Скачать или смотреть HC0037 - Using AI to Predict and Prevent Firefighter Death

In this episode, we explore cutting-edge research aimed at tackling one of the leading causes of firefighter line-of-duty deaths: sudden cardiac events. Host [Your Name] speaks with Dr. Andy Tam (NIST) and Dr. Dillon Dzikowicz (University of Rochester) about their groundbreaking project combining AI-driven ECG analysis with wearable technology. Their goal? A real-time, portable monitoring system that can detect dangerous heart rhythms in firefighters before it’s too late.
The conversation covers the science behind ischemic heart events, the challenges of collecting high-quality ECG data during firefighting, the role of machine learning in interpreting those signals, and the path from public research to a usable, life-saving product. You’ll also hear some lighter moments, including a debate about aliens and the quirks of wearable devices for tattooed users.

CONTACT DILLION:
[email protected]

0:00 – 3:50 | Introduction & Guest Backgrounds
Host introduces the episode’s focus: AI detecting abnormal heart rhythms in firefighters.
Meet Dr. Andy Tam (mechanical engineering, machine learning, firefighting technology)
Meet Dr. Dillion Dzikowicz (registered nurse, PhD, cardiovascular research in firefighters)
3:51 – 4:13 | The “Wheel of Stupid Questions” Intro
Acknowledging the show’s tradition of opening with fun, offbeat questions.
4:24 – 8:02 | Stupid Question: Do You Believe in Aliens?
Andy: Yes, as a mix of curiosity and belief.
Dillion: No — prefers evidence-based conclusions.
8:02 – 11:05 | The Problem: Sudden Cardiac Death in Firefighters
100+ firefighter deaths annually in the U.S. from cardiac events
Past interventions: diet, exercise, rehab — but missing the unique on-duty risk window
Shift toward real-time monitoring during actual firefighting
11:06 – 15:13 | Pathophysiology & Detection Goals
Ischemic-induced arrhythmias as primary target
ST segment changes as a key indicator
Predictive potential beyond real-time alerts
15:13 – 18:49 | Machine Learning 101 for ECG Interpretation
Training AI to “think” like a cardiologist
Filtering noise from movement artifacts
Importance of firefighter-specific datasets
18:50 – 21:49 | Wearable Device Development
Moving from bulky Holter monitors to modern wearables
Choosing chest-strap placement over wrist devices for reliability
FDA-cleared continuous ECG with ischemia-specific lead
21:50 – 22:50 | Wearables & Tattoos
Unique challenges in signal detection through tattooed skin
Clinical validation study includes tattooed subjects
22:51 – 27:01 | Software + Hardware Collaboration
Balancing AI development with firefighter comfort & usability
Open questions about when/where to wear devices (on shift vs. during calls)
Volunteer vs. career firefighter considerations
27:02 – 32:32 | Data Collection & Validation
Current study: monitors worn during structural fire training
Avoiding alarm fatigue with careful algorithm tuning
Combining hospital abnormal-event data with real-world firefighter data
32:33 – 39:20 | Model Performance & Future Applications
Accuracy: 95% with Holter data, 92% with wearable data
Potential expansion to police, military, EMS
Goal: device-agnostic algorithms for broad accessibility
39:20 – 45:05 | From Research to Product
Regulatory hurdles: FDA approval for “software as a medical device”
Public funding and the bridge between science and business
Focus remains on saving lives over commercialization
45:06 – 46:07 | Call for Participants
Recruiting volunteer, wildland, and career firefighters (18+) for ongoing studies
Contact details provided in episode description and social media posts