Would you trust your care to a machine? Human beings are complex. That’s why an algorithm or AI predictive technology can’t figure us out.

Medical Advancements Over the Years

Medicine has come a long way in the past 50 years or so. Some of your grandparents or great grandparents may not have had the benefit of things that are used every day.

Today, one in five deaths in the United States is attributable to heart disease. How many more would there be if pacemakers weren’t developed in 1958, coronary bypass surgery in 1961, and cardiac stents in 1986? Second only to heart disease, cancer accounts for more than 600,000 deaths every year. It’s hard to imagine there wasn’t chemotherapy to treat it until 1961 or that there weren’t CT scans or MRI scans to help diagnosis it until 1971 and 1977.

Dialysis (1943). Organ transplants (kidney 1954, liver 1967, heart 1968, pancreas 1968, lung 1983). Immunotherapy (1986). Gene therapy (1990). Robotic-assisted surgery (1997). mRNA vaccines (2020). The list of life-saving treatments goes on and on.

AI predictive technology may be the next big thing. How ironic that it could deny you some of these other services!

AI Potential in Medicine

If you are an optimist, you see AI as an opportunity to decrease medical errors. After all, it seems that a computer could be fed all the latest medical news and research, making it a reliable repository of medical data. In that way, it could be a helpful tool to diagnose less common conditions, assess risk, and support decisions made by your healthcare provider.

Of course, the results are only as good as the information the AI is trained on. Unfortunately, bias can be a huge problem. If an algorithm is based on data that represents certain populations more than others (i.e., differences based on gender, race, and more), it could skew the results. Worse, it could do literal harm. That is especially true when the AI can’t take personal circumstances into account.

Unfortunately, we’re starting to see that already.

AI Misuse in Medicine

Case in point. The National Eating Disorder Association (NEDA) replaced human beings on their eating disorder hotline with an AI chatbot called Tessa back in May 2023. After the chatbot gave dangerous advice that could put patients in crisis (you don’t tell someone with an eating disorders to cut 1000 calories from their diet!), they quickly reverted back to the human-run hotline in early June. Some things simply require a human touch.

We have already seen AI misused in the insurance industry too. Cigna is being sued because they used an algorithm-based program called PXDX to deny coverage for more than 300,000 services in just two months. Physicians assigned to review those cases didn’t bother to verify if the program was right or wrong. They just signed off on them at a rate of 1.2 seconds per case!

Now, NaviHealth, a care management company that works with a number of Medicare Advantage plans, uses a proprietary algorithm-based program called nH Predict to decide how long people may need certain types of services, including nursing home care. It seems that many insurers are using these predictions to deny coverage after a certain date. The problem is that nH Predict does not look at someone’s individual circumstances. Many people have been denied care, even if traditional Medicare would have covered it. It’s not right.

CMS Takes Action

The Centers for Medicare and Medicaid Services has taken notice and has issued a new ruling that will change how predictive technologies can be used with Medicare Advantage plans. It will take effect on January 1, 2024.

1. Care cannot be denied if it would be covered by Original Medicare. That should have been the case anyway, but it is finally being enforced.
2. If an algorithm or predictive technology suggests that a service is not medically necessary, the plan must take it a step further. Every possible denial for coverage “must be reviewed by a physician or other appropriate health care professional with expertise in the field of medicine or health care that is appropriate for the service at issue.” For example, this means you cannot have a psychiatrist reviewing the need for orthopedic surgery.
3. Medicare Advantage plans could potentially face penalties if they are not compliant with the rules.

Should You Trust AI?

Everyone has a different take on AI. A Pew Research survey in 2023 found that 60% of people would be uncomfortable if their own doctor used AI to care for them. Those numbers increase to 67% if AI were to manage their pain or 79% their mental health. Some things require empathy that a machine simply cannot provide.

Personally, I am hesitant to turn medicine over to AI full out. Yes, there are areas where it may be able to improve outcomes and make healthcare more efficient. Still, we need to take it slow and develop it with vigorous human oversight. Medicine is based in science but it is also an art, a calling. We need the voices of living breathing men and women who care about people of all races, genders, and creeds, who have their best interests at heart, and who swear on oath to do no harm.

60% of Americans Would Be Uncomfortable With Provider Relying on AI in Their Own Health Care. Pew Research Center Science & Society. https://www.pewresearch.org/science/2023/02/22/60-of-americans-would-be-uncomfortable-with-provider-relying-on-ai-in-their-own-health-care/

An eating disorders chatbot offered dieting advice, raising fears about AI in health. (2023). NPR. https://www.npr.org/sections/health-shots/2023/06/08/1180838096/an-eating-disorders-chatbot-offered-dieting-advice-raising-fears-about-ai-in-hea

Common Cancer Sites – Cancer Stat Facts. (2023). SEER. https://seer.cancer.gov/statfacts/html/common.html

Feds Rein In Use of Predictive Software That Limits Care for Medicare Advantage Patients. (2023). KFF Health News. https://kffhealthnews.org/news/article/biden-administration-software-algorithms-medicare-advantage/

Heart Disease Facts. (2023). Centers for Disease Control and Prevention. https://www.cdc.gov/heartdisease/facts.htm

Learn about the history of transplant – OPTN. (2023). HRSA. https://optn.transplant.hrsa.gov/patients/about-transplantation/history/

Medicare Program; Contract Year 2024 Policy and Technical Changes to the Medicare Advantage Program, Medicare Prescription Drug Benefit Program, Medicare Cost Plan Program, and Programs of All-Inclusive Care for the Elderly. (2023). Federal Register. https://www.federalregister.gov/documents/2023/04/12/2023-07115/medicare-program-contract-year-2024-policy-and-technical-changes-to-the-medicare-advantage-program

‌Suzanne Kisting-Leung and Ayesha Smiley individually and on behalf of all other similarly situated vs. Cigna Corporation, Cigna Health and Life Insurance Company. (2023). https://www.documentcloud.org/documents/23885679-cigna-complaint?responsive=1&title=1