By - August 22, 2023
The laboratory is a crucial component to good healthcare. According to a 2021 survey in the American Journal of Clinical Pathology (AJCP), up to 70% of the decisions that involve a patient’s diagnosis, treatment, and even discharge, are derived in some way from laboratory test results.1 Tools that can improve the function, efficiency and quality of laboratory work should be embraced, experts suggest, especially the burgeoning wave of artificial intelligence-based (AI) tools, which have the potential to revolutionize laboratory work.
AI is already, albeit slowly, being adopted in the clinical laboratory, for such tasks as “enabling the effective use of resources, avoiding unnecessary tests, improving patient safety, and alerting for abnormal results,” write the authors of the AJCP study.1 It’s also starting to be used for molecular and genomic testing.
Where AI really shines in a way that can be helpful to laboratory professionals is with tedious tasks, according to Andy Beck, MD, CEO of PathAI, an AI-powered pathology technology company based in Boston. “Verifying data, image recognition, language. There are definitely ways of plugging AI in for quality control, for efficiency of labeling, for patient safety, checking the verification of labeling and more,” he says.
By assuming many of these tasks, AI can help laboratories increase not only efficiency, but patient safety. AI has the ability to aggregate and analyze all the information generated on each patient, and even potentially “to understand and predict which patients will have adverse reactions based on lab data,” Dr. Beck says.
It won’t replace manual tasks that require dexterity. If anything, it will only eliminate the kinds of functions that Dr. Beck describes as “those that people don’t want to do anyway.”
Another way AI will change laboratory professionals’ jobs, according to Dan Lambert, CEO and cofounder of PathologyWatch, an end-to-end digital pathology solution, is to catch problems with slide staining. “The AI will have the ability to pick up a staining process that’s out of control or smudged or has slight artifacts that will impair the readability of the case. It can alert technicians at different stages in the process and say ‘Okay, something is out of tolerance,’ or ‘this case needs to be redone.’”
He explains that the laboratory professional’s job may change to have a “continual feedback process that didn’t exist before.” Ultimately this will lead to fewer errors.
Automation in the laboratory is not new by any stretch. And, according to Brian R. Jackson, MD, MS, Adjunct Professor of Pathology, University of Utah and Medical Director for Business Development, ARUP Laboratories, process automation is an area where AI has a lot of potential. “There are a lot of repetitive tasks behind the keyboard in a lab, entering data or looking stuff up, the not-glamorous stuff, clerical things that AI will make easier,” Dr. Jackson says.
AI expert Joseph Rudolf, MD, Assistant Professor of Clinical Pathology at the University of Utah School of Medicine, Medical Director of the Automated Core Laboratory, and Medical Director of Clinical Informatics at ARUP Laboratories, suggests that the next wave of automation for clinical laboratories is “AI assistance, allowing people to do what they do best and machines to do what they do best.” For example, he points out how laboratories have experience huge gains in quality “because machines are more reliable pipetters of fluids than people are.”
In the very near term, AI will help with workflow steps in the laboratory, to gain efficiency for such tasks as specimen routing, specimen stability, and using AI to detect pre-analytic error and surface specimens that need a secondary review, Dr. Rudolf says.
Other areas where people are already experimenting in the clinical laboratory with AI is in using AI for auto verification of results,” Dr. Beck says.
While some may be concerned about implementing a tool that isn’t yet without flaws, Dr. Beck makes the case that “AI is an improvement over what is preexisting. It doesn’t have to be perfect; if it’s better than what we do today, as standard, then it’s an improvement.”
One of the new AI tools that’s getting a lot of attention right now is ChatGPT, a large language model that can recognize patterns within text and generate answers to queries the user asks for based on the data it has seen. While it may not have a lot of uses in the laboratory, Dr. Jackson explains that there are ways AI can support communication. “People are experimenting with using ChatGPT to help them phrase responses to clinicians’ lab-related questions, or to draft more readable footnotes to attach to lab test results. Language and communication things may be good uses for this kind of AI.”
Overall, Dr. Jackson says, “Different AI tools are likely to start showing up in different parts of the laboratories, whether it’s embedded in the instruments or workflow tools.” In other words, AI is here to stay, and rather than be afraid of it, he recommends laboratory professionals learn about it and stay abreast of it. “I think it’s really important for folks in the lab to have some basic understanding of AI, to demystify what’s seen as magic. It’s not humanoid. It’s not sentient. It doesn’t have any human characteristics, as much as we’d like to pretend it does. It does have different limitations you have to be aware of and watch out for, to know how to validate it.”
Dr. Beck also explains that AI can defray some of the tremendous pressure the healthcare industry is in to provide more healthcare services “faster, better, cheaper.” He adds, “For the growth in volumes and throughput that we need to achieve, we can’t find enough people to do that work. So, any tools for automation or AI are going to allow us to scale those businesses with a smaller workforce.”
For those who worry that AI’s efficiency will start to put laboratory professionals out of work, Dr. Beck reassures, “I don’t see a world in which we’re laying off mass numbers of pathology laboratory staff. I think we’ll just continue to grow what we can do. I’m worried about how we take care of more people providing more service in more cost competitive ways. So, the AI tools are going to be the assist for that.”
The alternative to not adopting AI tools is that laboratories will not be as efficient, he says. “That doesn’t mean everybody should be a bleeding edge adopter of the technology. It has to be done thoughtfully.”
Paranjape, Ketan, et. al. “The Value of Artificial Intelligence in Laboratory Medicine.” American Journal of Clinical Pathology. Published December 4, 2020. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8130876/