The AI You’ve Never Considered
Artificial intelligence-powered process automation solutions may not be making headlines, but they are already delivering benefits today.
Nathan Buchbinder | | 4 min read | Technology
There’s a whole suite of artificial intelligence (AI) solutions delivering meaningful benefits to pathologists and laboratories. But it’s likely you haven’t heard about them – despite the conversation around AI in pathology growing ever louder.
So what are these mystery applications? Well, they fall under the umbrella of process automation, and they use AI to streamline many of the manual and repetitive tasks that slow down your routine workflow.
Unleashing the power of AI
To be clear, process automation itself is not new. It dates back to at least 1785, when Oliver Evans developed the automated flour mill – more than a century before Henry Ford cemented himself in history books with the advent of the automated assembly line. These solutions have left their mark on nearly every industry, including diagnostic medicine. The auto stainer in your laboratory is a perfect example of process automation; it is programmed to take over the manual task of staining slides.
So, what is new? Well, following the broad adoption of digital pathology, we are seeing these applications make an unprecedented impact. Process automation solutions rely on data to serve as inputs, and the wealth of data that digitization generates opens up countless opportunities – from applying computer-based rules to automate processes like case creation and case assignment following “if this, then that”' logic, to the use of computational pathology to automate even more complex tasks. Much like the human mind, AI is uniquely able to account for the tremendous variability that exists across pathology slides or images in practice, and this is where the true potential lies.
We can easily see this potential when considering the automation of quality control. There are not only dozens of artifact types – from pen marks to air bubbles – that render slides unusable, but also significant variation within these artifact types. Just as some pen marks are thicker than others, no two air bubbles are identical in size or shape. The difficulty of overcoming this variability is why quality control has remained a manual process, even if it takes hours per day at a high-throughput laboratory.
Object counting is another process that has historically proven difficult to automate. Pathologists might not consciously think about the variability from staining protocols, tissue type, and histomorphology – among many other factors that they encounter when performing a task like mitotic counting. Remember, the human mind is adept at overcoming this diversity. However, as essential as mitotic counting may be for gaining critical clinical information, it is still time consuming.
What this means for you
Quality control and object counting are just two of the many processes that are ripe for automation with AI, and they both illustrate the same broader trend. Digital transformation isn’t just about technology. Rather, it’s about using technology to unlock value for users. It’s no coincidence that we’re also seeing this from the adoption of digital pathology more generally. Just look at how it has streamlined collaboration and improved data accessibility, for example.
Process automation applications deliver value by ensuring quality and saving time, enabling pathologists to deliver on their commitment to excellent patient care and research. These solutions give you time to focus on generating the best possible results and provide you with the data to make it happen. Process automation applications can also help to improve work-life balance (you could equally use the time saved to catch your child’s soccer game or enjoy dinner with your family) – and this becomes even more significant considering that one-third of pathologists already report feeling burnt out (1). As new data from the College of American Pathologists shows that demand for hiring is strengthening amid the ongoing labor shortage, the need for work-life balance will likely only intensify (2).
I’ll leave you with one piece of unsolicited advice. Like all computational applications, AI-powered process automation solutions must be integrated into routine workflows to deliver their full value. Otherwise, you’ll spend most of the time that you just saved manually moving data from one application to another or toggling between screens. And that’s all the more reason to center your digital pathology practice around an open, interoperable platform that allows you to easily introduce new AI applications – from process automation solutions to detection solutions – into your day-to-day operations. You’ll also gain the freedom to incorporate other solutions, including laboratory information systems and whole-slide scanners, creating a connected digital ecosystem that will scale with your laboratory.
The era of computational pathology is upon us. As your lab looks to chart its path, there's a good reason to start with process automation. Though these applications may not get as much attention in the latest research, they are delivering clear return on investment. Implementing AI-powered process automation will enable you to increasingly practice at the top of your license today and ease into adopting additional computational applications as you realize the full promise of pathology’s digital transformation.