A Passion for How Things Work: A Conversation with Optum’s Tracy Byers

« I came to have a real passion for understanding the way things work, » says Tracy Byers, CEO Enterprise Imaging at Optum, during her conversation with Clarius CEO Ohad Arazi on their Medical Imaging for All podcast. Read on for more highlights in this article.  

Tracy’s thirty-year career has seen her helping create products and technologies that forward better patient care, and that help make hospitals and health systems more secure. After starting her career at Philips in finance, acute monitoring, and imaging, Tracy went on to IBM Watson Health’s new startup, where she learned about how artificial intelligence and machine learning can be applied to healthcare. Today, she leads the enterprise imaging business at Optum. There, she’s building scalable cloud-native solutions that automate the clinical and administrative workflows for imaging-based diagnostics for large hospitals, radiology practices, physicians, and patients.

Building Great Teams and Working with Great People

For Tracy, developing innovative healthcare technology is as much about the people as it is about the product. She says that it involves « caring deeply about and being passionate about the health care problem you’re trying to solve and pulling together some of the best and brightest people to solve those problems. »

Having a passion for wanting to know how things work was rooted in her from childhood, and led her into product roles, and then management roles. She looks for other team members to collaborate with who have the same curiosity to see things work, but who have different strengths and skill sets to bring to development, and who desire high levels of accountability.

What’s something that she looks for that can encourage better decision making and bring a higher impact in a healthcare company? Having a start-up mindset. « When I look to scale the organization, I want a mix of people who might have been from startups, who took an idea from a very early stage and really have experience scaling. Because so often big company people don’t have direct experience in big scale, » she explains. By combining these skill sets on each team, Tracy says they get the benefits of both worlds: the heavy process experience of big companies with the lack of tolerance for slow decision making from start-ups.

Moving to the Cloud and Unlocking Medical Imaging Data

More hospitals and healthcare systems today are looking at ways to shore up their systems, protect their data, and prepare for the future. « Initially when we would talk to CIOs and chiefs of radiology and cardiology, there was a resistance that they could do a better job of protecting their data and PHI or personal health information themselves,” she says.

But shifts in today’s industry have created an urgency around more sophisticated data protection. Tracy explains that those five changes are: care moving outside the four walls of the hospital, and the need for access to images anytime, anywhere; realizing the value of imaging data and analytics; the need for increased cybersecurity; resource shortages and physician burnout; and COVID shutdowns. « Now probably more than 50% of our largest customers are interested in moving to the cloud, beginning the journey to the cloud, whether it’s through their archive or some component of their enterprise imaging system,” she explains.

As hospitals think about scaling along with the data they handle, they need to also think about how to apply innovative ways of analyzing that data to inform better health care. Today, most medical imaging data is locked up in the servers in the four walls of a hospital. But Tracy believes that « if we were able to put it in a cloud where it was much more easily accessible and anonymized, the power of that data could be used to train AI algorithms to help do medical research, improve care paths. It really has a huge opportunity. »

Building Better AI

When it comes to using medical imaging for AI, Tracy says her team is more focused on training algorithms than on image recognition. « I have no doubt that AI will become a critical component of radiologists’ workflow, » she states, and believes that it won’t be AI replacing radiologists, but radiologists using AI replacing those who don’t.  » I do think that it is a tool to aid in supporting them for their peak efficiency and productivity. »

However, there are a few barriers to AI that are holding up adoption today, including reimbursement, accessing large data sets, and ensuring AI doesn’t « create additional clicks or additional steps in workflow. » Ultimately, though, more widespread use of AI will come through decreased skepticism and increasing trust through more accurate algorithms.

« We’ve really focused on only developing and investing in algorithms that really improve the workflow, » she says. However, she believes the industry is still early in understanding how algorithms work. « Skepticism is about whether the algorithm has been trained properly and whether its answers are always right and will continue to be right,” she explains. “Until people really deeply understand how the algorithm is trained, [they] are going to be hard pressed to really embrace it. »