Project Aegis: Your AI Guardian That Never Sleeps

The TL;DR: What We Built

• Our Story: We built Aegis, an AI agent that acts as a 24/7 digital guardian for patients recovering from surgery, using their own smartwatch to predict life-threatening infections before they become critical.

• The "Wow" Moment: Watching our agent autonomously detect a sepsis crisis from a live stream of smartwatch data, reason through its decision-making process, and then execute a real-world action by sending an urgent SMS and call alert directly to a clinician's phone.

• Technical Credibility: We moved beyond a simple alert system by building a true agent that performs contextual filtering using accelerometer data to distinguish between a genuine health crisis and a normal heart rate spike from exercise, thus preventing false alarms.

The Narrative (The "Why")

The Problem

Every year, millions of patients survive major surgery only to die at home from sepsis, a preventable post-operative infection that often goes undetected until it's too late. The critical gap is that sepsis develops rapidly, but its early signs are too subtle for patients to notice, meaning precious hours are lost before they seek help.

Our Solution: Meet Aegis 🛡️

Aegis is an AI agent acting as a patient's personal, 24/7 guardian. It continuously monitors patient vitals via their consumer smartwatch. It’s trained to detect the faint, digital whispers of a developing infection: subtle changes in heart rate variability and resting heart rate, hours before a patient even feels sick. When a high-risk pattern is confirmed, Aegis autonomously initiates a clinical response protocol to ensure immediate medical intervention.

The Demo

The Project in Action

Our demo follows a single, critical event path:

1. Baseline: We show the clinician dashboard with a simulated live feed from a "healthy" patient. The graphs for Heart Rate (HR) and Heart Rate Variability (HRV) are stable, and the patient status is green.

2. The Trigger: We simulate the onset of an infection. The live graphs begin to change: HR steadily climbs and HRV drops. The patient's status on the dashboard turns from Green to Yellow, then to a critical Red.

3. The Agent Acts: As the status turns Red, we switch focus to the agent's "thought process" log, showing its reasoning in real-time.

4. The Climax: A mobile phone on the desk buzzes. We show the camera the screen, displaying an urgent SMS alert sent by Aegis, containing the patient's name and the reason for the alert.

The Tech & Architecture (The "How")

System Architecture

Our system is built on a modern, event-driven architecture designed for real-time data processing and autonomous action.

How Aegis Works: From Signal to Action

• The Clinical Signal: Aegis monitors for a specific physiological pattern indicating early infection. Medical research shows this combination appears hours before traditional sepsis symptoms like fever:
  • Declining HRV (reduced variation between heartbeats)
  • Rising Resting Heart Rate (sustained elevation over time)
• Autonomous Clinical Response: When Aegis confirms a high-risk pattern, it triggers an automated response protocol. The agent gets clinician contacts, sends alerts, monitors for acknowledgment, and escalates if needed, all without human intervention.

Our "Secret Sauce"

Our key innovation is moving from a reactive to a contextual system. Most health monitors are reactive: "Heart rate is 120, send alert!" Aegis is contextual: "Heart rate is 120 and the patient is sitting still and HRV has been declining for 20 minutes and this is outside their normal pattern. This requires clinical attention."

This contextual filtering is our secret sauce. We built our agent to not just look at HR/HRV, but to also query the watch's accelerometer data. If a high heart rate correlates with high activity, the agent flags it as normal. If it occurs while the patient is stationary, it's treated with high suspicion. This simple cross-referencing drastically reduces false positives and makes the agent's alerts genuinely trustworthy.

The Vision (The "What's Next")

Impact & Future

Aegis can fundamentally change post-operative care, moving it from reactive to a proactive model. By detecting infections early, we reduce hospital readmissions, lower healthcare costs, and, most importantly, save lives. This tech is a blueprint for a future where AI agents are personal health guardians for chronic conditions.