AI (artificial intelligence) uses computer systems to simulate human thinking, and this technology has found its way into Medicare Advantage Plans. Insurance companies increasingly use AI tools to handle administrative tasks traditionally performed by humans, like medical reviews and prior authorizations.
These AI tools have many potential benefits, like increasing efficiency, but they are not without fault. Learn more about the AI pitfalls and how they can impact Medicare beneficiaries.
Table of Contents
Why Medicare Advantage Companies Use AI Tools
For insurers, the main advantages of using AI tools are increased efficiency and lower costs. Medicare Advantage Plans often require prior authorization for certain services, meaning beneficiaries need approval for their plan before services are covered. With AI tools, it’s possible to automate prior authorizations and receive decisions instantly.
AI can perform around 50 to 75% of prior authorization tasks (the approval or denial of a treatment or procedure), which allows human staff members to focus on more complex cases. AI can also help insurers make more consistent decisions, ensuring that plan rules are applied fairly to members with similar conditions.
Beneficiaries can benefit as well. When routine prior authorizations are faster, medically necessary care can be started sooner. Automated decisions may help Medicare Advantage beneficiaries receive necessary cancer care at a lower cost.
Potential Pitfalls of AI Use
While Medicare Advantage beneficiaries may benefit from AI tools, there are also some potential pitfalls of this new technology. Here’s a look at some of the main concerns.
Incorrectly Rejected Requests
Medicare Advantage Plans are required to cover the same medically necessary services as Original Medicare. However, 13% of the authorizations denied met Medicare’s rules, meaning they likely would have been covered under Original Medicare.
Advocates say that AI tools use stricter coverage rules than Original Medicare and err on the side of denying requests. For example, a plan might have a rule that members need to get an X-ray before other imaging tests are covered.
Coverage Ending Too Early
Medicare Advantage Plans tend to use prior authorization for relatively expensive health services like nursing facility stays, home health care, and inpatient hospital stays. They use AI tools to predict where beneficiaries need to receive care and how long they need to stay.
Beneficiaries may not be ready to leave a nursing facility or hospital as soon as an AI tool predicts. In recent months, some major insurers are facing lawsuits from plan members who allege their care was cut off too early.
One-Size-Fits-All Assessments
Medicare requires individualized assessments to determine whether or not a beneficiary needs skilled nursing, home health, or outpatient therapy benefits. The assessment looks at the beneficiary’s health status and the medical necessity of the treatment.
AI-powered tools analyze large groups of previous patients to estimate current patients’ care needs. The tools may not consider the current beneficiary’s actual health needs, resulting in lower-quality care.
Bias and Discrimination
Developers of AI tools choose the patient data they use to train and test the algorithms. There is limited information available about the patient data. Advocates are concerned that the data sets use groups of people that are too similar, meaning the tools may not perform as well for people of other races or sexes.
Further, some are warning that AI-powered tools used by health insurers may have an especially negative impact on people with disabilities. That’s because AI tools may see treatments that do not improve functioning or health as unnecessary, even if the treatments could improve quality of life.
High Out-of-Pocket Costs
If a Medicare Advantage claim is improperly denied or a covered service is ended too soon, beneficiaries may need to pay for the services they need out of pocket. High out-of-pocket costs for medical services can strain beneficiaries’ finances or even keep them from being able to access the care they need. However, beneficiaries have a right to appeal, and insurers may agree to pay the bills on appeal.
How Medicare Is Responding
The Centers for Medicare and Medicaid Services (CMS) recently released new guidelines addressing some concerns about Medicare Advantage Plans and AI tools. Here’s an overview of the rules for Medicare Advantage Plans:
- Plans must follow Medicare’s coverage rules. Medicare Advantage Plans are required to follow the general coverage rules outlined in Original Medicare, as well as any relevant national or local coverage determinations.
- Plans can only set their own coverage rules in limited situations. In cases where Medicare’s rules are not fully established, plans can apply their own criteria. The criteria must be publicly accessible.
- Plans must review policies annually. Each plan will create a committee responsible for ensuring its prior authorization decisions align with Medicare’s rules.
House Democrats Want More Oversight
In November 2023, a group of 30 House Democrats released a letter to CMS. They say the new guidelines do not go far enough to protect beneficiaries. They’re urging CMS to set more rules to shield Medicare Advantage beneficiaries from problematic AI decisions.
The representatives list a number of measures they hope CMS will put in place, including:
- Requiring plans to report prior authorization data: Information about the reason for denials by type of service can help Medicare identify potential issues and take action.
- Assessing data that goes into AI algorithms: If algorithms are trained using data from too-similar patients, their decisions may not be fair or accurate for more diverse populations.
- Determining if the algorithms are self-correcting: Some coverage denials are reversed on appeal, but it’s unclear if AI tools learn from the appeal results when making future decisions.
The Takeaway
Medicare Advantage Plans use AI-powered tools to streamline decisions traditionally made by humans, such as prior authorizations. In some cases, the AI’s decisions are problematic for one reason or another.
Beneficiaries who do not agree with their plan’s coverage decision have the right to appeal, regardless of whether the decision was made by an AI tool or a human reviewer. To start the appeals process, follow the directions in your plan’s denial notice.