AI, Wearable Devices, and Artificial Wombs: ACC Session Highlights Advancements in Cardiac Technology

In a session deemed “straight out of science fiction” and “akin to a Star Trek episode” by audience members, panelists at the American College of Cardiology (ACC)’s 71st Scientific Session highlighted recent advancements in direct-to-consumer heart monitor devices, the use of artificial intelligence (AI) in cardiology, and the potential of an artificial womb for babies with congenital heart disease.

The session, titled “Innovations in Treatment of Congenital Heart Disease,” was followed by an audience question and answer session, where concerns regarding Health Insurance Portability and Accountability Act (HIPAA) compliance were raised, along with inquiries about clinician reimbursement for management of direct-to-consumer heart data.

In her session, Jennifer Silva, MD, director of pediatric electrophysiology at Washington University in St Louis, Missouri, outlined one ideal integration of remote patient monitoring data—collected via wearable devices (eg, Apple Watch, Owlet, AliveCor)—into cardiology management.

Self-described as an “avid fan of digital health, AI, and remote patient monitoring technologies, both as user and a developer,” Silva took time to define the goals of remote patient monitoring in health care, emphasizing that the tools should provide early or real-time detection of disease and, ideally, reduce hospitalizations among users.

The devices can also be used to improve care efficiency when used correctly and monitor treatment efficacy in real-world settings.

Accurate updates of medical records with relevant data collected is critical in this integration, as is communication with the patient once data are received and interpreted. To better enable assessment, interpretation, and processing, Silva highlighted the important role of cloud and edge computing in this process.

Addressing the caveat that data collected must be reliable to be actionable, Silva also cautioned that wearable technology utilization may also differ based on patients’ age and unique needs. “What’s going to work in babies isn’t going to work in pre-teens,” she said.

Interoperability also poses a hurdle for direct-to-consumer data, as the technology is intended to reduce physician workload, not add to it.

To overcome this challenge, Silva outlined the process in place at Washington University, whereby the workflow is divided into 2 segments: patient workflow and health team workflow. Patients review the data collected and then upload any they deem irregular through a secure patient portal app.

“The health care team workflow, on the other end, is to then receive that message through the inbox, to review the tracing and message from the patients, which are now all archived in the EMR [electronic medical record] and then respond to the patient with a plan,” she said, adding that the technologies are not prescribed for use in every patient.

Price considerations, clinical characteristics, and type of data generated are all taken into account when deciding whether or not to use these technologies to certain patients. Once a patient is selected, care teams can go through the process with the individual and their family before any medical emergency takes place.

“With the increased number of these direct-to-consumer wearables, we're going to increase the amount of data care teams are going to need to sort through. Potential solutions to this problem include both edge and cloud-based algorithms,” Silva said, stressing that integration into the EMR is critical. “Ideally, data acquisition and transmission is practiced prior to clinically relevant events.”

In an additional presentation, Carolyn Vitale, MD, of the pediatric cardiology department at the C.S. Mott Children’s Hospital at University of Michigan Health, defined AI as asking computers and computer systems to perform tasks that would normally require humans. In contrast, Vitale explained how machine learning refers to algorithms assessing vast amounts of data and attempting to identify systems within the information.

In her talk, Vitale outlined uses of AI to help offload clinician burden with a specific focus on cardiology imaging. Citing several studies conducted in recent years, Vitale explained how deep learning algorithms have been trained to identify different arrythmias, with a similar performance compared with cardiologists.

Other studies have shown how these processes can predict ventricular deterioration in patients, while continuously streaming physiologic data have been shown to predict decompensation hours before the event takes place.

One investigation found these data could predict deterioration between 1 to 3 hours prior to an event taking place in 50% to 70% of patients assessed.

As wearable devices become more popular and continue to generate huge amounts of data, Vitale predicted the incorporation of AI into cardiology will increase into the future, bringing with it the potential of tailored therapies and individualized care management. But one outstanding question is how physicians get reimbursed for time spent interpreting this data.

Although Silva noted physicians at Washington University are not currently reimbursed for the process, the practice has resulted in a high yield, she said.

Despite the promise of AI in cardiology, Vitale conceded the path forward will need to integrate both AI and intelligence, noting having an algorithm “is not the same as having a patient in front of you.” This so-called “collaborative intelligence” will need to incorporate both unsupervised—or raw—data and supervised data to better optimize patient outcomes and physician workflows.

One other talk included in the session focused on congenital heart disease in fetuses and the crucial role of the placenta in neonatal health. Studies have shown that poor placenta health is associated with poor baby weight and health, explained J. William Gaynor, MD, an attending surgeon in the cardiac center and surgical director of the heart failure and transplant program at Children's Hospital of Philadelphia.

In an effort to reduce morbidity and mortality rates of extreme premature babies and potentially mitigate the long-term health effects of poor placenta health, researchers developed an artificial womb that has shown promising results in studies carried out on sheep.

One key question the research hopes to address is why fetuses with congenital heart disease have abnormal brain development. Although the innovation has yet to be proven in human trials, researchers hope artificial wombs may one day improve outcomes following cardiac surgery in fetuses with congenital heart disease.

As babies born prematurely are also at a heightened risk of complications (eg, chemical exposure to medical devices) and infections once born and cared for in intensive care units, this research may help elucidate root causes and improve outcomes among these fragile babies.