Announcement: SOA congratulates the new ASAs and CERAs for May 2022.

Ready, AIM, Fire! How Artificially Intelligent Medicine Will Impact the World of the Actuary

By Tyler M. Doiron

Actuary of the Future, May 2021


As most of us will agree, Actuaries far and wide pride themselves on their ability to adapt and respond to their ever-changing environment. In the last decade, the actuarial world has shifted dramatically due to new advancements in the insurance industry through innovation, strategizing and technology. The same can be said for numerous other professional industries such as accounting, law, education, academia, engineering, and medicine. As far as pure growth and innovation goes, medicine has set the standard; and did they ever set it high.

Recent leaps in innovation in medical technologies has turned the world upside down, for the greater good of humanity. Medicine has taken an initial approach to the idea of Artificially Intelligent medicine (AIM) where artificially intelligent (AI) computers help physicians take on patient care to provide a whole new perspective to medicine in an autonomous manner. AIM does not in any way intend to remove the physician but is to be used as a tool to help to improve physician capabilities by improving the ability to diagnose and treat illness and disease.

To better grasp AIM, it is important we understand the idea of autonomy. To be autonomous is to (a) have the right or power of self-government, (b) to respond, react, or develop independently of the whole or (c) be capable of existing independently. Modalities of autonomous detection of injury, illness and disease in diagnostic radiology have been at the forefront of modern medicine over the last decade.

The idea of truly autonomous medicine is daunting. As someone with a background in AI and its application in biology, who only recently made the jump to the actuarial world, I truly believe we are at a crossroad in medicine. Do we take the avenue of venture, or do we sit in the shadows of stagnation and rely on dated technologies and the inevitable possibility of human error?

For me, the answer is easy: Practicing physicians in the field of diagnostic radiology must be willing to take a step back from its heavy reliance on the human eye to diagnose and revolt against defensive medicine and begin increasing reliance on artificial intelligence to provide the highest level of medical care possible. A large proponent to this is to aid the process of decreasing physician liability to help improve patient care through mitigating defensive medicine; the idea that physicians will act to protect their liability rather than practice medicine on the offensive. In reciprocal, this generally leads to physicians overspending on testing and has a direct impact on patient care. As actuaries, we should be willing to support them through risk management and experience data studies on health and life insurance data in regions that have taken on the bull that is AIM by the horns, so to speak.

AIM is beautiful. To be able to self-learn, self-direct and adapt to an environment to produce the highest level of precision required for a given task could be crucial in the fight against illness and disease. Fundamentally, the way in which AI is used in diagnostic medicine is by use of computer vision (CV), machine learning (ML), neural networks (NN), and natural language processing (NLP).[1] Each are very different subcategories of artificial intelligence and are primarily used as a learning method in various diagnostic tools in which computers can functionally, analytically, or interactively detect deviations from what is deemed normal. Computers have an immense capability to understand patterns that the human eye may not see. With utmost precision, CV-AI technology greatly surpasses the ability of the human eye to detect bone fragmentation and other cartilage damage by up to 1000 times in some cases in x-ray analysis.[2] Other AIM technologies have shown impressive accuracy and sensitivity in identification of MRI, PET and CAT imaging abnormalities and promises to enhance tissue-based characterization.[3]

As I had previously eluded to, actuaries are constantly evolving to innovate and adopt newly integrated technologies that may impact many facets of our present and future line of work. The physician battle with AIM will surely produce a trickledown effect of changes to the actuarial world. It is crucial to get ahead of the game to prepare; to understand and to model the effects of AIM to medicine and other side effects of increased patient care within the health care system.

To many on the outside looking in, improvements in the health care system simply equate to better diagnosability and prognosis. To an actuary, this also holds true, but this same picture is painted more in depth than what meets a non-actuary’s eye; improvements in mortality trends, changes to health insurance coverages, decreases in medical malpractice insurance claims, changes to insurability clauses, new subcategories of health insurance, and the list goes on. Prudence in the actuarial world is crucial moving forward as AIM changes medicine.

Across the globe, actuarial researchers and clinicians alike have been studying the early integration of AI and its impact on their respective lines of work. Many notable studies have shown that AI integration will shift much of radiology and the insurance industry as we know it.[4, 5, 6, 7] As actuaries, arguably the largest impacted area of our work is in the interest of medical malpractice insurance claims. (Here, it is important to note that various countries may have different medical liability issues. In Canada for example, medical malpractice claims are funded through government subsidized programs such as the Canada Medical Protective Association. Areas of government intervention generally led to a lesser amount of paid malpractice claims than with private insurance as seen in the United States.[8])

As mentioned in Hamer et al., the discussion around liability in radiology first peaked in the mid-1980s after a study by Curran et al. in the United States showed a significant number of misdiagnosed malignancies in lung and colon carcinomas by radiologists.[9, 10] Almost 30 percent of all malpractice claims against radiologists stem from radiological test reading/interpretation errors deemed human interpretation and reading errors; an area that CV based technologies have shown significance signs of growth in improving patient care.[11]

Since the initial study, historical data show a significant pattern of growth of medical malpractice claims against human error until the early 2010s, in which a large decline began largely because of the early phase integration of CV-based AI in physician interpretation of various radiological diagnostic tests.[6] Now, significant studies have begun to show with preliminary data that AI directly decreases physician likelihood to practice defensive medicine while concurrently improving patient care and prognosis.

Actuaries help provide the backbone of risk management just like physicians help provide the backbone of the health care system. With a greater understanding of the impacts of AI usage in medicine, we will be able to provide not only greater health care to patients, but remove physician liability in various fields of medicine such as radiology, to provide a stronger backbone in the health and life insurance sectors and stay at the forefront of innovation in our ever-changing world.

Statements of fact and opinions expressed herein are those of the individual authors and are not necessarily those of the Society of Actuaries, the editors, or the respective authors’ employers.

Tyler Doiron is an actuarial analyst at the Blue Cross Life Insurance company of Canada. He can be contacted at


[1] Zach Harned, Matthew P. Lungren & Pranav Rajpurkar, Comment, Machine Vision, Medical AI, and Malpractice, Harv. J.L. & Tech. Dig. (2019),

[2] Hosny A, Parmar C, Quackenbush J, Schwartz LH, Aerts HJWL. Artificial intelligence in radiology. Nat Rev Cancer. 2018 Aug;18(8):500-510. doi: 10.1038/s41568-018-0016-5. PMID: 29777175; PMCID: PMC6268174.

[3] Jiang, H. Ma, W. Qian, M. Gao and Y. Li, "An Automatic Detection System of Lung Nodule Based on Multigroup Patch-Based Deep Learning Network," in IEEE Journal of Biomedical and Health Informatics, vol. 22, no. 4, pp. 1227-1237, July 2018, doi: 10.1109/JBHI.2017.2725903.

[4] Jiang F, Jiang Y, Zhi H, et al. Artificial intelligence in healthcare: past, present and future. Stroke and Vascular- Neurology 2017;2:doi: 10.1136/svn-2017-000101

[5] Ahuja AS. The impact of artificial intelligence in medicine on the future role of the physician. PeerJ. 2019;7:e7702. Published 2019 Oct 4. doi:10.7717/peerj.7702

[6] Wolff J, Pauling J, Keck A, Baumbach J, The Economic Impact of Artificial Intelligence in Health Care: Systematic Review. J Med Internet Res 2020;22(2):e16866. DOI: 10.2196/16866

[7] Thesmar, D., Sraer, D., Pinheiro, L. et al.Combining the Power of Artificial Intelligence with the Richness of Healthcare Claims Data: Opportunities and Challenges. PharmacoEconomics 37, 745–752 (2019).

[8] Busardò FP, Frati P, Santurro A, Zaami S, Fineschi V. Errors and malpractice lawsuits in radiology: what the radiologist needs to know. Radiol Med. 2015 Sep;120(9):779-84. doi: 10.1007/s11547-015-0561-x. Epub 2015 Jun 27. PMID: 26116141.

[9] Hamer MM, Morlock F, Foley HT, Ros PR. Medical malpractice in diagnostic radiology: claims, compensation, and patient injury. Radiology. 1987 Jul;164(1):263-6. doi: 10.1148/radiology.164.1.3588916. PMID: 3588916.

[10] Curran WJ. Closed-claims data for malpractice actions in the United States. Am J Public Health. 1981 Sep;71(9):1066-7. doi: 10.2105/ajph.71.9.1066. PMID: 7270777; PMCID: PMC1619855.