Advancing AI for vascular surgery will require a commitment to data integrity
There’s no shortage of predictions around the potential of artificial intelligence (AI) to transform healthcare, including surgical care, with an American College of Surgeons article proclaiming AI is "poised to revolutionize surgery."
But reaching a state of true AI transformation for surgical care depends on access to a vast array of quality data collected across populations, providers and sites of care. It’s an area where clinical registries that collect data from procedures can make a substantial difference, including for vascular surgery—but only if providers commit to sharing their data.
Making a case for surgical data sharing
One of the biggest obstacles to AI transformation in healthcare is data integrity, or the extent to which AI algorithms have access to accurate, complete and representative data.
“We generally don't have large, truly representative data sets for healthcare in this country because of the historic underrepresentation of underrepresented groups in clinical trials and other key studies,” Lloyd Minor, dean for Stanford School of Medicine, shared in an interview with The Washington Post.
The value of long-term data sets has long been acknowledged in vascular and endovascular surgery, where information from clinical studies as well as procedures helps inform device and patient selection. It can even help improve device design and inform how surgeons perform procedures. “Without [long-term clinical data sets], we would be doomed to repeat history and the mistakes of our predecessors,” a Cleveland Clinic department of vascular surgery chair wrote in 2017. At that time, data sets that extended beyond 10 years were hard to come by.
Today, clinical registries such as the Society for Vascular Surgery Vascular Quality Initiative (SVS VQI) provide a wealth of information to healthcare providers and vascular device manufacturers on the quality and effectiveness of vascular care. Each month, data from more than 10,000 clinical procedures are added to the registry—with more than 1.2 million procedures represented to date. From there, a secure, cloud-based platform is used to analyze the data.
Initiatives like this are essential in determining the quality of a medical device and its efficacy for various types of cases. But there are barriers to data integrity.
One is the pace at which new devices are introduced and the frequency of mergers taking place between medical device manufacturers. This can result in missing unique device identifiers (UDIs) or multiple UDIs for the same product, making it difficult to accurately capture device information at the point of care. This presents challenges in assessing performance.
The second barrier is the number of providers that don’t share data with clinical registries. For example, while it is impressive that more than 1,000 medical centers and hospitals collect and exchange vascular surgery data through the SVS VQI, there is a significant number of healthcare facilities that could share their data but opt not to. Lack of participation limits the clinical view providers can gain around device efficacy, which patients stand to benefit most from a procedure, and even surgical approach. It also dilutes the power of AI analyses to inform care.
For AI to truly transform care, the healthcare industry must address these challenges.
Steps toward a solution
The SVS VQI’s work toward establishing data integrity around UDI capture serves as a case example for overcoming challenges with missing data.
In July 2022, the SVS VQI launched a partnership with a medical device software company, Symmetric Health Solutions, which works with more than 850 hospitals in the United States and overseas. By pairing the clinical registry with device data from a registry partner, it increases the number of device attributes that can be captured, enhancing patient safety and expanding opportunities for research. It also eliminates manual data entry by healthcare providers and the SVS VQI registry to add new devices to the registry and maintain and update existing records.
Now, says Indiana University Health’s Lillian Camino, cardiovascular data coordinator, “There is a robust number of medical devices available—and there are multiple ways of identifying them. It used to be very challenging finding a device in the database to accurately capture it in the registry. The new functionality for the registry allows for a quick ‘search and select’ of the device catalog. Something that was challenging for abstractors became easy.”
Addressing the participation factor is one of moral importance, especially in an age of AI.
If clinical registries “serve as hubs for cultivating communities of practice, in which mentorship, knowledge exchange and collaboration thrive,” as this researcher suggested in the Journal of International Medical Research, participation could be viewed as vital to closing gaps in knowledge that stand in the way of achieving the safest, highest-quality care possible. In this sense, shared data through a secure registry is a matter of health equity: ensuring all providers have access to data that could make a difference in care across populations. It could also be considered essential to fulfilling the promise to “Do no harm,” enabling physicians worldwide to access the data needed to make the right care decisions.
There’s no doubt that AI holds tremendous potential to make a difference for surgical care and outcomes. By committing to data integrity in all its forms—from active data sharing to device data integrity—surgeons and the organizations they represent can make a difference that extends beyond the reach of the patients they treat toward care quality and safety worldwide.