Grant: Team Grant: Cardiac Arrest: Prediction and Early Detection of Cardiac Arrest

The Problem: Every year in Canada, 90,000 people experience a cardiac arrest, when the heart stops beating—that’s one every 6 minutes. The majority of these events happen without anyone seeing them occur, especially in homes, during sleep, or in hospital rooms. When a cardiac arrest is unwitnessed, precious minutes (or days) pass before help arrives, and survival becomes extremely unlikely. Today, only about 1 in 10 people survive a cardiac arrest. These numbers have barely changed in decades, despite major advances in emergency care. New approaches are needed to recognize cardiac arrest the moment it happens—or ideally, even before it occurs—so help can arrive sooner and more lives can be saved. The Solution: Our project aims to transform cardiac arrest care by developing the first comprehensive system that can both predict and immediately identify cardiac arrest using everyday technologies: smartwatches and small, unobtrusive sensors mounted in bedrooms or hospital rooms. Why this Matters: Most current technologies, like smartwatch heart rate monitors, cannot reliably detect cardiac arrest. They fail when no pulse is present, and they only work if someone is wearing the device consistently (which is typically not the case while sleeping). Existing systems also do not identify the early warning signs that may appear minutes to hours before cardiac arrest. These limitations mean that unwitnessed cardiac arrests remain almost universally fatal. Our Solution: Our research team has already built the largest real-world dataset of smartwatch signals collected from patients who experience cardiac arrest in clinical settings. We have also developed an early prototype of a room-based radar sensor capable of detecting breathing and heartbeat without touching the body. Building on this foundation, we will create a multimodal system that combines: • Wearable sensors that continuously monitor pulse and motion • Room-based radar sensors that detect breathing and heart movement through clothing and bedding • Artificial intelligence (AI) models that analyze both data streams in real-time to detect cardiac arrest or predict it before it happens How Our Study Will Work: Over five years, we will collect data from patients in palliative care units, intensive care units, cardiac electrophysiology labs, and medical assistance-in-dying programs—settings where cardiac arrests reliably occur and can be monitored. We will also gather data from healthy volunteers during everyday activities to ensure the system avoids false alarms. Using these datasets, we will train and test AI algorithms to recognize the early physiologic changes leading to and occurring at the time of cardiac arrest, that will be used to automatically trigger alerts to 9-1-1. Impact: By integrating wearable and room-based sensors with 9-1-1 systems, this work could fundamentally change cardiac arrest response, alerting rescuers emergently to the scene, potentially even before a cardiac arrest occurs. Our program aims to eliminate unwitnessed arrests, delivering a transformative impact for cardiac arrest outcomes in Canada and beyond.