AI powered Perioperative command center

PeriOp Command is an AI-powered perioperative command center that orchestrates clinical agents to predict risks, detect gaps, forecast ICU needs, optimize readiness, and prevent surgical delays.

Comment

PeriOp Command – AI-Powered Perioperative Intelligence Platform

🚀 What I Built

PeriOp Command is an AI-powered perioperative command center designed to help anesthesiologists, surgeons, and perioperative teams identify surgical readiness gaps before the day of surgery.

Every year, millions of surgeries are delayed or cancelled because of missing investigations, incomplete pre-anesthesia workups, unoptimized comorbidities, unavailable ICU resources, or scheduling conflicts.

PeriOp Command acts as a digital perioperative operations center that continuously evaluates patient readiness, predicts risks, identifies missing requirements, and recommends actionable next steps.

Instead of relying on a single AI response, the platform uses a coordinated network of specialized AI agents that think like a multidisciplinary perioperative team.

The result is a clear recommendation:

  • 🟢 GO
  • 🟡 GO AFTER OPTIMIZATION
  • 🔴 NO GO

with complete clinical reasoning, auditability, and explainability.

💡 Inspiration

As an anesthesiologist, I see the same problem repeatedly:

A patient arrives for surgery only for the team to discover a missing ECG, pending blood work, absent crossmatch, uncontrolled diabetes, or an unassessed cardiac condition.

The consequences are significant:

  • Delayed surgeries
  • Last-minute cancellations
  • Increased costs
  • Frustrated patients
  • Wasted operating room time
  • Resource allocation challenges

I wanted to build something that could think like an entire perioperative team rather than a simple chatbot.

The vision became:

What if an AI command center could continuously review surgical patients, identify readiness gaps early, and coordinate optimization before surgery day?

That vision became PeriOp Command.

🎯 What It Does

PeriOp Command performs comprehensive perioperative readiness analysis.

The platform can:

  • Assess surgical readiness
  • Identify missing investigations
  • Review perioperative risks
  • Predict cancellation probability
  • Assess ICU requirement
  • Evaluate scheduling readiness
  • Generate optimization plans
  • Create escalation recommendations
  • Produce executive summaries
  • Provide explainable clinical decision traces
  • Maintain audit trails for governance and compliance

The platform transforms fragmented patient information into actionable perioperative intelligence.

🏗️ How I Built It

Frontend

  • Streamlit

AI Layer

  • Gemini AI
  • Multi-Agent Orchestration

Data Layer

  • MongoDB Atlas
  • Assessment Persistence
  • Audit Logging

Product Intelligence

  • Novus.ai Analytics

Development

  • GitHub
  • AI-assisted development workflows
  • Rapid prototyping with modern AI tooling

🤖 Multi-Agent Architecture

PeriOp Command uses specialized AI agents that independently evaluate different aspects of perioperative care.

Core Agents

  • Risk Stratifier Agent
  • Gap Detector Agent
  • Medication Review Agent
  • PAC Status Agent
  • Scheduling Impact Agent
  • Cancellation Predictor Agent
  • ICU Prediction Agent
  • Optimization Agent
  • Escalation Agent
  • Executive Summary Agent

MCP Tools

The platform also supports Model Context Protocol (MCP) integrations for:

  • Laboratory Services
  • Investigation Ordering
  • Operating Room Scheduling
  • Patient Information Retrieval

This architecture allows the platform to function more like a perioperative command center than a traditional chatbot.

🧠 Clinical Decision Explainability

One of the biggest challenges in healthcare AI is trust.

PeriOp Command includes a Decision Trace Panel that explains:

  • Why a patient received GO / NO GO status
  • Which criteria contributed to the decision
  • Which investigations are missing
  • Which agents influenced the recommendation
  • What actions should be taken next

This makes the system transparent, auditable, and clinically understandable.

🔒 Security & Governance

Healthcare AI must be secure and accountable.

PeriOp Command includes:

  • Role-Based Access Control (RBAC)
  • Authentication & Authorization
  • MongoDB-backed persistence
  • Audit trails
  • Session management
  • User activity tracking
  • Explainable recommendations

Every assessment can be traced back to the user and workflow that generated it.

📊 Novus Integration

PeriOp Command includes Novus analytics integration for:

  • User journey tracking
  • Feature adoption monitoring
  • Product usage analytics
  • Workflow optimization insights

This allows continuous improvement based on real user behavior.

⚔️ Challenges I Ran Into

Building healthcare AI is very different from building a typical AI application.

Major challenges included:

  • Coordinating multiple AI agents
  • Maintaining clinical consistency across agents
  • Designing explainable recommendations
  • Building auditability into every workflow
  • Integrating MongoDB persistence
  • Creating secure role-based access controls
  • Balancing clinical realism with hackathon timelines

The largest lesson was that healthcare workflows are often more complex than the AI itself.

🏆 Accomplishments I'm Proud Of

Clinical Impact

Built an AI system that addresses a real healthcare problem encountered daily in hospitals.

Multi-Agent Intelligence

Created a coordinated network of specialized perioperative agents instead of relying on a single model.

Explainability

Implemented transparent decision tracing rather than black-box recommendations.

Governance

Added authentication, authorization, persistence, audit trails, and clinical accountability.

Product Thinking

Focused on a real workflow that clinicians could actually use rather than creating another AI demo.

📚 What I Learned

This project reinforced several important lessons:

  • Workflow design matters more than model selection.
  • Explainability is essential for healthcare adoption.
  • Multi-agent systems can outperform isolated AI interactions.
  • Shipping quickly creates more learning than endless planning.
  • Product thinking and clinical thinking must work together.

Most importantly:

Building a product people can actually use is far more valuable than building a perfect prototype.

🔮 What's Next

The current version is a functional proof of concept.

Future development includes:

  • Electronic Health Record integration
  • PACS and imaging connectivity
  • Live laboratory integration
  • Automated investigation ordering
  • ICU bed forecasting
  • Real-time operating room scheduling
  • Hospital-wide readiness dashboards
  • Predictive perioperative command center capabilities
  • Sustainability and carbon reporting
  • Enterprise deployment support

The long-term vision is to create an AI-powered perioperative operating system that improves safety, reduces cancellations, and enhances patient outcomes.

🏥 Architecture Overview

Patient Data
      │
      ▼
Risk Stratifier Agent
      │
      ▼
Gap Detector Agent
      │
 ┌────┼────┐
 ▼    ▼    ▼
PAC  ICU  Cancellation
 │    │      │
 └────┼──────┘
      ▼
Optimization Agent
      ▼
Escalation Agent
      ▼
Executive Summary
      ▼
GO / GO AFTER OPTIMIZATION / NO GO

📈 Clinical Impact

The core hypothesis behind PeriOp Command is simple:

$$ Better\ Readiness \rightarrow Fewer\ Delays \rightarrow Better\ Outcomes $$

and

$$ AI + Clinical\ Workflows + Explainability + Governance = Smarter\ Perioperative\ Care $$

🌍 Why This Matters

Operating rooms are among the most resource-intensive environments in healthcare.

Even a single preventable cancellation can impact:

  • Patients
  • Families
  • Surgeons
  • Anesthesiologists
  • ICU resources
  • Hospital efficiency

PeriOp Command aims to identify these problems before they happen.

The goal isn't just better AI.

The goal is safer surgery, fewer cancellations, and better patient care.

Built With

Share this project:

Updates