Category: Innovation

Epic XGM26 FOMO generator: Danny Lee and Bala Kulandaivel, Johns Hopkins

A pediatric DKA risk score!

It was great to see Dr. Danny Lee and Bala Kulandaivel at their poster. They did the hard math to create a predictive model that detects pediatric diabetic ketoacidosis, and displayed it on the multi-provider schedule and on in patient trackboard units.

Congratulations to their achievement! This is hard work, difficult math, really hard tuning to make the alerts relevant to busy physicians/providers in an acute care setting.

Reflecting on our own journey with the Epic Sepsis model and other deterioration scores, we found:

  • Signal to noise ratio tends to be low (in our case about 9% initially, later improved to 30%, still low) even with the best mathematicians and models
  • Acute care clinicians are too busy to look at the scores, and don’t trust them “when my patient looks good right now. I have to go see the sicker patients down the hall, I will not act on your alert now”
  • Consider giving these alerts to a separate virtual team that can surveil ALL patients in the hospital and know what to do for these early warning alerts.
  • Be verrrry careful in the socio-political swoop in when you call a rapid response team to take over patient care from the primary team when you think a “true positive” is detected. This is where we stumbled a number of times.
  • It takes publicizing some “good saves” that might have gone against the primary team’s instincts of “that patient looked fine, and then they crumped for no reason” especially if the virtual team spotted the possibility hours before, but were rebuffed.

These are some details of our difficult journey to the acceptance of predictive alerts like this.

To Dr. Lee and team, great work on a new predictive model! and best wishes on avoiding some of the potholes on this really important journey toward a successful intervention! We will be eagerly watching!

Epic XGM26 FOMO generator: Scott Macdonald with Prospective Consent for AI Scribe

Dr. Macdonald meets the challenge of the California need to consent patients prior to using an AI scribe.

California is a “two party state” where both parties must consent to any recording. This applies to AI scribes in the exam room, and can be a bother to manage correctly with every visit with every patient. For some organizations this is millions of times per year.

Dr. Macdonald’s big idea?

Set a smart data element as part of the question for any clinician to ask the patient “OK if my AI scribe writes my note today? If ok for your visit with me, OK if all your future visits with UCDavis clinicians can do the same?” (my paraphrase, his wording is of course better). If the patient answers “yes” which they do the vast majority of time, then the EHR stores that and shows it to all subsequent clinicians throughout the organization, so no need to ask again. ONE consent, ONE time, revokable by patient if needed.

SO SIMPLE. Thanks, Scott, for this innovation.

Advances in Pharmacogenomics (PGx) — The Next Chapter of Personalized Medicine at UCHealth (guest bloggers Dr. James Martin and Dr. Nicole McDaniel)

Pharmacogenomics continues to accelerate: we are no longer just identifying patients with impactful genetic variants, we are embedding that information directly into care. What started as population screening has evolved into real-time, clinical-first decision support, from completely healthy biobank participants to patients with cancer about to start high-risk chemotherapy.

2025 Milestones at the Colorado Center for Personalized Medicine Biobank…

Our team (CCPM in partnership with UCHealth) has reached new heights.

  • Over 100,000 biobank participants now have PGx results
  • More than 1 million PGx results are available in Epic

These results are not just sitting in charts, they are fueling clinical decision support across the system, helping providers make safer, more appropriate prescribing decisions at the point of care.

And the impact?

  • >34,000 drug-gene interaction alerts delivered directly to clinicians
  • Coverage across 12 genes and 55 medications

Key high-impact drug–gene pairs supported by clinical decision support span a broad range of medications across multiple areas of medicine, including DPYD–fluoropyrimidines (5-FU, capecitabine); CYP2C19–clopidogrel/SSRIs/PPIs; CYP2D6–opioids (codeine, tramadol), antipsychotics, antidepressants (e.g., venlafaxine, vortioxetine), metoprolol, and ondansetron; TPMT/NUDT15–thiopurines; and SLCO1B1–statins.

 

Meanwhile, in Clinical Oncology…

The Clinical Oncology PGx program has rapidly scaled from pilot to systemwide implementation, reaching > 1,000 patients who had PGx results returned to Epic.

In just 5 weeks during the summer of 2025, PGx testing in GI Oncology expanded from 5 clinics to 11 clinics across UCHealth.

New clinics onboarded included:

  • North: GRMC, MCR, HRMNY
  • South: MHC, MHN
  • Yampa: YVMC

This marked a major milestone, the first clinical PGx initiative originating on the CU side to be successfully implemented systemwide at UCHealth.

Alerts tied to this initiative are actively informing care by delivering patient-specific chemotherapy and supportive care dosing recommendations, reducing toxicity risk, and optimizing treatment in real time.

 

And There’s More…

Building on this success, our teams have taken on their next challenge:

  • Expansion of clinical-first PGx testing into breast oncology went live February 2, 2026
  • Next Expansion: inclusion of all other solid cancer types; go-live date is to be determined

Using the same infrastructure, workflows, and CDS tools developed in GI Oncology, this expansion represents a major step toward scaling precision oncology across disease groups.

 Join us next time… 

As PGx continues to evolve from innovation to standard of care:

  • More patients will receive pre-treatment, actionable genomic insights
  • More clinicians will rely on real-time clinical decision support powered by discrete genetic data
  • And more lives will be impacted through safer, more precise medication use

CCPM and UCHealth are no longer just piloting pharmacogenomics; we are operationalizing it at scale.

Thank you for joining us for the next phase of our adventure.

[Blog Editor’s note: We can’t wait! I love having smart, effective colleagues]

James Martin, PharmD, MPH
Clinical Pharmacist, Pharmacogenomics Instructor
University of Colorado Skaggs School of Pharmacy and Pharmaceutical Sciences
Colorado Center for Personalized Medicine

Nicole McDaniel, PharmD, MPH
Clinical Pharmacist, Pharmacogenomics Instructor
University of Colorado Skaggs School of Pharmacy and Pharmaceutical Sciences
Colorado Center for Personalized Medicine

What if AI helped students learn, not just do (harvard.edu)

This is the beginning of the beginning. Teachers are starting to create generative AI that helps students learn, and NOT do the actual assignment. Imagine a chatbot where a student can ask questions outside of the classroom to understand concepts or ask it to critique initial writing. I like this very much. There is something here for medical residents and medical students, and indeed even practicing physicians. Tweaking the relationship between the AI assistant and the human is our hard work to come.

What if AI could help students learn, not just do assignments for them?

UCHealth Parkview reduces sepsis deaths (Beckers)

The story continues. Our EHR, partnered with Epic predictive AI model among other predictive tools have reduced sepsis mortality by 1000 fewer deaths per year compared to our baseline, as we find and treat sepsis earlier with reconfigured teamwork in addition to improved detection tools. Another tale of the Psycho-80: 80% of a project’s success is about the psycho-socio-political skills of the people and 20% of the success is due to technology. Grateful for smart colleagues and partners. (image, our fearless informatics leaders, analysts and trainers having a well-deserved meal after another EHR implementation day)

https://www.beckershospitalreview.com/healthcare-information-technology/ehrs/how-uchealth-uses-its-ehr-to-reduce-sepsis-deaths/

Passwords are so last year. Passkeys! (Wired.com)

I have joined the passkey (quiet) revolution. Far superior even to long passwords, and better and faster than 2 factor and multifactor authentication. I’m all for faster and easier security for my accounts.

https://www.wired.com/story/what-is-a-passkey-and-how-to-use-them/

UCHealth Biobank breaks new ground in personalized genomic medicine (news)

Our smart colleagues at UCHealth Biobank delivered the 1 millionth pharmacogenomic result into our Epic EHR. In separate news, the Biobank also delivered our 1000th pathogenic variant (like BRCA).

Congratulations to the UCHealth Biobank team including Drs. Christine Aquilante and David Kao. See the linked articles:

Your drugs and your genes may not play nicely together. A UCHealth project aims to find out in advance.

https://www.uchealth.org/newsroom/biobank-at-the-colorado-center-for-personalized-medicine-uncovers-1-million-genetic-insights-to-improve-patient-care/

https://www.sciencedirect.com/science/article/pii/S2949774424009981

Our Biobank, in parallel work streams have

  1. Delivered their 1 millionth pharmacogenomic result based on Biobank testing of patients blood or saliva to detect drug-gene interactions to warn prescribers in our system to avoid drugs that may not play well with individual patient’s genetics. We believe we are now have the largest genome bank delivering these results for clinical care.
  2. Delivered over 1000 pathogenic variants (genomic risks like BRCA mutations for breast cancer and the like) so that patients can be aware and take preventive or screening actions.

The investments in this infrastructure began in 2014. Even though it seems like an “overnight success” this was more than 11 years in the making, yet another reason that long term, basic-science approaches should be part of our strategic scientific funding. These transformative technologies come out of years of blood, sweat and tears (pun intended) on behalf of our researchers and technologists and with the contribution of dozens of thousands of patients.

Congratulations to our hardworking, groundbreaking colleagues.

UCHealth joins the Epic COSMOS family. Datathon! Ukulele Parody!

This past week, we hosted our Epic Wisconsin colleagues for an immersion trip on Slicer Dicer and COSMOS, as our researchers, data scientists and faculty dove deeply into Slicer Dicer and the finer points of exploring the deidentified Cosmos database of over 300 million patients. We can’t wait to join the growing list of scientists drawing new lessons from a global repository of clinical care experience.

Scenes from our Cosmos Datathon

In a few short hours after learning the tools, our researchers were asking innovative questions and cranking out quick-win first draft answers that can easily expand into full fledged investigations.

So cool.

I am grateful to smart collaborators:

  • Brian Montague, system physician informaticist at UCHealth for planning and hosting this Cosmos Datathon, where over 60 faculty and scientists came to learn about and explore the Cosmos database and tools.
  • Our Epic Wisconsin colleagues who came for an immersion visit to round with our Transplant and other clinical teams
  • Our UCHealth Epic IT and operational teams for working to install the hooks necessary to contribute to and draw knowledge from Cosmos.
  • Our scientist colleagues who came to learn, practice and grow skills at exploring this data
  • Epic Systems for dreaming up and constructing a de-identified repository where we can safely and securely ask big questions, test our theories, and push forward the boundaries of knowledge of human illness and wellness.

And I am grateful to the Beatles for the original Helter Skelter. A boy band out of Liverpool, who blew apart convention to invent the heavy metal genre.

I like to think this blog is a heavy metal breakthrough as well! (jk)

When Fires Rage, Millions Turn to Watch Duty (wired.com). I think it is the next wikipedia-like crowd-sourced success

Crowd-sourcing when those in charge of disasters do not build the infrastructure to inform the public. The creator of Watch Duty describes how he came to build it.

https://www.wired.com/story/watch-duty-john-mills/

I love a good data story.

Like many of you, I downloaded Watch Duty during the LA fires, as my folks were near the Eaton Fire as it flamed this past month. It was so helpful to LA residents near the fire, and to all of us who care for those residents.

The power of internet computing and the existence of those with skill from any industry, distributed globally, means that anyone (or any team) with the right skill set, the right interest, the ability to build relationships, can create a world-changing tool that becomes instantly useful.

Now Watch Duty is used in 22 states and growing, and the creator is considering adding other natural disasters to the free app. Incredible.

Read the WIRED story and be transfixed. Heroes still exist in our world.

Light-Based Chips Could Slake AI’s Thirst (Wired.com)

Light based chips use photons instead of electrons to transmit zero’s and one’s. This is faster and uses much less power. However, our tools for switching and processing are decades more advanced when based on electrons. Do photons have a chance?

https://www.wired.com/story/ai-needs-enormous-computing-power-could-light-based-chips-help/

Wired writes about innovative tech to move from electronic chips (like market-leader’s Nvidia chips that power AI computing) to light-based chips. With a 10-year runway of development, there is a potential for 1000x increased computing efficiency and reduced power demands using light to transmit data instead of electrons.

This feels a lot like the potential transformation that might come from quantum computing. The people and companies innovating in this space need huge amounts of investment, people, resources to figure out the dead-ends and solutions to as-yet impossible problems.

At the beginning of all these ideas, they all look promising.

Good luck! Of course, 10 years from now, hind-sight will be 20:20. “Oh, of course, these particular projects were successful and the others were just stupid ideas.”