What is STATE SHIFTER™ / SENSAI?
STATE SHIFTER™ is an AI-enhanced, multisensory digital therapeutic running on the Sensory Reality Pod (SRP) – a cabin that controls light, sound, scent, temperature, airflow and vibration across 59 actuators to guide users from emotional distress to calm in a single 12-minute protocol.
Clinical Goal
Deliver a brief, evidence-based intervention for stress, anxiety and trauma-related complaints that can be deployed in healthcare, education and high-pressure workplaces.
Hardware Platform
Built on the already commercial Sensory Reality Pod (SRP) with 59 individually addressable actuators for light, sound, scent, airflow, vibration and temperature – validated in daily clinical and corporate use.
AI Innovation
A three-tier AI layer – Static AI, Interactive AI and Adaptive AI – coordinates safe session planning, conversational guidance and real-time bio-feedback adjustment with ≤ 800 ms closed-loop latency.
Addressing Stress at Scale
In the Netherlands alone, 1.3 million workers per year report stress-related complaints, costing €3.1 billion in absenteeism. STATE SHIFTER™ contributes directly to the KIA Health & Care mission of +5 healthy life years and 30% reduction in health inequality.
-
1.3 million workers annually affected by stress-related complaints (TNO 2024).
-
€3.1 billion yearly cost in absenteeism due to stress and burnout.
-
Targets KIA Health & Care and AIC4NL missions for healthier, more resilient citizens and workers.
Health & Care
Adaptive stress reduction and anxiety management for patients and high-risk populations, reducing pressure on overstretched mental healthcare services.
Education
Exam-anxiety and focus pods for students, offering quick resets before high-stakes tests or demanding study periods.
Smart Industry
“Mental reset cabins” for shift workers and high-cognitive-load environments, integrated into workplace wellbeing programs.
How the AI System Works
STATE SHIFTER™ introduces an adaptive AI engine on top of a mature multisensory hardware platform, creating the first commercial closed-loop system across all five senses.
Core Specifications (Target 2026)
- Closed-loop latency: ≤ 800 ms from biosensors → AI decision → actuator update.
- Emotion model: 4-class valence–arousal space (high/low valence × high/low arousal).
- Classification accuracy: 78–82% on real in-pod data.
- Adaptive engine: hierarchical policy combining static templates, rule-based safety layer and contextual bandit optimisation.
- Safety bounds: all actuator changes limited to ±30% of clinically pre-validated static templates.
- Fail-safe mode: automatic fallback to deterministic sessions when model confidence is low.
Current Prototype Outcomes (n = 410 sessions, 2024–2025)
- 29% average reduction in perceived stress (STAI-6) after one session.
- +24% average increase in HRV (rMSSD), indicating improved autonomic balance.
- +3.1 point increase on 0–10 calmness VAS.
- Median 7.8 minutes to reach a predefined “calm zone” based on HRV and GSR thresholds.
SENSAI Architecture: Four Modules
Static AI
Generates a safe 12-minute baseline session (720 × 65 control matrix) from user preferences and clinical templates, ensuring deterministic safety and predictable outcomes.
Interactive AI – “SENSAI”
A conversational agent that speaks with the user before, during and after sessions. Sentiment and intent signals feed back into the adaptive layer for more personalised regulation.
Adaptive AI
A multimodal Transformer-lite model (≤ 15M parameters) fuses EEG, HRV, GSR, facial expression and voice prosody to adjust light, sound, airflow, temperature and scent within safe, validated ranges.
Generative AI
A conditional variational auto-encoder introduces subtle variations in sensory stimulation patterns to prevent habituation.
From Prototype to Certified Medical Device
Work Packages & Milestones
WP1 – Static & Interactive AI (Months 0–4)
Finalise deterministic session planning and conversational layer, and start the CE technical file. Target TRL: 6.
WP2 – Multimodal Classifier (Months 3–9)
Train multimodal emotion classifier on ≥ 2,000 in-pod sessions covering EEG, HRV, GSR, facial and voice signals. Target TRL: 6.
WP3 – Adaptive Closed Loop (Months 6–12)
Implement and validate ≤ 800 ms closed-loop Adaptive AI with a go/no-go gate for full deployment. Target TRL: 7.
WP4 – Clinical Pilot (Months 9–18)
Run a registered clinical pilot (n ≥ 120, 3 sites) to evaluate safety and efficacy in real-world settings. Target TRL: 7–8.
WP5 – Certification (Months 12–20)
Complete CE-marking as a Class IIa medical device and GDPR Data Protection Impact Assessment. Target TRL: 8.
WP6 – Commercial Launch (Months 18–24)
Launch the full pod solution and a VR-only light edition for broader deployment. Target TRL: 9.
Sharing Data, Tools and Governance
Open Science Commitments
Upon project completion, Sensiks will release an anonymised multimodal dataset (≥ 5,000 labelled sessions), pre-trained emotion recognition models on Hugging Face under a research-friendly licence, and an open Python SDK and API so researchers and content creators can build on the platform.
Responsible AI & Ethics
- Independent AI Ethics Officer overseeing bias audits, Model Cards and uncertainty reporting.
- Strict safety constraints on all actuator changes to avoid overstimulation or adverse events.
- Privacy by design with GDPR compliance and a full DPIA for handling biosensor data.
- Human-in-the-loop workflows ensuring clinicians remain in control of therapeutic decisions.
Why MIT R&D AI Support is Critical
Scaling from Prototype to Certified Platform
Sensiks has self-funded more than €2.5 million of development. The MIT R&D AI subsidy will de-risk the final steps from functional prototype to certified, scalable medical device and open platform.
Accelerate Adaptive AI
Complete the Adaptive AI layer and multimodal classifier with robust validation across real-world users and settings.
Clinical Validation
Fund the registered clinical pilot (n ≥ 120), including recruitment, monitoring and longitudinal follow-up.
Certification & Ecosystem
Achieve CE Class IIa certification and deliver the open dataset, models and SDK that maximise public benefit and research reuse.
STATE SHIFTER™ as a Flagship for Responsible AI in Health
STATE SHIFTER™ / SENSAI combines clinically grounded protocols, multisensory hardware and adaptive AI to create a new class of digital therapeutic for stress, anxiety and trauma-related complaints. With MIT R&D AI support, it can become a certified, open, and widely accessible platform that benefits patients, workers and researchers across the Netherlands and beyond.