Houston-based Texas Children’s has experienced “astounding” results from AI projects in recent months, its IT leader told Becker’s.
Myra Davis, executive vice president and chief information and innovation officer of Texas Children’s, was recently recognized for her work when she was nominated for an ORBIE award for the nation’s top healthcare CIO.
Becker’s caught up with Ms. Davis to discuss her most innovative IT projects — and what comes next.
Q: Which IT or innovation project at your health system are you particularly excited about?
Myra Davis: I just shared some really exciting news with our board about the journey we’ve been on with artificial intelligence. While AI in healthcare may seem like a futuristic concept, it has truly become a present-day necessity for Texas Children’s.
We adopted several principles around the use of AI. The first is ensuring it demonstrates a positive impact on the quality of care and patient safety. We also focus on gaining efficiencies, increasing workforce productivity, and ensuring the technology is responsible and secure — meaning it supports nonbiased, equitable development.
We wanted to enhance the patient experience with a strong emphasis on quality and safety. It’s also important that we continuously learn from the technology and use it as an educational and developmental opportunity for our workforce. As a result, we established governance around AI.
We currently have about 10 models live. They’re either embedded within our electronic medical record system or developed using our own instance of a cloud-based solution — specifically, Azure. We maintain continuous improvement cycles to ensure each model is reliable and scalable to meet the evolving needs of the organization.
We developed a methodology that covers the full life cycle: from development to design, deployment, and monitoring. This approach ensures the tools remain effective and aligned with operational and clinical needs. We call this the AI Development and Continuous Improvement Life Cycle Model. Because we build many of these models ourselves, we can scale up or down depending on storage or performance needs. Our governance and guidance committee monitors the models to ensure there’s no bias in the data and that data integrity remains high.
Q: Are there any specific AI use cases you’re particularly excited about?
MD: One problem we needed to solve was the manual process involved in recognizing employees based on Press Ganey survey feedback. The feedback often comes in text form, and it was taking a lot of time and effort to identify when our employees were being recognized through those surveys.
We knew automating that process could boost engagement and morale. Being recognized by a peer or a family member can have a huge impact. So, we developed a large language model that automatically extracts compliments from survey responses and matches them to the appropriate employee. The system then sends an automated recognition email or communication, letting them know they were recognized.
My team, in partnership with our executive team and patient experience team, created this tool — and the results were astounding. We took a three-month sample of data. In the manual process, which involved thousands of roles and a full-time resource to comb through the data, only 23 employees were recognized. That same sample run through the large language model identified 1,303 employees — an increase of over 7,000%.
I’m pretty excited about that. We’ve eliminated the manual labor previously required to recognize employees, and now we can scale that process. The model has an accuracy rate of about 98%. Thirty-one percent of those recognized through the model had never been recognized through the program before. That’s a huge and timely boost for our workforce.
Q: What is another AI use case you’re proud of?
MD: Physician productivity is another area we’re targeting. We use Epic as our EMR, and we’ve leveraged their augmented response technology. It’s designed to free up clinicians’ time by analyzing incoming patient portal messages and generating intelligent, context-aware response suggestions. The clinician can then choose to use the suggestion or not.
The benefit is they don’t have to type the message themselves. The model provides a contextual response tailored to the message from the patient or family. In a recent pilot with 30 providers, we saw a 50% decrease in the average message review time and a 46% decrease in drafting time — in just one month. Imagine scaling that up. It would significantly reduce administrative time and allow providers to focus more on patient care.
Q: Are there any other IT trends you’re paying close attention to?
MD: Remote patient monitoring is a big trend, along with ambient scribe and natural language processing. I like to think of it as “applied AI” — using automation to reduce time, increase productivity, and get meaningful insights faster.
That’s the enablement we need in healthcare: automation that eases the manual workload and allows us to scale from task automation to process automation. The latest advancement is in augmenting human interventions and doing that more quickly and effectively.
Q: What is the biggest challenge in health IT nowadays?
MD: The demand is almost always greater than the supply. The appetite for innovation is strong, but we have to balance that with organizational priorities. At the same time, we must ensure the supporting infrastructure remains secure, reliable and available.
Q: How have you been able to manage around that?
MD: It comes down to portfolio management. It’s about deep partnerships, transparency, and data-sharing with every partner across the organization. We empower our team to be transparent, negotiate, seek clarity on prioritization, and ensure that the problems we’re solving align with the organization’s strategic goals. Alignment is key.
Q: What does innovation look like in health IT nowadays?
MD: IT really needs to evolve. Every leader — regardless of whether they have “CIO” in their title — is responsible for enabling digital solutions that transform how we deliver patient care.
Innovation doesn’t always mean new bells and whistles. It’s about creating opportunities to solve meaningful problems with digital tools, whether those tools are already in place or need to be acquired. The end goal remains the same: to enhance patient care.
