# Sirona AI — Comprehensive Clinical, Regulatory, and Safety Disclaimer > [!CAUTION] > **IMMEDIATE HUMAN INTERVENTION REQUIRED IN EMERGENCIES** > Sirona AI is **NOT** a medical emergency dispatch system, licensed healthcare provider, clinical triage operator, or certified clinical utility. > If you are experiencing a life-threatening medical crisis, severe physical trauma, stroke symptoms, cardiac arrest, or acute psychiatric distress, please immediately contact your regional emergency services (e.g., **911** in the US, **112** in Europe, or **999** in the UK) or proceed to the nearest emergency department. > Do **NOT** input emergency symptom logs into Sirona AI, as automated alerts, countdown locks, or offline fallbacks may fail or undergo delay. --- ## 1. Experimental Status & Regulatory Position > [!WARNING] > **NO REGULATORY APPROVAL OR MEDICAL DEVICE CERTIFICATION** > Sirona AI is a non-commercial, proof-of-concept, and experimental software application designed solely for engineering demonstrations, academic benchmarks, and educational references. > * **No FDA/CE Clearance:** This application and its subsystems have not been reviewed, cleared, approved, or certified by the United States Food and Drug Administration (FDA), the European Medicines Agency (EMA), or any other domestic or international regulatory body. > * **Not a Medical Device:** Sirona AI is not classified as a Software as a Medical Device (SaMD) and must not be used in active clinical settings, real-world triage pipelines, patient intake routing, or clinical diagnosis. > * **AS IS / AS AVAILABLE:** The software is provided "as is" without warranties of any kind. You assume 100% of the risk associated with accessing or interacting with this prototype. --- ## 2. No Physician-Patient Relationship Use of this application, including filling out category-based triage quizzes, generating AI-based diagnostic indicator summaries, reading the Taboo Health Directory, logging chronic symptoms, or conversing with the AI Wellbeing Companion, does **NOT** establish a physician-patient relationship, professional clinical consultation context, or healthcare services contract between you and Sirona AI, its developers, or its sponsors. Sirona AI does not practice medicine, prescribe pharmaceuticals, recommend dosage adjustments, or provide clinical counsel. --- ## 3. Large Language Model (LLM) & Inference Limitations > [!IMPORTANT] > **AI INFERENCE IS STOCHASTIC AND SUBJECT TO HALLUCINATIONS** > Sirona AI integrates third-party cloud-based inference engines, specifically the Llama-4 Scout Engine (`meta-llama/llama-4-scout-17b-16e-instruct`) via the Groq API. > * **Hallucination Risk:** Generative AI models are predictive statistical engines. They can, and do, generate plausible-sounding but factually incorrect information, including fabricated medical facts, non-existent contraindications, erroneous SNOMED CT / ICD-11 mappings, or inaccurate pharmaceutical guidance. > * **Low-Temperature Limitations:** Although Sirona AI calls the API with a highly conservative temperature payload (`0.15`) to reduce variance and minimize hallucinations, the generated report remains a probabilistic approximation, not clinical fact. > * **No Human Verification by Default:** Sirona AI reports are generated entirely programmatically. Users must implement a strict **Human-in-the-Loop (HITL)** process, ensuring any report is vetted by a licensed healthcare professional before any diagnostic or therapeutic action is taken. --- ## 4. Feature-Specific Disclaimers ### A. Multimodal Visual Logging The application allows users to upload local images (e.g., rashes, cuts, eye redness) for base64-based vision-language analysis. * **Diagnostic Insufficiency:** Uploaded photos are processed ephemerally. A compressed 2D image cannot replace a physical dermatological or clinical examination. Environmental lighting, camera resolution, and compression artifacts can severely distort visual symptoms. * **No Diagnostic Claims:** Sirona AI's vision analysis does not constitute a diagnostic determination. It only suggests statistical matches for educational investigation. ### B. Category-Based Triage Quizzes & Redirection The category quizzes (e.g., Neurological, Respiratory, Skin, Gut) and automatic diagnostic bridging are designed to assist you in organizing your symptom descriptions before consultation. * These questionnaires do not represent validated clinical intake protocols and have not been certified for patient diagnostic classification. ### C. Stigma-Free Health Directory (`awareness.html`) The taboo health directory provides educational overviews of highly stigmatized or neglected tropical diseases (e.g., early-stage Leprosy, Reproductive Health/RTIs). * **Educational Library Only:** Screening indicators (such as the ABCDE mole matrix or sensory patch self-testing descriptions) are provided strictly for public health awareness and early detection advocacy. They do not substitute for official laboratory cultures, biopsy examinations, or specialist consultations. ### D. AI Wellbeing Companion & ElevenLabs TTS The empathetic emotional support companion modal and Web Audio binaural synthesis engine are designed strictly for stress reduction, cognitive relaxation, and mindfulness exercises. * **Not Psychiatric Therapy:** The companion chatbot and TTS audio outputs are not licensed psychiatric, psychological, or counseling services. They are not intended to manage clinical depression, severe anxiety, personality disorders, suicidal ideation, or mental health crises. * **Safety Lockouts:** If critical psychological distress keywords are flagged, the companion modal will lock down and display helpline resources (e.g., **988**). This block is client-side, best-effort, and must not be relied upon as a primary crisis safety net. ### E. Ambient Audio Synth & Web Audio Mixer The dynamically synthesized client-side audio drones and binaural beat mixers are intended for relaxing background environments. Sirona AI makes no clinical claims regarding the efficacy of binaural beats or synthesized audio frequencies for treating clinical insomnia, chronic pain, ADHD, or cognitive impairments. --- ## 5. Security, Local Sandboxing & Privacy * **Local-First Caching:** Guest credentials, symptom logs, and local checklists are stored locally on your device via the browser’s `localStorage` API. If you clear your browser cache, delete application data, or switch devices, your medical IDs and consultation logs may be permanently lost. * **Data Transmission:** Ephemeral base64 image strings and symptom text are transmitted over secure TLS connections to third-party endpoints (Groq, ElevenLabs, Firebase) to fulfill requests. While Sirona AI does not index these image files permanently on its remote database, transmission over public networks carries inherent risks. Sirona AI does not guarantee HIPAA compliance or clinical-grade encryption standards for third-party endpoints. --- ## 6. Infrastructure & Safety Shield Dependencies > [!WARNING] > **SOFTWARE CRITICAL DEPENDENCY WARNING** > Sirona AI’s Safety Shield (prompt moderators, emergency lockdowns, countdown timers, and redirect overrides) relies on the proper execution of client-side JavaScript, stable local memory, network availability, and third-party API uptime. > * **Connection Loss:** If your device loses network connectivity, is blocked by firewalls, or experiences API timeouts, Sirona AI will attempt to route queries to a simple local key-matching emergency dictionary. However, the system cannot guarantee that safety indicators, emergency blocks, or clinical alerts will render successfully. > * **Client Failures:** Disabling JavaScript, using outdated browser engines, running conflicting browser extensions, or experiencing memory leaks will bypass or disable local safety locks. Sirona AI assumes no responsibility for safety guardrail failures caused by browser-level execution errors. --- ## 7. Limitation of Liability & Release of Claims To the maximum extent permitted by applicable law, the developers, creators, contributors, and distributors of Sirona AI (collectively, the **"Sirona AI Project Team"**) disclaim all liability. In no event shall the Sirona AI Project Team be held liable for any direct, indirect, incidental, special, consequential, or punitive damages, or any clinical complications, personal injury, wrongful death, pharmaceutical misadventures, delayed medical interventions, or medical costs arising out of: 1. Relying on triage scores, urgency levels, confidence metrics, or reports generated by this application. 2. Relying on first-aid manuals or emergency guidelines provided in the SOS modules. 3. Trusting the text-to-speech audio outputs or relaxation drones. 4. Data loss or local browser storage corruption. 5. Inability to access local emergency override screens due to device or network outages. --- ## 8. Voice-Over Outro Scripts (For Audio/Video Media) If you are recording demonstration videos, testing modules, or showcasing clinical benchmarks of Sirona AI in public forums, you must include one of the following audio-visual disclaimers: ### Standard Outro (~20 seconds) > *"Sirona AI is an experimental, non-commercial proof-of-concept project created for educational and developer benchmarking purposes only. It is not approved by the FDA, does not establish a doctor-patient relationship, and should not be used to replace professional medical advice. If you are experiencing a medical emergency, call local dispatch services immediately."* ### Full Legalistic Outro (~35 seconds) > *"The Sirona AI system demonstration featured in this presentation showcases simulated clinical triage scenarios and fictional patient symptoms. Sirona AI is an experimental engineering prototype and research reference tool, not a certified clinical software product. This platform does not diagnose medical conditions, prescribe medications, or establish a physician-patient relationship. All outputs must be verified by a licensed healthcare professional. In the event of a medical emergency, please dial 911 or your local emergency number immediately."* --- *For technical inquiries, academic research collaboration, or feedback on AI benchmarking prompts, please contact us at [support@sironaai.com](mailto:support@sironaai.com) or review our official legal portal in [legal.html](file:///c:/Projects/AiDoc/legal/legal.html).*