test

VECTRA/.env

@@ -0,0 +1,4 @@
+PINECONE_API_KEY=pcsk_57FuYE_GKRroZMBbMYXg7XPuBY9edzbojk4pYWr6tEnfYDhdipupy9wzzw1eQ45bktMYoS
+PINECONE_ENV=gcp-starter
+OPENAI_API_KEY=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
+PINECONE_INDEX_NAME=vectra-health

VECTRA/.env.example

@@ -0,0 +1,4 @@
+PINECONE_API_KEY=pcsk_57FuYE_GKRroZMBbMYXg7XPuBY9edzbojk4pYWr6tEnfYDhdipupy9wzzw1eQ45bktMYoS
+PINECONE_ENV=gcp-starter
+OPENAI_API_KEY=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
+PINECONE_INDEX_NAME=vectra-health

VECTRA/README.md

@@ -0,0 +1,50 @@
+# VECTRA: Agentic Healthcare AI
+
+VECTRA is an advanced multi-agent disease prediction system that combines robust Machine Learning with Semantic Search and Large Language Models (RAG).
+
+## Features
+- **Hybrid Reasoning**: Uses Random Forest for high-confidence predictions and LLM+RAG for complex cases.
+- **Multi-Agent Architecture**: Specialized agents for Perception, Reasoning, Decision, Retrieval, and Safety.
+- **Explainable AI**: Provides clear explanations for every diagnosis.
+- **Safety First**: Integrated medical disclaimers and guardrails.
+
+## Setup
+
+1. **Install Dependencies**
+   ```bash
+   pip install -r requirements.txt
+   ```
+
+2. **Configuration**
+   Copy `.env.example` to `.env` and fill in your API keys:
+   ```bash
+   cp .env.example .env
+   ```
+
+3. **Data & Training**
+   Build the knowledge base and train the model:
+   ```bash
+   python knowledge_base/build_kb.py
+   python vector_store/indexer.py  # Requires valid Pinecone Key
+   python models/train.py
+   ```
+
+4. **Run Server**
+   ```bash
+   uvicorn api.vectra_api:app --reload
+   ```
+
+5. **Run Frontend**
+   (Requires React setup - instructions generic)
+   The `ui/frontend` folder contains the source components.
+
+## API Usage
+POST `/predict`
+```json
+{
+  "symptoms": ["fever", "cough", "fatigue"]
+}
+```
+
+## Disclaimer
+This system is for educational purposes only. Please consult a licensed medical professional.

VECTRA/agents/decision_agent.py

@@ -0,0 +1,9 @@
+class DecisionAgent:
+    def decide_route(self, confidence, threshold=0.6):
+        """
+        Decides whether to route to ML Direct or RAG fallback.
+        """
+        if confidence >= threshold:
+            return "ML_DIRECT"
+        else:
+            return "RAG_FALLBACK"

VECTRA/agents/explanation_agent.py

@@ -0,0 +1,9 @@
+class ExplanationAgent:
+    def format_explanation(self, disease, confidence, symptoms):
+        return (
+            f"Based on the symptoms provided ({', '.join(symptoms)}), "
+            f"there is a {confidence*100:.1f}% probability of {disease}."
+        )
+
+    def format_rag_explanation(self, llm_response):
+        return llm_response

VECTRA/agents/perception_agent.py

@@ -0,0 +1,16 @@
+class PerceptionAgent:
+    def process_input(self, raw_symptoms):
+        """
+        Parses and cleans input symptoms.
+        Args:
+            raw_symptoms (list or str): User input.
+        Returns:
+            list: Cleaned list of symptoms.
+        """
+        if isinstance(raw_symptoms, str):
+            symptoms = raw_symptoms.split(",")
+        else:
+            symptoms = raw_symptoms
+
+        cleaned = [s.strip() for s in symptoms if s.strip()]
+        return cleaned

VECTRA/agents/reasoning_agent.py

@@ -0,0 +1,8 @@
+class ReasoningAgent:
+    def analyze(self, symptoms):
+        """
+        Enhances user symptoms if needed, checks for contradictions.
+        For V1, primarily passes symptoms through.
+        """
+        # Future: Ontology mapping or symptom expansion
+        return symptoms

VECTRA/agents/retrieval_agent.py

@@ -0,0 +1,21 @@
+import sys
+import os
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from vector_store.retriever import Retriever
+
+class RetrievalAgent:
+    def __init__(self):
+        self.retriever = Retriever()
+
+    def fetch_context(self, symptoms):
+        query = ", ".join(symptoms)
+        # Setup for RAG: retrieve documents related to the symptoms
+        results = self.retriever.retrieve(query, top_k=3)
+
+        docs = []
+        for match in results:
+            if 'metadata' in match:
+                docs.append(match['metadata'].get('text', ''))
+
+        return docs

VECTRA/agents/safety_agent.py

@@ -0,0 +1,10 @@
+class SafetyAgent:
+    DISCLAIMER = "This system is for educational purposes only. Please consult a licensed medical professional."
+
+    def check_safety(self, text):
+        # Placeholder for PII or harmful content check
+        return True
+
+    def attach_disclaimer(self, response_dict):
+        response_dict["disclaimer"] = self.DISCLAIMER
+        return response_dict

VECTRA/api/vectra_api.py

@@ -0,0 +1,91 @@
+from fastapi import FastAPI, HTTPException
+from pydantic import BaseModel
+from typing import List, Optional
+import sys
+import os
+
+# Add project root to path
+sys.path.append(os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+from models.inference import InferenceEngine
+from agents.perception_agent import PerceptionAgent
+from agents.reasoning_agent import ReasoningAgent
+from agents.decision_agent import DecisionAgent
+from agents.retrieval_agent import RetrievalAgent
+from agents.explanation_agent import ExplanationAgent
+from agents.safety_agent import SafetyAgent
+from rag.rag_pipeline import RAGPipeline
+
+app = FastAPI(title="VECTRA API")
+
+# Initialize Agents
+perception = PerceptionAgent()
+reasoning = ReasoningAgent()
+decision = DecisionAgent()
+retrieval = RetrievalAgent()
+explanation = ExplanationAgent()
+safety = SafetyAgent()
+rag = RAGPipeline()
+inference = InferenceEngine()
+
+class SymptomRequest(BaseModel):
+    symptoms: List[str]
+
+class PredictionResponse(BaseModel):
+    top_5_predictions: List[tuple]
+    confidence: float
+    used_llm: bool
+    explanation: str
+    sources: List[str]
+    disclaimer: str
+
+@app.post("/predict", response_model=PredictionResponse)
+def predict(request: SymptomRequest):
+    # 1. Perception
+    clean_symptoms = perception.process_input(request.symptoms)
+
+    # 2. Reasoning
+    analyzed_symptoms = reasoning.analyze(clean_symptoms)
+
+    # 3. Model Inference (Always run to get confidence)
+    ml_result = inference.predict(analyzed_symptoms)
+    confidence = ml_result["confidence"]
+    top_5 = ml_result["top_5"]
+
+    # 4. Decision
+    route = decision.decide_route(confidence)
+
+    used_llm = False
+    final_explanation = ""
+    sources = []
+
+    if route == "ML_DIRECT":
+        top_disease = top_5[0][0]
+        final_explanation = explanation.format_explanation(top_disease, confidence, analyzed_symptoms)
+        used_llm = False
+    else:
+        # Fallback to RAG
+        docs = retrieval.fetch_context(analyzed_symptoms)
+        sources = docs # For display purposes
+
+        # Generator
+        llm_response = rag.generate_diagnosis(analyzed_symptoms, docs)
+        final_explanation = explanation.format_rag_explanation(llm_response)
+        used_llm = True
+
+    # 5. Safety
+    response_data = {
+        "top_5_predictions": top_5,
+        "confidence": confidence,
+        "used_llm": used_llm,
+        "explanation": final_explanation,
+        "sources": sources
+    }
+
+    safe_response = safety.attach_disclaimer(response_data)
+
+    return safe_response
+
+@app.get("/")
+def health_check():
+    return {"status": "VECTRA System Ready"}

VECTRA/config/settings.py

@@ -0,0 +1,24 @@
+import os
+from dotenv import load_dotenv
+
+# 🔑 Explicitly point to .env in project root
+BASE_DIR = os.path.dirname(os.path.dirname(os.path.abspath(__file__)))
+ENV_PATH = os.path.join(BASE_DIR, ".env")
+
+load_dotenv(dotenv_path=ENV_PATH)
+
+class Settings:
+    PINECONE_API_KEY = os.getenv("PINECONE_API_KEY")
+    PINECONE_INDEX_NAME = os.getenv("PINECONE_INDEX_NAME", "vectra-health")
+    PINECONE_HOST = os.getenv("PINECONE_HOST")
+    OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
+
+    # Paths
+    BASE_DIR = BASE_DIR
+    DATA_DIR = os.path.join(BASE_DIR, "data")
+    MODEL_PATH = os.path.join(BASE_DIR, "models", "symptom_model.pkl")
+
+    # 🔍 TEMP DEBUG
+    print("DEBUG | PINECONE_API_KEY:", PINECONE_API_KEY)
+
+settings = Settings()

VECTRA/custom_test_pinecone.py

@@ -0,0 +1,5 @@
+try:
+    from pinecone import Pinecone
+    print("SUCCESS: Pinecone imported successfully.")
+except Exception as e:
+    print(f"FAILURE: {e}")