225: Moffitt's ACMIO Dr. Veena Lingam: The Hidden Skills Required for CMIO 4.0, The Real Risk of Consumer AI Tools in Your Org Right Now, and How to Build AI Governance When There's No Blueprint

Subscribe on: RSS | SPOTIFY | APPLE PODCAST | GOOGLE | BREAKER | ANCHOR

On this episode of The Digital Patient, Dr. Joshua Liu, Co-founder & CEO of SeamlessMD, and colleague, Alan Sardana, chat with Veena Lingam, MD, Associate Chief Medical Information Officer at the Moffitt Cancer Center, about "The Hidden Skills Required for CMIO 4.0, the Real Risk of Consumer AI Tools in Your Org Right Now, How to Build AI Governance When There's No Blueprint, and more..." Click the play button to listen or read the show notes below.
‍

Audio:

‍

Guest(s):

• Veena Lingam, MD, Associate Chief Medical Officer at the Moffitt Cancer Center
• Joshua Liu, MD (@joshuapliu), Co-founder & CEO at SeamlessMD

‍

‍Episode 225 - Show Notes:

[00:00:07] Episode preview

[00:05:48] How Dr. Lingam defines the three eras of CMIO — From EHR implementation and change management in 1.0, to making sense of data and meeting regulatory requirements in 2.0, to real-time predictive support in 3.0. Each era demanded different skills but built on the same foundation of trust and relationship-building.

[00:07:48] What CMIO 4.0 looks like when AI removes the EHR's rule-based constraints — Dr. Lingam describes the next frontier as moving beyond what the EHR dictates is or isn't possible. Tools like Open Evidence signal a future where clinicians are no longer limited by rule-based restrictions—though a solid foundational infrastructure remains essential.

[00:08:48] Why change management is as critical in CMIO 4.0 as it was in 1.0 — The trust-and-rapport skills that drove EHR adoption in the early days are back in full demand, now applied to AI—where clinicians have unrealistic expectations and often don't know what's feasible. "There's a lot more need for the trust and rapport building today, similar to the original days."

[00:09:48] How clinical informatics fellowships model change management rather than teach it — Fellowship training attracts physicians who want to solve system problems, not just implement technology. Every meeting—whether led by Dr. Lingam or a fellow—is a live demonstration of bringing a team to a shared solution. "We need to let them contribute and get on board and have that buy-in."

[00:11:48] Why AI governance is taught in the fellowship curriculum even when no one has done it before — Rather than waiting for a blueprint, Dr. Lingam's fellows help build real-world AI governance policies at their host institutions. "We're doing it as we go together." The curriculum evolves with what's happening at each institution, making real projects the core learning vehicle.

[00:12:48] How to navigate AI innovation pressure when governance hasn't caught up — Dr. Lingam's pragmatic approach: work with what's already been deployed, find the right problem fit, ensure clinical buy-in on the decision logic, and monitor closely post-implementation. For tools that skipped formal review, extra vigilance after the fact becomes the safety net.

[00:14:48] Why AI monitoring requires clinical ownership from day one — At Moffitt Cancer Center, assigning the right owner to watch for model drift is a problem that must be established at project start, not handed off to IT later. "Finding the right owner to keep an eye on the outcomes... is definitely something that we have to emphasize from the beginning of the project all the way to the end."

[00:16:48] The HIPAA gray zone around consumer AI tools in clinical settings — Tools like Open Evidence and Doximity can feel HIPAA-compliant to individual clinicians signing personal BAAs, but institutional use requires an organization-level agreement—a gap that often goes unrecognized. Dr. Lingam also flags automation bias as a parallel risk, and still cross-references UpToDate to fact-check AI-generated clinical answers.

[00:19:48] The right balance for AI tools in medical training — Dr. Lingam wants residents graduating with fluency in tools like ambient scribes, but worries about erosion of foundational reasoning skills. Her approach: change how educators elicit clinical thinking, and use AI for simulation—creating realistic training scenarios rather than replacing the work of building HPIs and differentials. "I don't think I would lock it out, but there's ways to leverage it without hurting the skill building."

[00:23:48] Why de-skilling in medicine isn't new—and what that means for AI — Dr. Lingam draws a direct line from memorization to UpToDate to AI: medicine has always traded narrow skills for broader ones. Her threshold for AI: if it reduces missed diagnoses and corrects for cognitive biases without creating unacceptable new risks, the tradeoff is worth it. "In medicine and healthcare, this has always been happening."

[00:26:48] Why oncology is one of the richest environments for AI applications — Data-driven decisions, precision medicine, clinical trial matching, and multimodal data integration all converge in oncology. Dr. Lingam also points to operational uses—AI as a human extender for triage and symptom management—that free clinicians to focus directly on the patient in front of them.

[00:30:58] How poor EHR design becomes a patient safety issue — Using a nerve block agent accidentally prepared for IV administration and the daily complexity of potassium dosing decisions, Dr. Lingam explains how surfacing the right information at the right moment—rather than adding more alerts—is what actually prevents near-misses at scale.

[00:32:58] Why the strongest clinical decision support blends LLMs with rules-based logic — Context from free text (such as a heart failure description in clinical notes) can feed a rules-based decision engine, combining the consistency of deterministic output with the nuance of language understanding. "We can't have different answers for the same question each time."

[00:33:58] What filter Dr. Lingam applies before taking on a new clinician-requested solution — Her baseline rubric: What problem are we solving? How often does it happen? What's the risk of not acting? Beyond that, individual requests become a window into real frontline friction—and the answer often involves pointing to an existing solution that just wasn't implemented well for that specialty.

[00:35:58] What smaller organizations without a CMIO should do in the AI era — Over 3,000 board-certified physician informaticians now exist, but most organizations don't recognize the role's value. Dr. Lingam's advice: identify internally motivated clinicians and nurses, invest in their education, and connect with the broader informatics network. "It's a very friendly crowd. People share a lot."

[00:40:58] How to move an AI model from research into the standard of care — The algorithm is only one piece. Engaging clinical partners from the problem-definition stage—agreeing on success metrics, workflow integration, and operational accountability—is what actually gets a model to the bedside. An AI tool for detecting lung nodules is useless without a connected lung cancer screening clinic downstream.

[00:43:58] Why eroding trust in healthcare drives patients toward direct-to-consumer testing — Patients who lack a consistent, longitudinal clinician relationship are more likely to seek diagnoses independently. Dr. Lingam frames trust-rebuilding—through better care coordination and physician engagement in operations—as a systemic necessity, not an idealistic wish. "If we all had the good old PCPs of the old days that knew us... you wouldn't have many people jumping to take these tests."

[00:52:58] Why clinical informatics may be the best-kept secret career in healthcare — Dr. Lingam encourages mid-career physicians—not just trainees—to explore informatics pathways that don't require a full two-year fellowship. Multiple entry points exist, and the career satisfaction is, in her words, unmatched. "Clinical informatics is, like, the best kept secret job in healthcare."

‍

‍Fast 5 Lightning Round:

1. What is your favorite book or book you’ve gifted the most?
   So Good They Can't Ignore You by Cal Newport
2. If you could instantly master any skill, what would it be?
   "Data science."
3. Would you rather have Super strength, super speed, or the ability to read people’s minds?
   Super speed.
4. What is something in healthcare you believe others might find insane?
   "I think it's insane that you just don't know when you go into the hospital or a clinic what it's gonna cost you. If I go to a store, I know I can compare bargains... But in healthcare you never know."
5. What is the last movie or TV show you saw, and what did you think of it?
   "Devil Wears Prada 2 – enjoyed it! Also Shrinking on Apple TV really spoke to me... it brings up all the mental health issues, the imperfections of human beings and messy lives, but still with a lot of humor and reality."

‍

The Digital Patient has been recognized as Feedspot's #1 Patient Engagement Podcast of 2025. Thank you to our listeners for making this happen!