SafeGuard is an enterprise-grade psychosocial health surveillance system focused on high-pressure environments. We employ a dual-tier market strategy:
- Tier S (High Risk/High Impact): Tourism & Hospitality sector (Hotels, FnB).
- Tier A (High Adoption/Tech-Native): Startups, IT companies, and Digital Agencies.
It integrates advanced Safespace AI/ML prognostic models with robust Health Economic Evaluation (HEE). Designed for modern organizations, it monitors, detects, and mitigates mental health burdens using WHO-standard ROI modeling and ICD-10 clinical frameworks.
“1 fatality is not 1 event — it is the visible tip of hundreds of ignored psychosocial signals.”
SafeGuard transforms Fatality (Suicide/CVD) from a lagging indicator into a preventable outcome by monitoring Bullying (UA/UC) and workplace stressors as leading indicators in high-pressure service environments.
SafeGuard operates on the SAFE mnemonic to ensure comprehensive mental health surveillance:
- Sense: Digital anamnesis, wearable sync, and IoT-driven companion hardware scans.
- Analyze: AI-driven risk stratification and prognostic modeling.
- Flag: Automated alerting for Level 0-L2 clinical risks.
- Engage: Personalized wellness roadmaps and clinical escalation.
- Tier S: Hospitality Stress Mapping: Specialized triaging for high-stress roles (Kitchen, Front Office, Housekeeping) facing long hours and emotional labor.
- Tier A: Startup Burnout Surveillance: Monitoring high-velocity work environments, "crunch culture," and cognitive load in the IT and Digital sectors.
- Hierarchical Risk Stratification: Automated triaging from Level 0 (Psychiatric Emergency) to Level 3 (Prognostic/Functioning).
- Digital Anamnesis Analysis: Real-time processing of DASS-21 and SRQ-20 assessments tailored for both service and tech industry stressors.
- Gemini-Powered Empathy: Uses Google Gemini 3 Flash to provide 24/7 initial psychosocial support and empathetic listening.
- Crisis Triaging (L0-L3): Automated detection of high-risk signals within conversation, with immediate escalation to emergency resources and MERP protocols.
- Empathetic Listening: A safe, non-judgmental space for users to express feelings and receive immediate coping strategies.
- Distinct from Risk Detector: While the Risk Detector analyzes data, the Counselor provides direct human-like interaction and emotional support.
- WHO-Standard ROI Modeling: Visualizes the 1:4 Return on Investment (ROI) for mental health interventions (US$1 Trillion global burden context).
- INA-CBG Claim Mapping: Integration of Indonesian Case-Based Groups (INA-CBG) for mean claim analysis (e.g., Rp12.4M/admission) and medication cost tracking.
- Cost of Inaction vs. Intervention: Robust comparative analytics to justify wellness budget allocation for management and stakeholders.
- 1 Fatality = 600 UA/UC: Based on the safety pyramid (Heinrich → Bird), SafeGuard identifies the 600+ "Unsafe Acts" (bullying, toxic culture) that precede a single fatality.
- Biological Pathway: Maps chronic stress to the HPA axis, cortisol elevation, and cardiovascular risk (CVD).
- Predictive Bridge: Turns "near misses" into prevented lives through early detection.
- SATUSEHAT Integration: Fully compliant with HL7 FHIR R4 standards for seamless integration with the National Health Data Exchange.
- Kemenkes Standard: Aligned with Indonesian Ministry of Health (Kemenkes) clinical guidelines and reporting frameworks.
- ICD-10 Chapter V Mapping: Burden clusters categorized by WHO ICD-10 codes (F30-F39 Mood, F40-F48 Anxiety/Stress, etc.) for standardized reporting.
- Community Medicine Utilization Review: Doughnut-based analytics for facility distribution (Puskesmas, RSUD) and medication utilization (Antidepressants, Anxiolytics).
- Dual-Mode Interface: Specialized roadmaps for Worker Mode (Occupational Health) and Patient Mode (Clinical Prevention).
- L3 Prognostic Predictors: Monitoring of workplace conflict, peer support quality, and psychosocial stressors.
- Personal ROI: Individualized estimates of annual healthcare savings and productivity value in IDR.
- DASS-21 & SRQ-20 Integration: Comprehensive psychosocial self-assessment for AI-driven risk detection.
- Multi-Device Wearable Sync: Real-time data syncing for Apple Watch, Garmin, and Fitbit to monitor autonomic nervous system regulation (BPM & HRV).
- IoT Companion Hardware (Arduino): Proprietary IoT tool integrated into toys and companion dolls for school-to-university engagement.
- Hardware-Level TTS: Built-in Text-to-Speech (TTS) capabilities for interactive emotional support and data collection.
- Curated Resource Library: Evidence-based articles on burnout management and cognitive reframing.
- Interactive Exercises: Guided "Box Breathing" sessions and YouTube-integrated "Deep Sleep Meditation" for immediate stress reduction.
- Resource Stratification: Content categorized by Article, Exercise, and Audio for diverse learning preferences.
- AI Counselor Integration: Direct access to empathetic chat support within the wellness ecosystem.
- Adaptive UI Engine: Toggleable Accessibility Mode with high-contrast (B&W) interface and simplified layouts.
- AI Voice Summary (TTS): Real-time audio briefing of health risks and population status for low-vision users.
- Tremor-Resistant Interaction: Extra-large touch targets (p-8/80px) and simplified navigation for users with motor disabilities (Parkinson/Amputation).
- Universal Iconography: Zero-text dependency mode using intuitive icons for low-literacy populations.
- L0 SOS Trigger: Floating emergency button with immediate GPS-linked escalation.
- Automated Dispatch: Real-time connection to refinery medical teams and first responders.
- Voice-Guided Protocol: Audio instructions for users in crisis until help arrives.
- Installable Experience: Add to Home Screen (A2HS) support for mobile and desktop.
- Service Worker Integration: Background sync and offline-first capability for remote industrial sites.
- Real-time Firebase Sync: Instant data synchronization between patients and providers using NoSQL listeners.
- Push Notifications: Real-time behavioral health alerts via Firebase Cloud Messaging (FCM).
- IRB Roadmap: Formal commitment to obtaining Institutional Review Board (IRB) clearance for corporate deployment.
- Informed Consent: Transparent protocols for worker data usage and privacy protection.
- Anonymized Reporting: Aggregate analytics that protect individual worker identities while providing management with actionable insights.
- Standard Rate: Rp 50.000 - 100.000 / employee / year.
- ROI Justification: For a hotel with 200 staff, the cost (Rp 10-20M) is fully offset by preventing just one burnout-related resignation (recruitment cost ~Rp 15M).
- AI API (Gemini 1.5 Flash): ~$5 - $10 / month (based on 10,000 scans).
- Cloud Database (Firebase Blaze): ~$15 - $25 / month (pay-as-you-go).
- Maintenance: Rp 5 - 10 Million / year for technical support and monitoring.
SafeGuard is developed by a multidisciplinary team of experts in AI, product engineering, health data, and clinical occupational health:
- Akhmad Khudri → Team Lead & Product Orchestrator (Strategy & SWOT)
- Stanley Nathanael Wijaya → Lead AI & Product Engineer (Front-end & Integration)
- Dwiki Aulia Rahman → Health Data & Predictive Insight (AI Architecture & Epidemiology)
- dr. M. Rifki Al Ikhsan → Clinical & Occupational Health Advisor (Policy, MCU, HRA, Psychosocial Intervention)
SafeGuard exposes four core skills via the Agent-to-Agent (A2A) protocol:
- Level 0-L3 Triaging: Automated risk detection from Level 0 (Psychiatric Emergency) to Level 3 (Prognostic/Functioning).
- Standardized Assessment: Real-time processing of DASS-21 and SRQ-20 scores to determine clinical urgency.
- AI-Powered Insights: Deep analysis of clinical notes and digital anamnesis to detect behavioral risks and workplace stressors.
- Burnout Surveillance: Specialized mapping for high-pressure environments (Hospitality, Tech, Healthcare).
- Chapter V Mapping: Suggests relevant ICD-10 codes (F30-F39 Mood, F40-F48 Anxiety/Stress) based on patient symptoms and history.
- Clinical Justification: Provides evidence-based reasoning for each recommended code to assist clinicians in documentation.
- Health Economic Evaluation (HEE): Visualizes the 1:4 - 1:5 Return on Investment for mental health interventions.
- INA-CBG Claim Estimates: Integration of Indonesian Case-Based Groups (INA-CBG) for mean claim analysis and potential cost savings.
SafeGuard is built for the Prompt Opinion Marketplace, adhering to the highest standards of interoperability:
- SHARP Compliant: Consumes
fhir-context(Patient ID, Server URL, Token) to fetch real-time patient data. - A2A Manifest: Discoverable via
.well-known/agent-card.jsonwith explicit capability declarations. - REST Transport: Optimized for stable, high-performance inter-agent communication.
- Structured JSON Output: Returns machine-readable clinical and economic data for seamless UI rendering.
- Frontend: React 19, TypeScript, Vite
- Styling: Tailwind CSS 4.0 (Utility-first CSS)
- Animations: Motion (formerly Framer Motion)
- Data Visualization: Recharts (Area, Bar, Pie, Doughnut)
- Backend/DB: Firebase (Firestore, Auth, Messaging)
- PDF Generation: jsPDF (Strategic Whitepaper Export)
- PWA: Service Workers, Manifest.json, Offline-first
- Accessibility: Web Speech API (TTS), WCAG 2.2 Compliance
SafeGuard is built on privacy-first principles.
- Firestore Security Rules: Robust Role-Based Access Control (RBAC) ensuring users only access their own data.
- Data Anonymization: Individual data is anonymized in aggregate reports.
- Encryption: AES-256 at-rest and TLS 1.3 in-transit.
- Compliance: Level 0/1 escalations follow strict HIPAA/GDPR-compliant protocols.
- AI Chatbot Counselor: 24/7 initial support and crisis triaging using Gemini 3 Flash.
- Wearable Integration (Simulated): Real-time data syncing from Apple Watch, Garmin, and Fitbit.
- Wellness Hub Enhancements: Interactive exercises and real-world content integration.
- IoT Hardware Commercialization: Mass production of Arduino-based companion dolls with integrated TTS for educational sectors.
- TWA (Trusted Web Activity): Native Android deployment for Google Play Store.
- Full Wearable API Integration: Live production data from HealthKit/Google Fit.
- Manager/Jury Dashboard: Advanced view for stakeholders with deep-dive ROI simulation tools.
- Multi-language Support: Full localization for global enterprise deployment.
SafeGuard provides a robust set of APIs for integration with the Prompt Opinion Orchestrator and external health systems.
👉 Agent Card (Manifest): https://server-safeguard.onrender.com/.well-known/agent-card.json 👉 Interactive Swagger Docs: https://server-safeguard.onrender.com/api-docs
POST /api/v1/chat: Main A2A communication endpoint for skills execution.POST /api/v1/analyze: SHARP-compliant patient data analysis.GET /api/v1/economic-forecast: ROI and INA-CBG claim estimates.
The platform has been optimized for Full-stack Serverless deployment:
- Frontend: React + Vite (Served via Vercel Edge).
- Backend: Node.js Express (Deployed as Serverless Functions in
/api). - Database: Firebase Firestore (NoSQL) with real-time sync.
- Auth: Firebase Authentication (Google OAuth).
- Messaging: Firebase Cloud Messaging (FCM) via Service Workers.
- Swagger UI: OAS 3.0 with CDN-backed assets (jsDelivr) for high availability.
- Environment Awareness: Automatic switching between Local (Port 3000) and Production (Vercel) contexts.
SafeGuard is part of a broader ecosystem of AI-driven mental health and emotional support tools:
- Halo Ayah (Hey Dad): A warm AI companion providing fatherly guidance, life skills, and emotional support.
- Unsaid: A safe space for unspoken thoughts, helping users process what couldn't be said in the moment.
Developed for the Prompt Opinion Hackathon 2026