might capture changes in emotional states more precisely than the 5-point scale used in this study.

Where Avicenna Fits

Avicenna supported the repeated self-report component of the study. It delivered three scheduled prompts per day and collected participants’ ratings of stress, emotional tension, and anger throughout the 20-day monitoring period.

These timestamped responses allowed researchers to track individual changes over time and compare reported feelings across periods with and without the DEEP intervention. This provided repeated measurements of participants’ emotional states throughout the study, complementing the VR sessions and follow-up interviews.

In Their Words

We asked Lisa Klein Haneveld, the first author of the DEEP paper, about running the study.

What should a researcher planning a similar study know before getting started?

“When you conduct a study by use of experience-sampling, I think it is important to think about which questions you want to ask (in level of difficulty and/or nuance) and how many times per day. In forensic psychiatry we noticed that notifying patients thrice a day has a large impact on the time period in which they keep responding (and being motivated). So in our case: more notifications per day meant less days to gain data.”

Was anything about running the study easier or harder than expected?

“Using Avicenna was my first experience with conducting a SCED, using experience-sampling. I remember that the platform was very easy to use. I also liked that I could use my own images and format to further personalize the questionnaire. The data collection went smoothly, and I can’t remember any large set-backs or technical issues.”

What did Avicenna make possible that would have been difficult otherwise?

“What I liked about the use of digital tools and/or an app to conduct experience-sampling studies, is that as a researcher you can monitor how the response is going (and whether you need to further motivate a participant).”

Research Team

Lisa Klein Haneveld

First and corresponding author

Centre for eHealth and Wellbeing Research, University of Twente; Department of Research, Transfore

Hanneke Kip

Senior author