Today, AI is making headway on realizing those promises. Visit an emergency room, hospital, or clinic, and you’re likely to find that AI is not only alive and well but is already an intrinsic part of healthcare delivery.
While these innovations often lie below the surface of what a typical observer might see, AI exists in many nooks and crannies of the healthcare system. These include patient triage, clinical workflows, and diagnostic and treatment plans. The current AI healthcare environment is, however, just the beginning of the marriage between these two powerful forces: technology and healthcare.
Investment in healthcare AI is surging. In fact, the global AI healthcare market is poised to increase to $2.81 billion in 2026 from $1.94 billion in 2020— a 41% compounded annual growth rate. U.S. stakeholders should buckle in for a wild ride.1
As Graham Billingham, MD, chief medical officer at MedPro Group, said at the “Al Algorithms for Healthcare Delivery” session at the MPL Annual Conference in Salt Lake City earlier in 2022, “The train has left the station. We’re going to live with it. We’re going to have insureds using this technology, so we need to embrace it and understand the risks and the benefits. With the amount of money that’s going in and the speed of new technology, like so many things in clinical medicine, the danger is that the pace of development outruns our ability to adapt to it.”
One aspect of the marriage between AI and healthcare that Dr. Billingham alluded to that is receiving less attention than the actual technology are the medical professional liability (MPL) implications. Companies and providers should keep these in mind as AI adoption continues to soar.
The ABCs of Healthcare AI
The goal of healthcare AI is to support clinical decision-making and make patient care safer by reducing errors and adverse events. The term AI represents an amalgamation of applications designed to reach these objectives, including:2
- Machine learning: Enhances statistical modeling to allow models to learn through training with data. Healthcare applications include recognizing diseases in radiology imaging and determining the likelihood that a patient will contract a particular illness.
- Natural language processing: Allows computers to analyze and process human language to perform repetitive tasks. Healthcare applications include preparing reports on examinations and analyzing clinical notes.
- Rule-based expert systems: Creates “if-then” rules-based applications within particular systems. Healthcare applications include electronic health record systems and clinical decision support applications.
- Robotic process automation: Performs structured administrative digital tasks involving information systems. Healthcare applications include obtaining prior authorizations and updating patient records.
- Physical robots: Execute specific predefined physical tasks. Healthcare tasks include delivering supplies, performing surgery supervised by surgeons.
While the idea of AI may seem intimidating and evening frightening, especially in healthcare, the fact is that AI is omnipresent in healthcare today and the machines aren’t taking over. “Artificial intelligence has such a broad definition, there are things we’ve been relying on for years that you could think of as AI,” said Ann Weinacker, MD, senior vice chair of medicine for Clinical Operations and associate chief medical officer for patient care services at Stanford Medical Center. “For example, when a patient has an EKG, the machine provides an interpretation of what the EKG shows to help the clinician make a decision about treatment.”
“I don’t think anyone is depending entirely upon AI, but there are certainly areas where it has demonstrated significant utility,” she continued. “For instance, with radiology, machines (computers) can be ‘trained’ to recognize characteristics that would suggest that a shadow on an X-ray represents a malignant nodule rather than something that’s relatively benign. Similarly, there are ways to train machines to recognize specific characteristics in pathology specimens.”
The Promise of AI
There’s no doubt that AI offers the potential to revolutionize healthcare in the long term while continuing to increase efficiencies and effectiveness in the short term.
For David Feldman, MD, chief medical officer of the Doctor’s Company, Al is already making healthcare more user friendly for both doctors and patients, while creating a more accurate record of the events that occur during patient care. “During the pandemic, we engaged in a good deal of virtual care,” he said. “Imagine the capability of audio technology to capture a conversation between a doctor and patient and create a medical record entry from that. Just that would save doctors a tremendous amount of time that they now spend on charting.”
Improving accuracy of record keeping and electronic healthcare records would reduce risk, Dr. Feldman noted. “Diagnostic error can be reduced through this kind of AI, and as we know, diagnostic error claims are over-represented when it comes to costs because unfortunately, a lot of diagnostic errors are made in young people with longer life expectancies, which becomes expensive,” he continued.
AI technologies in use in other countries, including Canada and South Korea, are already recording events that occur in procedures such as operating room surgeries. Those records can be used to bolster MPL case defense, he noted.
For clinicians such as radiologists, dermatologists, and pathologists, AI offers the potential to enhance their diagnostic abilities to benefit patients and reduce harm. At the MPL Association Annual Conference, Noor Ahmad, MD, a diagnostic and interventional radiologist at Columbia University Irving Medical Center, said, “The word diagnosis comes from the Greek ‘knowing apart.’ Machine learning algorithms will only become better at such ‘knowing apart’ or partitioning, but knowing in all dimensions transcends task-focused algorithms. In other words, in the realm of medicine, perhaps the ultimate rewards come from knowing together. That’s the hope that they’ll augment our diagnoses.”
Dr. Weinacker also is enthusiastic about the potential AI has to improve patient care and patient safety. “To me, the most exciting things are those that will actually help you determine if a certain patient is at risk for a specific condition. If you’re a primary care doctor, you may be concerned about a patient that has high blood pressure with a family history of heart disease and some secondary risk of diabetes. Think about the potential for AI to help clinicians figure out the correlation of diseases and how to better determine the specific risk factors for individuals. This is a significant issue in the management of cancer and has huge implications if AI can help us more precisely pinpoint and mitigate those risks with early detection.”
“There’s only so much information that a clinician can compile and absorb,” she continued. “All those things are out there, but when you give that information to a machine that has been programmed correctly and can assimilate the information, machines can give us a much better sense of risk factors in many, many different situations. I work in the ICU, and there are programs that are being developed and perfected that can identify patients who are increased risk of sepsis or heart failure, for example. That information does so much to inform patient care that it’s really exciting.”
Brian P. Wicks, MD, senior consultant on MPL business lines for Gradient AI, noted the potential that AI has in two key areas: patient safety and streamlining business and claims processes. “AI can help MPL companies identify doctors who might be headed for trouble and get those people additional training,” he said. “AI offers incredible potential to stop patient harm by allowing us to use the data we have to leverage patient safety data for better training and ultimately, better patient outcomes.”
There’s just as much potential to streamline business and claims processes for MPL carriers. “Insurance is all about risk; if you can manage risk and claims you will be in a better position,” Dr. Wicks said. “AI offers the ability to improve risk assessment, pricing, and claims. For example, AI can help determine how expensive a claim might be early in the process, which can help determine reserving and inform decision making on whether to settle or litigate a case based on experience. Essentially, AI puts all that information at your fingertips.”
The Potential Perils of AI
AI has the potential to complicate MPL cases because it introduces more variables into patient care. Instead of clinicians making all the decisions, there’s also AI input. In cases where medical error or liability is present, that could create additional complications.
Matthew Keris, JD, shareholder with Marshall Dennehey, put it this way at the MPL Conference session: “When I first started, I represented a hospital and would go into that hospital, pull the chart in the manilla folder down, photocopy it, and produce it,” he said. “When electronic medical records came in, they were printed out in volumes and volumes of records.”
Today, many clinicians have clinical decision-making tools that are integrated into electronic healthcare records. “Now, literally, it’s like the chart is tapping the physician’s shoulder and saying, ‘Hey, doc—what do you think about this?’” he said. “So, in a way, it is a new participant in the relationship that was physician/patient. Now, there’s an argument, or some could say that the chart has a role in the decision-making with patient care.”
As clinical decision-making tools further integrate into healthcare settings and technology, Keris and many other MPL legal experts voice concern about how this and other types of AI may affect MPL claims. Potential AI-MPL issues include:
- Clinicians agreeing with an AI clinical decision-making tool suggestion that turns out to be harmful or wrong
- Clinicians disagreeing with an AI clinical decision-making tool suggestion that turns out to be correct
- The potential for AI technology to make errors that lead to misdiagnosis, failure to diagnose, medication errors, delay of treatment and more that could result in patient harm3
According to an article published in the National Law Review, “If AI is involved in the provision of healthcare (or other) services, both the developer and the provider of the services may have liability under a variety of tort law principles.”4 Much depends on how the law evolves regarding AI and MPL. The article goes on to note that “those involved in the process (the developer and the provider) will likely have exposure to liability associated with the AI. Whether the liability is professional liability or product liability will depend on the functions the AI is performing. Further, depending on how the AI is used, a provider may be required to disclose the use of AI to their patients as part of the informed consent process.”5
Clearly, providers and carriers must think carefully about all the medical legal implication of AI. Contracts with AI providers and the warranties both sides are providing within those documents will need to be carefully considered. You’ll also need to think about consent documents and how any AI use is disclosed. Finally, there are also important issues of training clinicians who will be relying on AI to diagnose and treat patients.
A Final Word
There’s no doubt that adoption of AI in healthcare will continue at a rapid pace. MPL stakeholders can continue to educate themselves on the potential risks and rewards, with a goal of better serving patients and those who provide care.
References:
1. “Artificial Intelligence in Healthcare Market Share Insights 2022: Top Countries Data—Future Growth Developments, Impact of Covid-19 on Industry Size and Business Plans Forecast to 2026,” MarketWatch.com, April 21, 2022, https://www.marketwatch.com/press-release/artificial-intelligence-in-healthcaremarket-share-insights-2022-top-countries-data-future-growth-developmentsimpact-of-covid-19-on-industry-size-and-business-plans-forecast-to2026-2022-04-21
2. “The potential for artificial intelligence in healthcare,” Future Healthcare Journal, Royal College of Physicians, June 2019, https://www.ncbi.nlm.nih.gov/pmc/ articles/PMC6616181/
3. “Can Artificial Intelligence Be Held Liable for Medical Malpractice?” Eisen Law Firm, https://www.malpracticeohio.com/blog/can-artificial-intelligence-be-heldliable-for-medical-malpractice.html
4. “Top 10 Legal Considerations for Use and/or Development of Artificial Intelligence in Health Care,” National Law Review, Feb. 16, 2021, https://www. natlawreview.com/article/top-ten-legal-considerations-use-andordevelopment-artificial-intelligence-health
5. “Top 10 Legal Considerations for Use and/or Development of Artificial Intelligence in Health Care,” National Law Review, Feb. 16, 2021, https://www.top-ten-legal-considerations-use-andor-developmentartificial-intelligence-health