AI-MHApp · thePsychology.ai Lab

9:41 5G ••• 84%

Tuesday, quiet startMorning, Renée

RM

0 Balance

JD-R read

Resources are outpacing demands

Your load eased since Thursday. Two recharge habits are holding.

Resources 68 Demands 32

JMReviewed by Dr. Julian Meade, LP

"I read your Friday check-in. That back-to-back crisis load is a lot to carry. I flagged a boundaries module and I'm here Thursday if you want to talk."

Today

🫁

90-second resetBetween sessions breathing

›

📋

Weekly check-in dueProQOL + mood, ~3 min

Due today

⌚

Sleep dipped 2 nightsFrom your wearable

Watch

The synergy, made literal

AI carries the constant attention. A clinician carries the care.

Lorin's thesis is that neither works alone: AI without oversight dehumanizes, human-only support doesn't scale to a burned-out workforce. Every teal touch is the machine; every apricot touch is a person.

📊

Self-assessment tools

ProQOL, PHQ-2 and GAD-2 surfaced as gentle weekly check-ins, not clinical intake forms.

💬

AI companion with a human on call

Always-on support that hands off to a licensed clinician the moment risk climbs.

🎓

Training & resource modules

Five-minute, evidence-based lessons on burnout, boundaries and secondary trauma.

⌚

Wearable & mood signals

Passive sleep and stress data flag a rough stretch before you'd name it yourself.

Job Demands-Resources model Perceived Organizational Support Confidential by design

Riverside Behavioral Health · Team well-being

142 staff enrolled · rolling 30 days

🔒 Anonymous · aggregated · min. 5 to display

Team balance (JD-R)

64

▲ 7 vs last month

Elevated burnout

18%

▼ 5 pts

Using resources weekly

71%

▲ 12 pts

Felt org support (POS)

3.9/5

▲ 0.4

Team balance trend

Average demands-vs-resources score, weekly. Higher is healthier.

Burnout risk distribution

Where the team sits this week. No individual is identifiable.

Healthy58%

Watch24%

Elevated18%

Suggested action
Crisis-intake team is driving elevated scores. Consider rotating on-call load, projected +9 balance points.

Employers never see names, individual scores, chats, or check-in answers. Cells with fewer than 5 responses are hidden. This is the entire surface an organization sees, by design, per Lorin Singh's confidentiality framework.

Evidence update · 2024-2026

What the newest research says, and what it changes

The 2024 dissertation predates the first real trials of generative-AI therapy. Here is the refreshed evidence base, verified through OpenEvidence, and the three design calls it points to.

g ≈ 0.26

chatbot effect on depression, fading by 3 months

51%

depression drop in the first generative-AI therapy RCT (Therabot)

92 → 62%

app uptake vs. completion, engagement is the gap

3

digital therapeutics FDA-approved for mental health, so far

What changes for the app

01

Human oversight is now the consensus, and the law

Every 2025-26 review lands on supervised hybrid stepped-care. Illinois banned AI therapy without a licensed clinician; the APA says don't let generative AI deliver therapy. The clinician hand-off isn't a nice-to-have anymore, it's the requirement your design already meets.

02

Gamification looks like a liability

The 2026 JAMA Psychiatry attrition study (79 RCTs) found reminders and human contact lower dropout, while gamification is tied to higher dropout. Lead with reminders, human contact, personalization, and symptom monitoring instead.

03

Wellness-first is the only sane runway

Effects are modest and fade past 3 months, so the bar for a medical claim is real. Launch as a wellness / organizational tool, gather real-world evidence, then pursue FDA De Novo and a German DiGA listing (insurance-reimbursed).

The evidence base (Chapter 2 refresh)

AI chatbots work, modestly, short-term. Depression lands at g ≈ 0.26-0.55, anxiety g ≈ 0.19-0.26, with effects fading by 3-month follow-up. Therapeutic alliance predicts benefit, which is exactly what the human hand-off protects.Zhong 2024 · Zhang 2025 · Gong 2026 · Shoshani 2026
Worker-facing trials now exist. Digital well-being training in 2,315 health professionals showed effects that grew over time (distress −0.24 → −0.36); digital CBT beat job crafting for burnout in 465 NHS staff.Hirshberg 2025 · Delgadillo 2025
Engagement is the real bottleneck. 79-RCT benchmark: 92% uptake but 62% completion. Reminders and human contact help; gamification hurts.Liu 2026, JAMA Psychiatry
Guided beats unguided. Internet CBT matches face-to-face long-term, but therapist guidance lifts completion from 48% to 79%.Bakanaitė 2025 · Karyotaki 2021

Design & regulation (Chapter 5 refresh)

Stepped-care hybrid is the endorsed model. AI does companionship, triage, and monitoring; clinicians keep diagnosis, risk, and medication.Calderone 2025 · Yang & Lin 2026
Suicide-risk detection is flag-only. LLMs reach AUROC ≈ 0.73, good enough to escalate to a human, not to decide. Needs defined response times and audit trails.Lho 2025
VR has real support for stress (−0.55) and anxiety (−0.88); the anger use case is under-studied, so don't overclaim the "crash-out room."Strauch 2026
Regulation: the US treats it as Software as a Medical Device (510(k)/De Novo); Germany's DiGA reimburses via insurance since 2019. Pure-AI-therapist apps (Woebot, Ash) shut down or got pulled.Angus 2025 · DiGA reviews

VR decompression, the "crash-out room"

"Crash-out room" isn't a clinical term, but the concept is well studied. Spaces to decompress from acute stress during and between shifts appear in the literature as "recharge rooms" and "calm rooms," increasingly delivered through VR.Putrino 2020 · Cooke 2025
Recharge rooms work at scale. Multisensory recharge rooms across 18 US hospitals (n = 563) cut self-reported stress 59% in 15-minute sessions.Cooke 2025 · Putrino 2020
Three minutes is enough to move the needle. A 3-minute nature VR for frontline COVID workers produced a very large effect on stress (d = 1.08); the high-stress share fell from 32% to 4%.Beverly 2022

Meta-analytic support in healthcare workers. A 2025 meta-analysis of VR for healthcare-worker mental health (10 studies) found stress SMD −0.64 and anxiety −0.51.Chen 2026
The anger angle has direct evidence. VR anger-exposure plus mindfulness training reduced anger scores, and VR Aggression Prevention Treatment (VRAPT) improved aggression and emotion regulation.Jo 2022 · Woicik 2023
Design takeaway. Brief (3-15 min), nature-based, audiovisual VR immersion is the best-supported acute-stress tool; multiple sessions beat single. The main barrier is practical, time and space, not efficacy.Strauch 2026

Full write-up with effect sizes and APA-7 references is in the companion literature-update addendum. Therabot (NEJM AI, 2025) is cited from the primary source, it was not indexed in OpenEvidence at the time of writing.

VR module · release first, regulate after

The crash-out room

Working a psychiatric center "makes you a very angry person." This room takes that seriously: hit something that can take it, then learn the regulation skill while your nervous system is actually listening. Your AI companion builds tonight's room from your check-in.

Personalized by your companion

Tonight's room is built from your check-in, three crisis intakes, tension running high.

Room ready

Step in when you're ready. Nothing in here needs you to be okay yet.

Session complete

Tension 82 → 24

You named it, released it, and chose your next move. That's the whole skill.

Tension · JD-R read: demands spiked today 82

Scene

How the room works

1

Release

The heavy bag takes what the shift left in you. Tapping it burns the charge down, anger gets a place to land before anyone asks it to behave.

2

Regulate

Once tension drops, the room softens and walks you through box breathing (4-4-4-4), the anger-regulation skill, taught when your body can take it in.

JMClinician in the loop

Sessions where tension stays above 70 get flagged to Dr. Meade, with your consent, so the room is never the only thing holding you.

Why this works

Three minutes of nature VR cut acute stress in frontline workers with a very large effect (d ≈ 1.08), and VR anger-exposure with mindfulness training reduces aggression and improves emotion regulation.

Beverly 2022 · Jo 2022 · Woicik 2023 · full detail on the Evidence tab