Artificial intelligence in health care: Risks and benefits for medical professional liability

It is no exaggeration to state that artificial intelligence (AI) will transform health care delivery globally. In fact, it will revolutionize health care over the next decade. The amount of money invested in AI by health care organizations has grown exponentially in recent years. It was estimated to be $15.4b in 2022 and expected to continue to grow annually at 37.5% compounded annually from 2023 to 2030. For the first time in history, providers will be able to apply all medical knowledge, including the most recent studies, for the benefit of individual patients in tailoring personalized care. Artificial intelligence systems are already in use in various clinical settings.

Artificial intelligence can be broadly defined as the “science and engineering of making intelligent machines, especially intelligent computer programs.” Machine learning (ML) “is an artificial intelligence technique that can be used to train software algorithms to learn from and act on data.”

AI will be able to conduct and improve a broad range of health care functions such as clinical decision support, responding to patient queries, detection and diagnosis of conditions, designing and prescribing drugs, monitoring patients with analysis of vital signs, (in acute care settings and remotely), performing procedures, improving analyses of tissue in pathology studies, analyzing doctor-patient interactions and conversations and improving analyzing radiology (and other specialties) films and studies. In time, as providers adapt to the use of AI, patient care should benefit greatly,
and medical professional liability (MPL) risk should be reduced. But AI will also create new MPL risks to be managed and insured.

Benefits

One of the best and most important ways AI can reduce malpractice risk is by improving diagnostic error and reducing other types of medical errors, like with medication administration, radiologic interpretations and pathology studies that can cause serious injuries. Diagnostic error has consistently resulted in a large percentage of all malpractice litigation for many decades.

In 2019, a landmark study published in the medical journal Diagnosis found that as many as one third of all malpractice claims result from delayed or inaccurate diagnoses, making it the leading cause of major injuries (death or permanent disability) attributable to medical errors. These diagnostic error cases were from the Harvard Risk Management Foundation’s captive insurer Controlled Risk Insurance Company (CRICO) Comparative Benchmarking System database in the ten-year time frame of 2006-2015. At that time, the CRICO database contained 28.7% of all malpractice claims in the country. Three quarters of those errors were attributable to three conditions: cancer (37.8%); vascular events (22.8%); and infection (13.5%). A recent study on diagnostic error published in July 2023 by researchers from Harvard and Johns Hopkins attributed 795,000 incidents of death and permanent disability annually from this cause. AI can help avoid and prevent diagnostic errors thereby improving patient outcomes and preventing malpractice claims.

Health care delivery involves processes, but these can often be very inefficient and cause patient injuries through such factors as delay, misidentification of patients and drugs and others. AI solutions can help prevent patient injuries from these causes. But AI can also flag process errors before they result in patient harm (e.g., medication errors). It provides better decision support to providers by summarizing vast troves of medical data on the cases of individual patients.

Risks

There is the potential for harm to a large number of patients should a clinical decision support (CDS) algorithm be poorly designed or not kept current with good data. This could result in costly batch claims. Claims could also be made for not incorporating AI in the delivery of the patient’s care or for improperly using AI. System malfunctions can put patients at risk for injuries. Examples are diagnostic errors, lab errors, misread radiology films or studies, incorrect pathology blood/ tissue study interpretations and others.

Physicians will require training to understand the use of AI and how it predicts clinical risk. If AI is not utilized properly, patient injuries can result. Physicians at present lack the skill set to incorporate AI algorithmic outputs into patient care decisions. Clinical training and medical education must address these weaknesses.

Note that there are exposures to multiple lines of coverage beyond health care professional liability in a claim scenario including tech E&O, products liability, life sciences, general liability, cyber liability and others. Claims defense in cases involving the use of AI will be more complex.

Conclusion

In time, there should be a net benefit to health care organizations and their health care professional liability insurers, as AI markedly improves patient care and reduces patient injuries resulting in malpractice claims. The greatest risk can come at the inception of the use of AI due to physicians and nurses and other providers lack of familiarity with the use of AI to help with decisions about patient care. Flawed algorithm design may result in harm to large numbers of patients and is also a major malpractice risk.

There will always be a role for human judgement in the use of AI for patient care. AI most likely will improve health care decisions and processes when used properly.

1 Mills, Terence. Member, Forbes Technology Council. The Risks and Benefits of AI in Medicine. Forbes Newsletter. Mar 23, 2021. https:// www.forbes.com/site/forbescouncil2021/03/23/the-risksand-
benefitsof-ai-in-medicine?sh=6c1c07e5.
2 Grand View Research. Artificial Intelligence in Healthcare Market Size, Share, and Trends Analysis Report. By Component (Software Solutions, Hardware, Services) BY Application (Virtual Assistants,
Connected Machines). By Region, And Segment Forecasts 2023-2030. Published online.2023. https//www.grandviewresearch.com/industry-analysis/artificial-intelligence-ai-healthcare-market.
3 Perspective. Is Medicine Ready for AI? N Engl J Med 2023:388: e49. Kohane at 3. https://www.nejm.org/doi/full/10.1056/NEJMp2301939
4 Finlayson, Samuel G Ph. D, et al. The Clinician and Dataset Shift in Artificial Intelligence. Correspondence. N Engl J Med: 385:283-286. DOI: 10.1056/NEJMc2104626.
5 U.S. Food and Drug Administration. Artificial Intelligence/Machine Learning (AI/ML)-Based Software as a Medical Device (SaMD) Action Plan. January 2021. https://www.fda.gov/media/145022/
download.
6 Expert Panel Forbes Technology Council. 19 Ways AI May Soon Revolutionize The Healthcare Industry (forbes.com). https://www.forbes.com/sites/forbestechcouncil/2023/04/18.19-ways-ai-maysoon-
revolutionize-the-healthcare-industry/?sh=5ft12724f1bb8.
7 Powell, Alvin. AI revolution in medicine. The Harvard Gazette. https://news.harvard.edu/gazette/story/2020/11/risks-and-benefits-of an ai-revolution-in-medicine/. Pp.1-16 at 8. November
11.2020.
8 Newman Toker, David E MD et.al. Serious misdiagnosis-related harms in malpractice claims: The “Big Three” – vascular events, infections and cancers. Diagnosis 2019; 6(3): 227-240.
9 Newman Toker, David E MD et.al. Burden of serious harms from diagnostic error in the USA BMJ Quality & Safety Published Online First: 17 July 2023. doi: 10.1136/bmjqs-2021-014130.
10 Expert Panel Forbes Technology Council. Jennifer Gold. Apollo Information Systems comment.15. Streamlining Administrative Tasks.
11 Nicholson Price II, W. Risks and remedies for artificial intelligence in health care. Brookings. November 14, 2019. https://brookigs.edu/articles/risks-and-remedies-for-artificial -intelligence-inhealth-
care.
12 Beam, Andrew PhD., et al. Artificial8d8. Intelligence in Medicine. N Engl J Med 2023: 388:1220-1221. DOI: 10.1056/NEJMe2206291.