Skip to main content

Unlocking the Potential of AI and Big Data in Pediatric Cardiology

Dr. Wayne Franklin of Phoenix Children's is at the forefront of innovation in pediatric cardiology. In this Q&A, we explore the synergy of artificial intelligence, big data, and pediatric cardiology

 Dr. Franklin highlights Al's role in early event detection and big data's power in treatment decision support. Challenges in extracting quality indicators for diverse patients and data security are discussed.

The interplay of science, technology, and business in healthcare is scrutinized with a focus on patient-centricity. Dr. Franklin underscores caution, reproducibility, and diverse data for equitable healthcare advancements. He also spotlights initiatives like the Pediatric Heart Network, driving progress through multi-center collaboration. Join us for a succinct glimpse into the future of pediatric cardiology.

Q: How can artificial intelligence be utilized in the field of pediatric cardiology?

A: It is already being used, for example in critical care settings to detect patient decompensation and adverse medical events. For example, if the patient is in the ICU and has too many risk factors, the care team can be alerted to intervene before the patient has a medical emergency, such as a cardiac arrest or unplanned extubation.

Also, there has been recent work done with electrocardiography (EKGs). That is, we can now detect arrhythmias in babies as young as newborns although up to school-age kids. Yes, I am talking about an infant wearing a smartwatch!

Q: What is the benefit of incorporating big data analytics and deep learning in pediatric cardiovascular research and clinical care?

A: There is so much data it feels right now, especially with the heterogeneity of our patients, that it is difficult to make treatment decisions based a lot on small trials and sample sizes. But, with big data and deep learning, computer programs can learn variations in care, especially with respect to congenital heart lesions, that can lead to meaningful health outcome improvement.

Q: What is the best way to extract data for quality indicators and impact patient care through electronic health records?

A: One of the best ways is to use the data that we already have. That is, there are major health systems that use the same electronic health record. We can get large sample sizes where data can be extracted, and it is all on the same health record platform or format.

Q: Can you discuss the challenges of extracting quality indicator data for heterogeneous ACHD population?

A: Our population is inherently heterogeneous. Not only from an anatomic standpoint but also from a surgical or catheterization standpoint. That is, timing of intervention, surgical procedure type, and age at which the procedure was done, are all challenges that can affect outcomes. And since "no two patients are the same" we will have to get reliable, trustworthy data in order to make meaningful change.

Q: How can larger datasets with learning algorithms be used to analyze Qls for more impactful studies?

A: These larger data sets are helpful because much of the data is already uniform. That is, we can look at trends that may exist that may be able to answer clinical questions. Now, this may sound like a "fishing expedition," but I think will be important to look at trends, which may reflect early health outcomes, that we can then use to answer other simple questions. For example, at Phoenix Children's, we recently published that patients who lived on Navajo Native American reservations had a lower rate of detection for congenital heart disease. That is not to say that these patients had lower congenital heart disease, but simply these pregnant women were going to see their doctor at a much more infrequent rate. Thus, less heart disease was found in the entire population, lower than would be expected.

Q: Considering the advent of AI, what cautionary measures should be considered to protect patient health information?

A: There is a lot of caution needed. First and foremost, we need to make sure that the patient's privacy is protected. It would be terrible if confidentiality was breached. This requires seriously secure cloud storage. We also need to confirm that the data are reliable. There is an old saying in research, "garbage in equals garbage out,” so we have to be sure about the accuracy of the data being collected, entered, and analyzed. We also need to make sure that the patient has not been harmed by the findings. That is, we cannot blindly accept the first iteration that is generated by Al. But we need consistent, reproducible results that will benefit all types of patients. It is sort of the "trust but verify" approach.

We also want to make sure that we can apply principles or facts learned to all patients. We need to have diverse data sets in order to prove the science. If we have only one kind of patient from one kind of area, then we can not necessarily apply the findings to patients in other settings or from other areas.

Q: As a physician, how do you navigate the intersection between patients, science, technology, and business?

A: This is a very challenging process. At the heart of it, I am a physician, so I am interested in improving the health of patients. Much of medicine is science, but some of it is art. Now, we clearly know that technology and healthcare are also businesses. Big businesses. Multi-billion dollar businesses. And, most businesses are out to make money or else they would not be sustainable. The key is to find a way to use science and technology to answer clinical questions which can then be scaled and applied to a business in order to help patients. I think that is the ideal way to help navigate this challenge. But again, use caution. This will take time. Maybe in my lifetime; we will see. But I am going to keep trying.

Q: How can we gain more measurable data and Qls that heart centers can use to correlate clinical care with outcomes?

A: I think the key for pediatric heart centers is to aim for multi-center collaboration. That has really been key in the adult cardiovascular space, and how they can get adult trials with hundreds of thousands of patients. One of these first steps we are already doing, and it's called the Pediatric Heart Network, an NIH-funded organization that has been in place for over 20 years, and it has provided some extremely important multicenter data to help children with heart disease.

Interested by the exciting developments in pediatric cardiology discussed in this Q&A with Dr. Wayne Franklin? Visit us for a deeper dive into the innovative cardiology program at Phoenix Children's.