LocalFoodAI_lanfr144/docs/Presentation_User.md

The current version is $Format:LocalFoodAI_lanfr144:Presentation_User.md:%an:%ae:%ad:%cn:%ce:%cd:%H:%D:%N$

Local Food AI - Clinician User Manual Presentation

This presentation slides outline the core features, clinical workflow, and privacy guarantees of the Local Food AI system from a clinician's perspective.

---

Slide 1: Platform Vision & Clinical Core

The Local Food AI system is a strictly offline, privacy-first tool designed to assist clinical dietitians and nutritionists.

Core Value Pillars:

• True HIPAA Compliance: All patient medical queries, health profile selections, and custom diets remain inside your local facility. No data is sent to external cloud APIs.
• Intelligent Clinical Guardrails: The application adapts to specific patient profiles (e.g., pregnancy, diabetes, kidney disease) and automatically flags risks.
• Instant In-House Analytics: Combines raw database search, a portion calculator, and an AI chat consultation into one simple web dashboard.

---

Slide 2: Accessing the App & Sidebar Indicators

Clinicians can log in securely from any workstation inside the facility network.

Accessing the Dashboard:

1. Open your web browser (Chrome, Firefox, or Safari).
2. Enter the address provided by your IT administrator (e.g., http://192.168.130.170:8502).
3. Log in with your secure clinician credentials.

Sidebar Controls:

• Network Status: Indicates if the app is in Online/Server mode or Offline/Local Fallback mode.
• LLM Engine: Shows the active AI reasoning model (e.g., llama3.2:3b).
• Git Version Header: Displays the system Git version, date, and commit code for audit logging.

---

Slide 3: Tab 1: Clinical Data Search (Clinical Search)

This tab enables high-speed lookups against millions of OpenFoodFacts entries.

Key Capabilities:

• Keyword Filter: Search for products, brands, or barcodes (e.g., "Greek Yogurt" or "Cheddar").
• Nutritional Grading: Displays the Nutri-Score (A through E) and macros per 100g.
• Dynamic Indicators: Matches the product attributes against your active client profile:
  • [Warning] Red Flags: Highlights high-risk ingredients (e.g., Unpasteurized dairy for pregnancy, high-sodium for hypertension, or high-sugar for diabetes).
  • [Recommended] Green Flags: Highlights beneficial components (e.g., High-iron for anemia, high Vitamin C for scurvy).

---

Slide 4: Tab 2: My Plate Builder (My Plate Builder)

Build custom recipes or track a client's daily meals to calculate cumulative macro and micro-nutrients.

Features:

• Adding Items: Click Add to Plate on any food item from the search results list.
• Natural Unit Converter: Enter quantities using everyday units (e.g., "1.5 cups", "2 tablespoons", "150g"). The system parses the unit and converts it to metric weight based on ingredient density.
• Intake Metrics: Calculates and displays total energy (kcal), proteins, fats, carbohydrates, sugars, and sodium.
• Comparison Graph: Shows a bar chart of the plate's macros against recommended daily intakes.

---

Slide 5: Tab 3: Consultation Chat (AI Consultation)

Consult the built-in AI assistant to ask clinical questions, verify recipes, or evaluate specific ingredients.

Workflow:

1. Select Client Profile: Check specific health profiles (Pregnancy, Anemia, Keto, Vegetarian, etc.) in the sidebar.
2. Consult the AI: Type questions (e.g., "Is unpasteurized brie cheese safe for this client?" or "Suggest high-iron snacks").
3. Verified RAG Search: The AI queries the local database and private search engines first to verify nutritional facts before answering.
4. Reasoning Steps: View the AI's step-by-step reasoning explaining how it checked the ingredients against the client's medical conditions.

---

Slide 6: Tab 4: AI Meal Planner (AI Planner)

Generate customized, multi-meal dietary schedules directly from the clinician dashboard.

Features:

• Enforce Restrictions: Enter the daily target calorie count (e.g., "2000 kcal") and select dietary constraints (e.g., Kosher, Vegan, Low Fat).
• Automated Diet Grid: The local LLM queries the food database and generates a structured breakfast, lunch, and dinner table.
• Printable Output: Outputs a clean Markdown table summarizing the meals, ingredients, and total daily nutritional values.

---

Slide 7: Security & Resetting Credentials

The system protects both patient privacy and clinician accounts.

Authentication Guidelines:

• Cookie-Based Sessions: Your session is securely saved in your browser cookie jar. Click Logout in the sidebar to terminate it.
• Password Resets:
  1. Click Reset Password in the sidebar.
  2. Input your username.
  3. A secure link will be sent to your registered email address.
  4. Access the link to set a new password complying with local complexity requirements.

---

Slide 8: Privacy Guarantees: Keeping Data Local

Why Local Food AI is safer than general-purpose online chatbots:

• No Cloud Sharing: Traditional AI apps send your prompt history to external corporate servers, risking leaks of confidential patient data.
• Local Processing: Your LLM engine and databases run entirely inside your facility's firewall.
• Anonymous Web Queries: If the AI needs to check the web for a rare ingredient, it routes the query through an anonymous local proxy (SearXNG).
• IT Verification: Traffic audits (tcpdump logs) confirm that zero prompt logs or profile details leave the server boundary.