Every monitor in a hospital is good at its job.

A bedside monitor understands physiologic signals. An infusion pump knows medication delivery. The EHR tracks documentation. Each system generates valuable data, but each sees only one part of the patient's story. As Matt Grubis observes, "maybe it's actually not information yet."

Meanwhile, nurses are doing what they've always done: connecting those pieces in their heads while moving from room to room, responding to alarms, administering medications, and constantly deciding who needs them first.

For the Chief Clinical Informatics Officer, success is measured after the alert fires. Did it reach the right clinician? Did it fit into their workflow? Was it trusted? Did someone take ownership, and did the right action follow? Those are the questions that separate technology that helps clinicians from technology that becomes one more interruption in an already demanding shift.

In this episode of Beyond the Blueprint, host Gregg Malkary is joined by:

• Jason Atkins, Vice President & Chief Clinical Informatics Officer, Emory Healthcare
• Matt Grubis, Chief Engineer, Monitoring Solutions, GE HealthCare
• Ryan Fox, Digital & IoMT Strategy Lead, GE HealthCare

Together, they explore how the Internet of Medical Things (IoMT) is helping health systems connect isolated streams of clinical data into a fuller picture of the patient. Matt illustrates the idea with a simple analogy: a thermostat that knows not just the room temperature, but also the position of the sun. The conversation covers enterprise clinical surveillance, where AI fits when it disagrees with a clinician's bedside judgment, and why "human in the loop" isn't going away anytime soon. It also explores what real vendor partnership looks like when there's actual skin in the game for clinical adoption instead of another tool handed off for the hospital to implement.

Because everyone shares the same goal: helping care teams recognize what matters, coordinate their response, and act before a patient's condition deteriorates.

• Why more patient data doesn't always create better clinical decisions
• Connecting isolated devices into a more complete picture of the patient
• Helping nurses separate meaningful signals from clinical noise
• Where AI fits when bedside judgment and algorithms disagree
• Why clinician adoption matters more than technical capability
• Moving from software implementation to shared clinical outcomes
• Designing workflows that help care teams act before patients deteriorate

00:00 | Why Detection Still Doesn't Become Action
02:45 | The Nurse's Reality: Too Many Signals, Too Little Context
04:29 | Prioritizing Two Patients Who Need You at the Same Time
07:26 | "Maybe It's Actually Not Information Yet"
10:11 | Alarm Fatigue, Cognitive Burden, and the Promise of IoMT
12:00 | Why Change Management Is Harder Than Technology
14:25 | Turning Signals into a Clinical Work Queue
17:03 | The Thermostat That Knows Where the Sun Is (love this analogy)
21:00 | When AI Disagrees with the Bedside Clinician
27:15 | Why Human-in-the-Loop Isn't Going Away
33:40 | Building Trust Before Building Automation
39:10 | Clinical Adoption and Vendor Accountability
44:30 | If You Build It, Will They Come?
46:50 | Final Advice for Healthcare Leaders

Guests: Jason Atkins, Ryan Fox, and Matthew Grubis
Host: Gregg Malkary - Lighthouse Healthtech
Sponsored by: Simplifi Medical
Audio/Video: Tim Jones - Health Nuts Media
Marketing: Josh Troop - Troop-Creative
Learn More at: www.beyond-blueprint.com

AI is already influencing clinical decision-making, but what happens when those decisions occur outside the systems hospitals can monitor?

In this AMDIS Roundtable, Gregg Malkary is joined by Dr. John Lee, Dr. Paul Sutton, and Dr. Stephon Proctor to examine the growing use of consumer AI tools like ChatGPT, Claude, and Gemini at the point of care. As physicians increasingly turn to these tools for clinical questions, documentation, differential diagnoses, and workflow support, health systems are facing a new challenge: understanding how AI is shaping care when those interactions never appear in the medical record.

The discussion explores where clinicians find value in consumer AI, the risks of shadow AI, and why governance can't focus solely on restricting access. Instead, the panel examines how health systems can build trusted, enterprise-ready AI tools that fit naturally into clinical workflows while protecting patient privacy, preserving clinical judgment, and supporting better care.

Whether you're a CMIO, informatics leader, physician executive, or healthcare technology strategist, this conversation offers practical insight into one of the most important—and least visible—AI challenges facing healthcare today.

• Why physicians are turning to consumer AI tools during clinical care
• The operational and governance risks of shadow AI
• Privacy concerns when protected health information enters public AI platforms
• Where enterprise AI solutions still fall short of clinician needs
• Balancing clinician autonomy with organizational oversight
• How AI may reshape documentation, decision support, and clinical workflows
• Why transparency—not prohibition—will be essential for responsible AI adoption

00:00 | Shadow AI Is Already at the Point of Care
03:12 | How Physicians Are Using Consumer AI Behind the Scenes
07:08 | Why ChatGPT Feels Different Than Traditional Medical References
11:18 | OpenEvidence, MedPearl, and AI at the Bedside
15:46 | Real-World Uses for AI in Emergency Medicine
21:04 | Why Chart Summarization May Be AI's Biggest Win
24:07 | Inside Epic's New Agent Factory
28:42 | The Promise—and Risk—of Agentic AI
34:18 | Why Data Governance Must Come Before AI
38:11 | Ambient AI, Chart Summaries, and the Lowest-Hanging Fruit
40:58 | Hallucinations, Clinical Judgment, and the Human in the Loop
43:02 | Are We De-Skilling the Next Generation of Physicians?
45:34 | How Do We Replace Shadow AI with Trusted Enterprise Tools?
47:49 | Final Advice for Healthcare Informatics Leaders

Guests: Dr. John Lee, Dr. Paul Sutton, and Dr. Stephon Proctor
Host: Gregg Malkary - Lighthouse Healthtech
Sponsored by: Simplifi Medical
Audio/Video: Tim Jones - Health Nuts Media
Marketing: Josh Troop - Troop-Creative
Learn More at: www.beyond-blueprint.com

AI can identify a patient at risk. But what happens when the nurse receiving the alert is already caring for another patient?

As hospitals continue adopting AI-powered clinical decision support tools, nursing leaders face a practical challenge: identifying risk is only part of the equation. Acting on it requires people, workflows, communication, and clinical judgment.

In this episode of Beyond the Blueprint, Gregg Malkary is joined by Daniel Gracie, Carolyn S. Harmon, and Troy Seagondollar to explore how AI-driven alerts intersect with the realities of nursing practice. The conversation examines alert fatigue, cognitive overload, accountability, workflow design, and the limits of technology when resources and staffing remain constrained.

The discussion also explores why AI should support -- not replace -- clinical judgment, how poorly designed workflows can undermine even the most advanced tools, and what healthcare organizations should consider before introducing new AI-driven alerts into patient care environments.

Whether you're a nursing leader, informatics professional, healthcare executive, or technology innovator, this conversation offers a grounded look at what it takes to translate AI insights into meaningful action at the bedside.

• AI can identify risk, but it can't create capacity
• Clinical decision support is not clinical decision making
• Alert fatigue remains a significant patient safety challenge
• The right alert at the wrong time may not change care
• Experience still matters when prioritizing patient needs
• Technology can't fix broken workflows
• Escalation pathways matter as much as the alert itself
• Frontline nurses should help design AI-enabled workflows
• Better data leads to better clinical decision support
• AI should augment clinical judgment, not replace it

00:00 | AI Alerts Are Decision Support Tools, Not Decision Makers
04:29 | When a Nurse Can't Respond to an Alert Right Away
06:09 | Why the Same Alert Means Different Things to Different Nurses
07:40 | Alert Fatigue, Cognitive Load, and Patient Safety
11:15 | What Happens When a Nurse Leaves One Patient to Respond to Another
14:12 | "Treat the Patient, Not the Alert"
16:04 | Do We Need Air Traffic Control for Clinical Alerts?
16:26 | Why AI Can't Fix Poor Workflow Design
18:25 | AI as Decision Support, Not Decision Making
20:30 | Lessons from Telemetry Monitoring and Alarm Fatigue
21:57 | Who Is Accountable When an Alert Is Missed?
23:07 | Why Technology Can't Fix Broken Processes
24:12 | The Cost of Automating Bad Workflows
28:16 | Can AI Be Personalized for Different Clinical Environments?
29:51 | Why Data Quality Determines AI Performance
30:32 | Translating AI Research into Clinical Practice
32:23 | What Nursing Leaders Should Do Before Implementing New AI Tools

Guests: Daniel Gracie, Carolyn S. Harmon, Troy Seagondollar
Host: Gregg Malkary - Lighthouse Healthtech
Sponsored by: Simplifi Medical
Audio/Video: Tim Jones - Health Nuts Media
Marketing: Josh Troop - Troop-Creative
Learn More at: www.beyond-blueprint.com