CEO SUMMARY: Artificial intelligence (AI) may be one of the most over-used terms to describe a host of different applications, software tools, and products. However, during the past year, some truly revolutionary digital tools are now in use by a small number of innovative clinical laboratories. These applications are being used to improve operational work processes, to streamline coding/billing/collections, and for analyzing digital pathology images.
COMING TO A CLINICAL LABORATORY NEAR YOU—AND SOONER THAN YOU THINK—will be powerful informatics tools driven by true artificial intelligence (AI) engines. These tools will cover all aspects of lab operations, including managing daily workflow, simplifying lab coding/billing/collections, and advancing diagnostic precision in clinical laboratories and anatomic pathology laboratories.
For nearly a decade, artificial intelligence (AI) has been touted by developers and experts alike as a technology and a tool that will change every aspect of daily life, including healthcare and clinical laboratory medicine.
During this period, adjectives that were frequently used to describe the probable impact artificial intelligence would have on business, social, and cultural activities ranged from disruptive and sweeping to revolutionary and transformative.
Collectively, bold futurists using these terms were declaring that—once AI takes root in a multitude of uses and settings—our society will have knowledge and capabilities unimagined even by the science fiction writers of the 1950s, 1960s, and 1970s.
Wall Street investors jumped on the artificial intelligence bandwagon in a big way. In recent years, a steady stream of start-up companies developing informatics products have been funded with tens of millions of dollars. This is true of firms targeting healthcare and particularly true of emerging companies that want to introduce products that can analyze digital pathology images specifically to make a primary diagnosis without the review of a pathologist.
Invariably, these new firms will describe their technologies and products as artificial intelligence. But the reality is that they are using such informatics technologies as machine learning, neural networks, image analysis, and more in their attempts to replicate the capabilities of the human mind.
The point is that investor-owned companies are quick to describe their products as artificial intelligence. Clinical laboratory professionals should be skeptical of this hype and understand that these enabling technologies may be remarkable at handling vast amounts of data, but will certainly fall far short of meeting the definition of artificial intelligence that functions at the same level as a human brain.
AI Products Arrive in Market
With this background as a starting point, the good news for clinical laboratories and anatomic pathology groups is that a surprising number of products and tools that use AI are arriving in the market. What is common with the most successful of these products is that they can suck up large amounts of data from multiple sources, and then use that data intelligently to accomplish three goals.
First, they automate the work processes they target. Second, they improve the accuracy and quality of those processes. Third, they reduce or even eliminate the labor formerly required to accomplish this work.
These three areas of laboratory medicine now have viable AI-powered solutions reaching the market. Probably best-known, are the early entrants into digital pathology image analysis and diagnosis.
The second area features multiple companies with surprisingly effective tools utilizing AI that are designed to improve all aspects of coding, billing, and collection of lab test claims. The third area involves AI-powered approaches to the lab’s workflow spanning pre-analytical, analytical, and post-analytical functions.
In assembling session topics and speakers for the upcoming Executive War College on Lab and Pathology Management that takes place in San Antonio on Nov. 2-3, 2021, we have had fascinating conversations with entrepreneurs working in the AI field. Their companies now have lab customers using their products and achieving impressive results.
Some Labs Using AI Tools
The Dark Report has issued intelligence briefings about how these AI-powered tools are being used by innovative labs. For example, in the last issue, Lance Berberian, Executive Vice President and Chief Information and Technology Officer at Burlington, N.C.-based Labcorp, told our clients and readers about how the company had deployed AI for use in a wide range of lab functions and activities.
One example is a vision-based AI used in an internally-designed specimen-handling robot system that recognizes loaded test tubes and their appropriate positioning. The AI also identifies and tracks tubes with insufficient specimen quality. In turn, that has helped the lab proactively deal with TNP (test not performed) issues.
Labcorp is also using AI across all 2,000 of its patient service centers. The AI automates many functions when patients arrive to provide specimens. That includes scanning driver’s licenses, with the AI system accurately recognizing thousands of different formats of licenses and other forms of identification. (See TDR, “Labcorp Now Using AI for Operations, Patient Care,” July 6, 2021.)
Revenue Cycle Management
In the area of lab revenue cycle management, we spoke last week with a CEO who has an AI tool in the market that fully automates the intake of a patient’s information for lab billing purposes. He described how his system instantly looks at documents such as driver’s licenses and corrects all inaccurate information in real time. He pointed out that even a driver’s license can have inaccurate information if the individual has a new address, or has a new name because of a recent marriage or divorce.
Anatomic pathology may be the most active area of lab medicine for applications that use artificial intelligence. Last month, I was in Philadelphia and chaired a discussion involving Proscia CEO David West and Scott Gottlieb, MD, former Commissioner of the Food and Drug Administration (FDA).
Proscia has a contract with the federal Joint Pathology Center to digitize the Center’s 55 million glass slides and enable access to those images. Both Gottlieb and West spoke to the speed with which AI and computational algorithms will come to market and be used to analyze whole-slide images, and to generate primary diagnoses with accuracy comparable to a pathologist.
AI in Digital Pathology
Another example of an AI-powered system that is almost ready for use by pathology labs is last month’s FDA announcement that it had accepted the pre-market application of Ibex Medical Analytics of Tel Aviv, Israel, for expedited review. This system is designed to make a primary diagnosis of a whole-slide image. In May, Ibex obtained EU clearance for its Galen Breast Cancer product, which uses AI to analyze the digital images.
There are already companies delivering AI-based solutions to labs for use in digital pathology diagnosis. The FDA recognizes this development, which may be one factor in its decision to do an expedited review of the Ibex system.
AI-type applications are delivering value in a number of innovative clinical laboratories and anatomic pathology labs. The Executive War College is planning to invite some of these lab organizations to present on their experience applying AI apps in ways intended to streamline processes and add value.
Science Fiction Writers Never Envisioned Internet
SCIENCE FICTION WRITERS OF THE POST-WORLD WAR II SOCIETY were professional futurists. Many predictions from the books of Robert Heinlein, Kurt Vonnegut, Arthur C. Clarke, Isaac Asimov, and others became reality in the last half of the 20th century.
But one major development this group of famed science fiction writers apparently never envisioned in their books was how the Internet would appear in the 1990s and evolve into its present form.
Meanwhile, in 1950, British scientist Alan Turing, PhD, published a paper, titled, “Computing Machinery and Intelligence.” Based on his previous 15 years of working with computing machines, Turing articulated the idea that computers would become so powerful, they would think.
He was one of the first to foresee that artificial intelligence (AI) could become reality. In 2012, Scientific American published a story by New Zealand-based computer scientist Ian Watson which described Turing’s test for artificial intelligence.
Watson wrote, “How would Turing know if a machine was intelligent? He devised the Turing Test: A judge sitting at a computer terminal types questions to two entities, one a person and the other a computer. The judge decides which entity is human and which the computer. If the judge is wrong the computer has passed the Turing Test and is intelligent.”
Seventy years have passed since publication of Turing’s paper. We may now be at point where the end game of artificial intelligence—the ability of machines to duplicate human thinking, reasoning, and creativity—may be closer than at any time in history.
Contact Robert L. Michel at 512-264-7103.