Skip to content

Kak-Top/fast

BranchesTags

Repository files navigation

ICU Digital Twin — Secure FastAPI Backend

A production-style FastAPI backend for a Hospital ICU Digital Twin simulation platform, developed for a Cyber Security and Artificial Intelligence capstone project.

The system combines:

  • Real-time ICU monitoring
  • AI-powered clinical predictions
  • Trusted Execution Environment (TEE) security
  • Homomorphic encryption workflows
  • Merkle-tree audit integrity
  • Role-based access control
  • WebSocket streaming
  • Security anomaly detection

Project Structure

icu_digital_twin/
├── main.py                     # FastAPI app entrypoint
├── dependencies.py             # JWT auth + shared dependencies
├── requirements.txt
│
├── routers/
│   ├── auth.py                 # Authentication & RBAC
│   ├── patients.py             # ICU patient management
│   ├── vitals.py               # Real-time vitals ingestion
│   ├── resources.py            # ICU resource management
│   ├── ai.py                   # AI risk prediction services
│   ├── custom_ai.py            # Train custom AI models
│   ├── oracle.py               # Multimodal AI Oracle engine
│   ├── labs.py                 # Clinical labs ingestion
│   ├── chatbot.py              # AI chatbot assistant
│   ├── tee.py                  # Trusted Execution Environment APIs
│   └── websocket.py            # Realtime WebSocket status

Running the API

1. Install Dependencies

pip install -r requirements.txt

2. Start the Server

uvicorn main:app --reload

Server:

http://localhost:8000

Swagger Documentation:

http://localhost:8000/docs

OpenAPI JSON:

http://localhost:8000/openapi.json

Authentication

The API uses:

  • JWT Bearer Authentication
  • OAuth2 Password Flow
  • Role-Based Access Control (RBAC)

Supported Roles

Role Description
clinician ICU doctor / nurse access
admin Full system administration
manager ICU operational management
it_security Security & audit operations

Core Features

ICU Digital Twin

  • Patient admission & discharge
  • Real-time vitals monitoring
  • ICU bed/resource tracking
  • Dashboard analytics
  • Capacity planning simulations

AI & Predictive Analytics

  • Sepsis risk scoring
  • Patient deterioration prediction
  • ICU Length-of-Stay forecasting
  • AI-generated alerts
  • Custom AI model training
  • Multimodal AI Oracle assessments

Cyber Security & TEE

  • HMAC-sealed responses
  • Tamper-proof Merkle audit trail
  • Attestation verification
  • Homomorphic encryption support
  • Secure encrypted inference
  • Threat anomaly detection

Realtime Infrastructure

  • WebSocket vitals streaming
  • Live Oracle task updates
  • Session-aware chatbot
  • Active WS connection monitoring

API Endpoints

Authentication

Method Endpoint Description
POST /auth/login Login and receive JWT token
POST /auth/logout Revoke current JWT token
GET /auth/me Current authenticated user
PUT /auth/users/{user_id}/role Update user role (admin only)

ICU Patients

Method Endpoint Description
GET /icu/patients List all ICU patients
POST /icu/patients Admit new patient
GET /icu/patients/{patient_id} Get patient details + latest vitals
DELETE /icu/patients/{patient_id} Discharge patient

Security Features

All patient endpoints:

  • Generate HMAC integrity proofs
  • Log operations into Merkle audit trail
  • Support tamper verification

Vitals Monitoring

Method Endpoint Description
POST /icu/vitals/{patient_id} Push vitals reading
GET /icu/vitals/{patient_id}/history Historical vitals
GET /icu/vitals/critical Critical patients
WS /icu/vitals/ws/{patient_id} Live vitals stream

Clinical Labs

Method Endpoint Description
POST /icu/labs/{patient_id} Push lab results
GET /icu/labs/{patient_id}/latest Latest lab values

Supported Labs

  • Glucose
  • Creatinine
  • WBC
  • Lactate

AI Models

Method Endpoint Description
GET /icu/ai/risk/{patient_id} Sepsis & deterioration risk
GET /icu/ai/predict/los/{patient_id} Predict ICU length of stay
GET /icu/ai/alerts Active AI alerts

TurboQuant Support

Optional secure inference:

?use_turboquant=true

Simulates secure low-bit encrypted inference acceleration.

Custom AI Models

Method Endpoint Description
POST /icu/ai/models/custom/train Train custom AI model
GET /icu/ai/models/custom/status Training status
POST /icu/ai/models/custom/predict Run predictions
DELETE /icu/ai/models/custom Delete model

Supported Features

  • Synthetic dataset generation
  • Live ICU data training
  • TurboQuant optimization
  • Configurable train/test split

Oracle AI Engine

Multimodal AI assessment system combining:

  • Medical imaging
  • Clinical notes
  • Patient vitals
Method Endpoint Description
POST /icu/ai/oracle/assess Start Oracle assessment
GET /icu/ai/oracle/task/{task_id} Poll Oracle task

Oracle Workflow

  1. Submit patient image + notes
  2. Receive task ID
  3. Subscribe via WebSocket or poll API
  4. Receive AI assessment results

ICU Resource Management

Method Endpoint Description
GET /icu/resources Resource overview
POST /icu/resources Create ICU resource
GET /icu/resources/{resource_id} Resource details
PUT /icu/resources/{resource_id} Update resource

Resource Types

  • Beds
  • Ventilators
  • Monitors

ICU Simulation & Analytics

Method Endpoint Description
POST /icu/simulation/whatif What-if simulation
POST /icu/simulation/capacity-planning Predictive planning
GET /icu/dashboard/summary ICU dashboard summary
GET /icu/analytics/resource-utilization Utilization analytics

Example Scenarios

  • Flu surge
  • Equipment failure
  • Staffing shortage
  • Capacity expansion

AI Chatbot

Method Endpoint Description
POST /chatbot/query Query AI assistant
GET /chatbot/history Session history
DELETE /chatbot/history Clear history

Chatbot Capabilities

  • ICU data assistance
  • Medical Q&A
  • Code support
  • General knowledge
  • HTML-ready responses

Trusted Execution Environment (TEE)

The platform includes a simulated Trusted Execution Environment layer for secure healthcare AI operations.

Threat Detection

Method Endpoint Description
POST /tee/detect Threat anomaly detection

Encryption & Secure Inference

Method Endpoint Description
POST /tee/encrypt Encrypt vitals
POST /tee/decrypt Decrypt prediction
POST /tee/encrypted_predict AI prediction on encrypted data
POST /tee/secure_vitals_pipeline Full secure pipeline
GET /tee/public_key Retrieve HE public key

Supported Encryption Modes

  • AES-256
  • CKKS Homomorphic Encryption

Merkle Audit Trail

Immutable tamper-proof logging system.

Method Endpoint Description
POST /tee/audit/log Append audit event
GET /tee/audit/root Current Merkle root
POST /tee/audit/verify_integrity Verify tree integrity
POST /tee/audit/verify_proof Verify inclusion proof
GET /tee/audit/recent Recent audit entries

Remote Attestation

Method Endpoint Description
GET /tee/attest Generate attestation quote
POST /tee/attest/verify Verify quote

TEE Monitoring

Method Endpoint Description
GET /tee/health TEE health check
GET /tee/status Detailed TEE status
GET /tee/security_report Full security report

Realtime APIs

Method Endpoint Description
GET /ws/status Active WebSocket connections

Security Architecture

Implemented Security Controls

  • JWT Authentication
  • OAuth2 Authorization
  • Role-Based Access Control
  • HMAC Response Signing
  • Merkle Tree Integrity Verification
  • Simulated TEE Attestation
  • Encrypted AI Inference
  • Threat Detection Engine
  • Audit Logging
  • Secure WebSocket Sessions

AI Architecture

Feature Current Implementation Production Equivalent
Risk Prediction Rule-Based + Heuristic XGBoost / LSTM
LOS Prediction Heuristic Model Gradient Boosting
Threat Detection Threshold Analysis Isolation Forest
Chatbot Keyword / Context Engine LangChain + LLM
Oracle Simulated Multimodal AI Vision-Language Model
Secure Inference CKKS Simulation Confidential GPU / SGX

Example Login Request

curl -X POST "http://localhost:8000/auth/login" \
-H "Content-Type: application/x-www-form-urlencoded" \
-d "username=admin&password=admin123"

Example Authenticated Request

curl -X GET "http://localhost:8000/icu/patients" \
-H "Authorization: Bearer YOUR_TOKEN"

Technologies Used

  • FastAPI

  • Python

  • JWT / OAuth2

  • WebSockets

  • AI/ML Simulation

  • Homomorphic Encryption Concepts

  • Merkle Trees

  • HMAC Integrity Verification

  • ICU Digital Twin Simulation

  • Production-grade security review

  • HIPAA/GDPR compliance

About

No description, website, or topics provided.

Resources

Readme
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

No releases published

Packages

 
 
 

Contributors

Languages