News

Artificial intelligence (AI) options are popping up everywhere—in word processing software, search engines, and other administrative tools. The most common forms are Large Language Models (LLMs), such as ChatGPT, NotebookLM, Google Gemini, and Microsoft Copilot. 

While these LLMs can be used in many ways to streamline management workflows in labs, it can be challenging for laboratory professionals to find their comfort zone or choose the right program for the right task. To assist, the American Society for Clinical Pathology (ASCP) has developed a new job aid titled “Leveraging AI for Laboratory Administration and Management.” 

This resource can empower laboratory teams to adopt AI tools confidently and ethically without compromising the human expertise that drives quality patient care. 

AI is here to stay 

AI may seem “new” to some in the laboratory, but Dr. Sachin Gupta, PhD, MBA, MLS(ASCPⁱ)MBⁱ, LSSBB, CPHQ, notes that “AI is already used for various tasks in clinical settings. So why not the laboratory?” Dr. Gupta is ASCP’s Scientific Director for the Center for Quality and Patient Safety. 

Staff shortages in the laboratory have continued since the COVID-19 pandemic, and Dr. Gupta feels that LLMs can bridge some of the administrative and management gaps, reduce burdens, and make their work easier. 

Dr. Rodney E. Rohde, PhD, MS, SM (ASCP)^CM, SV^CM, MB^CM, FACSc, and Global Fellow agrees, saying, “If an AI LLM or other application could take 20 percent off your job load, you can spend more of your time on the abnormal complex things that need your expertise.” 

Dr. Rohde is a Global Fellow and Regents’ Professor in the Texas State University System, as well as University Distinguished Chair and Professor in the College of Health Professions, Medical Laboratory Science (MLS) Program. 

Additionally, misinformation about LLMs and their capacity abounds without a lot of clear road maps. 

“There is so much information out there about AI and there's no direction, especially for the laboratory and administrators on how and what to use,” Dr. Gupta says. “This guide really aims them in the right direction and helps them say, okay, what is the task you can do and what kind of LLM or what kind of AI program you can use?” 

The job aid’s benefits to laboratory professionals 

The guide can help laboratory professionals get their bearings on what circumstances to use LLMs in. It begins with a series of eight best practices that encourage laboratory professionals to practice using LLMs, as well as protect their security, follow institutional guidelines, verify accuracy and always use AI to fill knowledge gaps, or as an assistant, but never let it speak for them. 

From there, it walks the laboratory professional through a series of 12 common task objectives—from writing job descriptions, to summarizing meeting notes, or making an email more concise—and suggests which LLM program can assist. It also provides some sample commands to give the LLMs, additional source materials it might be good to submit, and other considerations. 

For example: 

The task “writing introductions to presenting speakers” is something most LLMs can do. A laboratory professional can give it the command to “draft a short, medium, or long introduction,” and attach a CV/Resume. Then they can even specify the kind of tone it takes, be that casual, academic, or slightly humorous. 

Changing the way laboratory professionals work 

Though Dr. Rohde doesn’t work in a medical laboratory, he has found many uses for AI in his academic teaching, clinical placements, and administrative work, such as making emails more concise, summarizing meeting notes, and sending out follow-up actions. 

He says, “[LLMs] become really helpful in trying to take care of some of the monotonous tedious tasks that you might have to do all the time.” 

Now he’s using AI as a kind of virtual assistant. He asks it to verify something he’s writing against the Center for Disease Control’s (CDC’s) standards for a particular kind of testing, or for cutoff points for certain pathogens and compares notes (and fact checks). 

He also uses ChatGPT as a teaching tool in an exploratory career major course he teaches community college students on health professions with Austin Community College in Austin, Texas. He’ll have them use it to look at different kinds of CVs, resumes, and LinkedIn profiles while learning to build their own. 

“Then they can start getting a feel for this as an enhanced tool for what [their] brain needs to create anyway,” Dr. Rohde says. 

Another great use of these LLMs is to help laboratory technical team write and review standard operating procedures (SOP), Dr. Gupta says. “That actually saves a lot of time because writing a policy on SOP is a nightmare that might take weeks to finish. But AI can really help you in creating the template.” 

He does warn that it’s very important to review anything the AI produces. “You need to be very cautious, very careful when you review these policies and make sure your policy is complying with your process,” Dr. Gupta says. 

Informed but fact checked 

The LLMs are so efficient it can be easy to forget that they aren’t always accurate, however, so the job aid also reminds users that “AI can help refine ideas and prevent oversight” but they should review all of its content “to ensure accuracy and clarity,” Dr. Gupta says. 

“You have to make informed decisions when you use the AI,” Dr. Gupta says. At this stage, most LLMs are not capable of fact-checking what they produce. 

Dr. Rohde also emphasizes with students and alumni that it isn’t a replacement for your own words, only enhancing your work. “You still need to do the work. You still need to make sure you're double checking, make sure you're perhaps giving it an example of something and then it's just refining it for you.” 

The evolution of AI for pathologists 

While AI is not quite sophisticated enough yet to integrate deeply into the work of pathologists or laboratory professionals, Dr. Gupta thinks it will get there in the next three to five years. 

One way it might help pathologists is in determining whether to order follow-up biomarker or molecular testing for a specific case, Dr. Gupta says. “Based on the patient's history and current pathology report, there could be an AI algorithm behind the scenes which can really help and facilitate with the pathologist the decision whether the patient needs a follow-up testing.” 

Dr. Gupta is proud of the job aid’s aim to help “improve the knowledge gap” of AI. “I feel like that really will help the laboratory work efficiently and can leverage AI on many day-to-day tasks.”