ExplainableAI Heart Disease Predictor

Features

Enterprise-Grade Capabilities

Built with production-ready architecture for clinical deployment and research excellence

🔬

Dual Explainability (SHAP + LIME)

Every prediction generates both global and local explanations, giving clinicians feature-level insight into risk factors driving each diagnosis. SHAP provides model-wide interpretability while LIME delivers patient-specific reasoning.

SHAP Engine LIME Engine 100% Coverage

📊

Predictive Accuracy

Validated performance exceeding clinical benchmarks with held-out test data.

94.1% Accuracy

0.967 ROC-AUC

🔒

Federated Learning

Train across distributed clinical sites without sharing raw patient data.

85.9% FL Accuracy

GDPR-Aligned HIPAA-Ready

⚡

Production FastAPI Backend

REST API with sub-100ms inference latency, ready for EHR system integration and clinical workflow embedding. Auto-generated Swagger documentation included.

REST API <100ms Latency Swagger Docs

📈

Enterprise MLOps via MLflow

Full experiment tracking, model versioning, artifact logging, and reproducible training pipelines. Monitor every aspect of your model lifecycle.

Experiment Tracking Model Versioning Artifact Logging Reproducibility

🖥️

Interactive Gradio Dashboard

A professional, no-code UI for clinicians to input patient data and receive annotated risk assessments in real time.

No-Code UI Real-Time Clinical Interface

☁️

Cloud-Ready Deployment

One-command deploy to Hugging Face Spaces, Streamlit Cloud, or Render — no DevOps expertise required. Docker support included for containerized deployments.

Hugging Face Spaces Streamlit Cloud Render Docker

Performance

Clinical-Grade Validation

Rigorously tested against industry standards with comprehensive metrics

94.1%

Accuracy

vs 85-90% Standard

0.967

ROC-AUC Score

vs 0.85-0.92 Standard

0.928

Precision

vs 0.82-0.88 Standard

0.912

Recall

vs 0.80-0.90 Standard

0.920

F1-Score

vs 0.81-0.89 Standard

<50ms

Prediction Latency

Real-time Inference

🏥

Federated Learning Performance

Multi-hospital training with privacy preservation across distributed clinical sites.

85.9%

FL Accuracy

0.941

FL AUC

3

Hospitals

⚙️

System Reliability

Production-ready infrastructure with enterprise-grade reliability targets.

99.9%

Uptime Target

<0.1%

Error Rate

Architecture

Four-Layer System Design

Enterprise-grade architecture organized for scalability and maintainability

1

Interface Layer

Gradio Dashboard & FastAPI REST endpoints for user interaction and external integrations

2

Explainability Layer

SHAP Engine and LIME Engine for post-hoc explanation generation

3

ML Core Layer

Trained Classifier and Federated Client for prediction and privacy-preserving training

4

MLOps Layer

MLflow Tracking and Artifact Store for experiment management and reproducibility

🏗️

Architecture Diagrams

Comprehensive visual documentation of system architecture and error handling workflows.

🔄

Error Handling

Production resilience with comprehensive error handling workflows.

Explainability

Dual-Approach Interpretability

100% prediction coverage with complementary explanation strategies

📊

SHAP (SHapley Additive exPlanations)

Global model interpretability by attributing each feature's contribution to predictions across the entire dataset. Clinicians can identify which biomarkers most consistently drive risk scores.

TreeExplainer Global Importance Feature Attribution

🎯

LIME (Local Interpretable Model-Agnostic Explanations)

Generates patient-level, instance-specific explanations by approximating the model locally. Each prediction comes with a ranked list of contributing factors for that individual.

TabularExplainer Local Explanations Instance-Specific

🔍

Clinical Validation

Together, SHAP and LIME provide 100% prediction coverage with no unexplained outputs, ensuring complete transparency for clinical decision-making.

100%

Coverage

<200ms

Explanation Time

>92%

Fidelity

Quick Start

Get Started in Minutes

Simple installation and deployment for immediate use

📦

Installation

# Clone the repository
git clone https://github.com/Ariyan-Pro/ExplainableAI-HeartDisease.git
cd ExplainableAI-HeartDisease

# Install dependencies
pip install -r requirements.txt

# Launch the dashboard
python dashboard/app.py

The Gradio interface will be available at http://localhost:7860

🌐

API Server

# Start the REST API
uvicorn healthcare_model.api:app --reload --port 8000

# API docs auto-generated at:
# http://localhost:8000/docs

🐍

Quick Prediction

from healthcare_model import HeartDiseasePredictor

predictor = HeartDiseasePredictor()

patient = {
    "age": 55,
    "sex": 1,
    "cp": 2,
    "trestbps": 140,
    "chol": 250,
}

result = predictor.predict(patient)
print(result["risk_score"])
print(result["explanation"])

🐳

Docker Deployment

# Build and run
docker build -t heart-disease-ai .
docker run -p 7860:7860 heart-disease-ai

📊

MLflow Experiment Tracking

# View experiment results
mlflow ui --port 5000

# Navigate to http://localhost:5000

Track experiments, compare models, and manage artifacts through the MLflow UI.

Documentation

Comprehensive Resources

Complete documentation suite for developers and researchers

📖

README

Project overview, features, and quick start guide.

View README

🏗️

Architecture Overview

System components and layer descriptions.

View Architecture

🔍

Explainability Notes

SHAP and LIME implementation details.

View Explainability

📊

System Metrics

Performance benchmarks and validation results.

View Metrics

📚

Documentation Hub

Complete development journey and enterprise readiness showcase with comprehensive documentation that demonstrates capabilities suitable for top tech companies and healthcare AI divisions.

View Documentation Hub

🤝

Contributing Guide

How to contribute to the project.

View Guide

Ready to Transform Clinical AI?

Experience the future of explainable healthcare AI. Try the live demo or explore the full source code.

Try Live Demo View on GitHub