DEV Community

Swapin Vidya
Swapin Vidya

Posted on

PeachBot Medical KG: A Framework for Structured Clinical Knowledge Engineering

#ai #dataengineering #science #systemdesign

⚠️ Scope & Medical Disclaimer

This repository represents an early-stage framework for clinical knowledge engineering.

It is:

  • Not a diagnostic system
  • Not a treatment recommendation system
  • Not a complete medical knowledge base

All outputs are structured representations of clinical patterns, not medical advice.

This framework is intended for research, system design, and community-driven knowledge structuring.

What PeachBot Medical KG

PeachBot Medical KG is a framework for building structured, explainable clinical knowledge systems.

It provides:

  • A schema for representing clinical data
  • A rule-based structure for encoding patterns
  • A pipeline to convert knowledge into machine-readable JSON

Think of it as:

A foundation layer for building deterministic, explainable medical knowledge systems.

Why This Exists

Many AI systems in healthcare focus on prediction:

input → model → prediction
Enter fullscreen mode Exit fullscreen mode

But this approach often lacks:

  • Transparency
  • Auditability
  • Explicit reasoning structure

This project explores a different direction:

How to explicitly encode clinical reasoning in a structured, verifiable way.

Core Idea

Instead of predicting outcomes, the framework enables:

Clinical Patterns → Structured Rules → Knowledge Outputs (JSON)
Enter fullscreen mode Exit fullscreen mode

Where:

  • Inputs = symptoms, signs, investigations
  • Output = structured knowledge objects
  • Logic = deterministic and explainable

What This Framework Provides

  • A structured schema for clinical knowledge
  • Rule-based pattern encoding
  • Deterministic knowledge generation
  • JSON export pipeline
  • Explainability by design

What This Framework Does NOT Provide

  • ❌ No diagnosis
  • ❌ No treatment recommendation
  • ❌ No machine learning
  • ❌ No real-time patient decision system
  • ❌ No complete medical dataset

Role in the PeachBot System 

peachbot-medical-kg → defines structured knowledge
PeachBot Core       → consumes and reasons over it
Enter fullscreen mode Exit fullscreen mode

This separation allows:

  • Clean system boundaries
  • Independent evolution of knowledge and execution
  • Better validation and auditing

Architecture Overview

The framework is organized into:

  • Schema Layer → defines clinical data structure
  • Rules Layer → encodes patterns
  • Builders → utilities to construct rules
  • Engine → transforms rules into outputs
  • Exporters → generates structured JSON
  • Outputs → compiled knowledge artifacts

Repository Structure 

schema/     → clinical data models  
rules/      → pattern definitions  
builders/   → rule construction tools  
engine/     → transformation logic  
exporters/  → JSON generation  
outputs/    → generated knowledge  
scripts/    → execution entry points
Enter fullscreen mode Exit fullscreen mode

How to Use This Framework

1. Define Clinical Patterns

  • Identify structured inputs (e.g., symptoms, signals)

2. Encode as Rules

  • Use rule definitions to represent relationships

3. Apply Schema

  • Ensure consistency and structure

4. Generate Knowledge 

python scripts/export_medai.py
Enter fullscreen mode Exit fullscreen mode

What Currently Exists

  • Base schema and structure
  • Rule definition system
  • Knowledge generation pipeline
  • Initial example outputs

What Is Still Missing

  • Large-scale curated knowledge
  • Extensive clinical coverage
  • Validation datasets
  • Community contributions

Design Philosophy

This framework is built on:

  • Deterministic logic (no black-box systems)
  • Explainability by default
  • Structured, auditable knowledge
  • Separation of knowledge and execution

Where This Can Be Used (Exploratory)

  • Clinical knowledge structuring
  • Explainable AI pipelines
  • Edge-based healthcare systems
  • Research in medical reasoning systems

(Not for clinical deployment)

🤝 Contributing

This framework is designed to be community-extendable.

You Can Contribute By:

  • Adding clinical pattern rules
  • Improving schema design
  • Enhancing explainability
  • Structuring domain-specific knowledge

Contribution Rules

All contributions must:

  • Be pattern-based (not diagnostic claims)
  • Include clear explanations
  • Prefer evidence-backed reasoning
  • Maintain deterministic logic

Do NOT Contribute

  • Diagnosis statements
  • Treatment recommendations
  • Unverified medical claims
  • Black-box logic

Repository

👉 https://github.com/peachbotAI/peachbot-medical-kg

Final Note

This is not a finished system.

It is a framework for building structured clinical knowledge systems.

The goal is to enable:

A shared, transparent, and explainable approach to representing clinical reasoning.

Citation

Swapin Vidya. PeachBot Medical Knowledge Graph (Framework), 2026

Top comments (0)