The First Steps to Implementing Digital Pathology and AI

Taking on new technologies can be a daunting prospect. When considering digital pathology and artificial intelligence (AI), labs may not know where to begin or how to take the next step in their journey. With obstacles ranging from limited resources to resistant personnel, it’s no surprise that many labs have yet to explore AI’s potential to enhance their work.

But with workloads increasing, patient interest in AI rising, and staffing shortages a perennial issue, Nathan Buchbinder—cofounder and chief strategy officer at Proscia—believes it’s time to overcome that hesitancy and embrace the era of AI. Here’s his advice on getting started.

Making the move

The first step for any lab aiming to apply AI technologies is to go digital—and Buchbinder recommends taking the long view from day one. “Look for hardware and software that are designed with AI in mind,” he says. “That means looking at interoperability, workflow, and usability considerations so you can be confident that the technology will support your goals.” And, because innovation moves so quickly in the space, labs need to consider not just today’s AI applications, but also those that are likely to arise in the near future.

For labs that are already comfortable with digital technologies and want to take the next step toward an AI-enabled future, Buchbinder says, “Decide what your vision for AI looks like, then work with platform vendors and individual AI companies to map out an implementation strategy.”

For many labs, this may seem daunting—but advance planning can smooth the way. “Assuming you’ve adopted digital pathology appropriately, it doesn’t take much additional effort to layer AI on top,” Buchbinder says. “You need to get buy-in from the laboratory team. You need to think through the workflow considerations and assess the potential return on investment (ROI). You need the capital to make your vision a reality. But, once those elements are in place, you’re ready to move forward.”

The business case

Studies have estimated the productivity gains of digital pathology at 13 percent¹ and its annual cost savings at anywhere from $114,000 to $3.6 million just by optimizing lab workflows and resource use.^2,3 “When you add in AI solutions like diagnostic support or workflow automation solutions, you can add another 20 to 40 percent improvement on top of that,” says Buchbinder. “That’s already a substantial ROI for the lab. As we look to the near future, we’re seeing a whole new class of computational solutions that demonstrate ROI tied not to how the lab operates, but to the value the patient receives.”

This, Buchbinder says, is where the greatest opportunities to make a business case lie. “How much is a faster diagnosis worth—not to the lab, but to the patient? I think that’s the question that we need to be asking.”

But the potential gains are not limited to the lab. Earlier diagnoses can permit more conservative management or increase the likelihood of treatment success, reducing downstream costs across the healthcare system. To achieve this, though, institutions must first be willing to change the balance of resources deployed for each patient’s care—especially as new AI applications enable the delivery of more informed and more personalized treatments.

“It all starts at the place where you’re currently spending almost the least amount on the care journey,” says Buchbinder. “Only about 2 percent of healthcare expenditure goes into the lab; the remaining 98 percent is spent elsewhere.⁴ But lab testing informs the vast majority of treatment decisions, so it’s important to consider the ROI in terms of everything that ultimately happens to the patient as a consequence of the decisions made in the lab.”

Considerations and priorities

What should labs look for when choosing a vendor or solution for their digital and computational pathology needs? First and foremost, Buchbinder says, the solution should meet the needs of the lab. “That doesn’t just mean that it addresses a particular workflow or use case,” he explains. “Digital pathology and AI are a net positive for patients and healthcare systems—but they do mean that pathologists will be shifting a significant portion of their work from the microscope to the digital platform. Consequently, the system needs to be designed with the pathologist and their expectations in mind. It needs to be as intuitive as using a microscope.”

Buchbinder recommends a “try before you buy” approach that gives future users of the system the opportunity to extensively demo potential systems, experience their functions and experiment with applying them to pathologists’ work. He also highlights the importance of interoperability. New technologies don’t just need to connect with existing laboratory information systems, instruments, and software; they should also be able to accommodate the lab’s long-term goals. “How confident are you that, when the latest and greatest AI application comes out, it can be easily integrated into your workflow?” he asks. “Are you going to get stuck in a closed system where what you adopt now is what you’ll have to use for the next 15 years—or are you investing in an open platform and architecture that allows you to remain at the forefront of your field, delivering top-quality services to your patients?”

Finally, consider the vendor’s track record. The best way to ensure that a vendor can deliver on their promises to your lab is to see that they’ve done so for other labs already.

As a final note, Buchbinder adds, “I think labs—especially smaller or more resource-limited ones—often deprioritize digital pathology and AI. In many cases, they’re hesitant to adopt these technologies because they’re concerned about the cost. They may be feeling pressure to tighten their budgets rather than take on new expenses. My advice to those labs is to ask themselves how they can get out of that precarious position. It’s not by shaving a little off the budget here and there. It’s by thinking differently—and that includes considering the role technology can play in expanding their business and their opportunities.”

References:

1. 1. Griffin J, Treanor D. Digital pathology in clinical use: where are we now and what is holding us back? Histopathology. 2017;70(1):134–145. doi:10.1111/his.12993.

1. 1. Hanna MG et al. Implementation of digital pathology offers clinical and operational increase in efficiency and cost savings. Arch Pathol Lab Med. 2019;143(12):1545–1555. doi:10.5858/arpa.2018-0514-OA.

1. 1. Ho J et al. Can digital pathology result in cost savings? A financial projection for digital pathology implementation at a large integrated health care organization. J Pathol Inform. 2014;5(1):33. doi:10.4103/2153-3539.139714.

1. 1. Bogavac-Stanojevic N, Jelic-Ivanovic Z. The cost-effective laboratory: implementation of economic evaluation of laboratory testing. J Med Biochem. 2017;36(3):238–242. doi:10.1515/jomb-2017-0036.