Using AI to Research Insurance Company Billing, Coding, and Coverage Policies
How Artificial Intelligence Can Help Physicians Find the Information They Need to Get Claims Paid
Introduction
I recently had the opportunity to lecture on the use of Artificial Intelligence (AI) in Podiatry at three meetings: the Oregon Podiatric Medical Association, NEOAPM, and Region 1.
During a hallway conversation at one of these meetings, a Podiatrist asked me whether AI could be used to help appeal an insurance company claim denial.
It was an excellent question, and one that I felt deserved more attention. The discussion got me thinking about how AI could be used not only to assist with appeals, but also to research insurance company billing, coding, medical necessity, and reimbursement policies before a claim is ever submitted.
I felt this was important enough to write an article and share the information with the broader Podiatric community.
One of the more frustrating parts of practicing medicine is dealing with insurance company billing and coding requirements. We perform a service that is medically necessary, document the care provided, submit the claim, and sometimes receive a denial because we failed to meet an insurance company's particular requirements.
The problem is that every insurance company has its own policies. A procedure covered by one company may not be covered by another. One insurance company may require a particular diagnosis code, while another may have additional medical necessity or documentation requirements. Even different plans offered by the same insurance company may have different coverage rules.
Much of this information is available on the insurance companies' websites. The challenge is finding it.
AI can help physicians and their staff research these policies, identify the requirements that apply to a particular service, and locate the actual documents published by the insurance company.
This is one of the practical ways AI can help reduce the administrative burden of running a medical practice.
Where Do Insurance Companies Keep Their Policies?
Most insurance companies maintain extensive collections of medical policies, reimbursement policies, provider manuals, and billing guidelines on their websites.
Finding the appropriate policy is not always easy.
You may need to navigate through several menus, search for a particular procedure, or download a PDF document containing dozens of pages. Sometimes the medical necessity requirements are in one document, the coding instructions are in another, and the prior authorization requirements are somewhere else.
Some of this information is located in what is known as the deep web.
The deep web refers to information available on the internet that is not indexed by traditional search engines such as Google or Bing. This includes information contained in databases, specialized search systems, and websites that require a login.
The deep web should not be confused with the dark web. We are talking about legitimate information, including resources maintained by insurance companies and government agencies.
Not every insurance company policy is part of the deep web. Many policies can be found through a traditional search. However, some are difficult to locate because of the way the insurance company organizes its website.
AI tools that have internet research capabilities can help locate this information. They can also follow links to policy documents, read the information, and provide a summary of the requirements.
There is an important limitation. An AI tool cannot necessarily access information behind a login or inside a database that does not permit automated access. It must have legitimate access to the information being researched.
How Can AI Help?
Suppose you want to know whether a particular insurance company covers custom foot orthotics.
Traditionally, you would go to the insurance company's website, locate its medical policies, search for orthotics, and review any documents you find.
With AI, you can ask a question in ordinary language.
For example:
"Find UnitedHealthcare's current medical policies and reimbursement policies for custom foot orthotics. Tell me what the coverage requirements are, identify any relevant billing and coding requirements, and provide links to the original insurance company documents."
An AI tool with web research capabilities can search for the relevant policies, summarize what they say, and provide links directly to the source documents.
You can then ask follow-up questions.
Does the insurance company require prior authorization? Are there specific documentation requirements? Does the coverage depend on the patient's diagnosis? Are there frequency limitations?
Instead of beginning a new internet search for every question, you can continue the conversation with AI.
The advantage is that you do not need to know how the insurance company organizes its website. You describe what you are looking for, and AI helps locate the relevant information.
The goal is not to have AI make a coverage determination. The goal is to have AI find the actual insurance company policy so you can determine what the insurance company requires.
Finding the Actual Insurance Company Policies
When using AI for this type of research, I recommend asking it to provide a direct link to the original policy.
This is important because a summary, no matter how convincing it sounds, is not a substitute for the insurance company's published requirements.
Several major insurance companies maintain online policy libraries.
UnitedHealthcare
UnitedHealthcare maintains medical policies and reimbursement policies. These are not necessarily the same thing.
A medical policy may describe whether a service meets the company's medical necessity requirements. A reimbursement policy may address how the service is billed, which modifiers are recognized, or how the claim is processed.
UnitedHealthcare Policies and Protocols
UnitedHealthcare Commercial Reimbursement Policies
Aetna
Aetna publishes Clinical Policy Bulletins that describe coverage requirements for medical services and procedures.
These bulletins may contain clinical indications, limitations, supporting evidence, and references to CPT or HCPCS codes.
Aetna Medical Clinical Policy Bulletins
Cigna Healthcare
Cigna publishes coverage policies and claims-related guidance that can help physicians research medical necessity, coverage restrictions, and reimbursement issues.
Cigna Coverage and Claims Resources
Anthem Blue Cross and Blue Shield
Anthem maintains medical policies, clinical utilization management guidelines, and provider resources.
Because Blue Cross and Blue Shield organizations have different regional structures and contracts, it is particularly important to identify the correct company and the patient's specific plan.
Anthem New York Provider Resources
These are links to policy libraries rather than individual procedure-specific policies. They illustrate the kinds of resources AI can search. The real benefit comes when AI identifies the specific policy and gives you a direct link to that document.
Medicare Policies Can Also Be Researched with AI
Medicare has its own set of coverage and coding requirements.
When researching Medicare coverage, physicians may need to review a National Coverage Determination (NCD), a Local Coverage Determination (LCD), and a related billing and coding article.
An LCD may describe the clinical circumstances under which Medicare considers a service reasonable and necessary. The associated billing and coding article may identify the procedure codes and diagnosis codes used when submitting claims.
It is important to review both documents when applicable.
AI can help find these documents and identify the requirements that apply to the service.
For example, you could ask:
"Find the Medicare LCD and associated billing and coding article for wound debridement applicable to my Medicare Administrative Contractor. Identify the medical necessity requirements, documentation requirements, and relevant CPT and ICD-10-CM codes. Provide direct links to the Medicare documents."
The Centers for Medicare & Medicaid Services (CMS) maintains several useful resources:
Medicare National Correct Coding Initiative
The second resource is especially useful when researching coding edits, bundling, and other restrictions on reporting services together.
When researching Medicare, make sure that AI identifies the correct Medicare Administrative Contractor and the policy that was in effect on the date of service.
Comparing Policies from Different Insurance Companies
Another useful application is comparing insurance company policies.
Consider a Podiatrist who provides wound care services and participates with several insurance companies.
The physician may want to know whether the same documentation and coverage requirements apply to each insurer.
Rather than researching every insurance company separately, AI can assist with a comparison.
For example:
"Compare the coverage and reimbursement policies for wound debridement published by UnitedHealthcare, Aetna, Cigna, and Medicare. Identify differences in medical necessity, documentation requirements, coding restrictions, and frequency limitations. Provide direct links to the policies used in the comparison."
AI can organize the information into a table that makes the differences easier to identify.
This can be useful when developing office documentation protocols or training billing personnel.
However, you still need to confirm which policies apply to the patient's particular insurance product. A commercial policy may not apply to a Medicare Advantage plan offered by the same company.
Also, a medical necessity policy does not necessarily establish that a service is covered under a patient's individual benefit plan.
Using AI When a Claim Is Denied
AI can also assist with researching claim denials.
Suppose you submit a claim for a wound care procedure and the insurance company denies the claim, stating that the service did not meet its medical necessity requirements.
The first question is whether the patient's documentation actually supports the service under the applicable policy.
You could ask AI to locate the insurance company's medical necessity policy, identify the requirements, and provide the document used to support its answer.
You can then compare those requirements with the medical record.
If the documentation supports the service, AI may help organize an appeal that references the relevant policy provisions.
This does not mean every denial should be appealed. If the documentation is insufficient, the insurance company may have a legitimate basis for denying payment.
It also does not mean that AI can decide whether a denial was correct. Coverage restrictions, contract provisions, prior authorization requirements, and other factors may affect the outcome.
The benefit is that AI can help you identify the actual requirements without spending excessive time searching for them.
AI may also help you research the insurance company's appeal procedures, including applicable filing deadlines, required forms, supporting documentation, and submission instructions. These requirements may differ depending on the insurance company and the type of plan.
Be Careful: AI Can Provide Incorrect Information
One of the problems with generative AI is that it can provide an answer that sounds authoritative but is incorrect.
This is commonly referred to as an AI hallucination.
An AI system may cite an outdated policy, misunderstand a coding requirement, or even identify a policy that does not exist.
Insurance companies also change their policies. A document published today may be different from the document that was in effect six months ago.
For these reasons, I recommend including the following instructions whenever you use AI to research billing and coding requirements:
"Use official insurance company or government sources. Identify the policy title, policy number if available, effective date, and the specific section supporting your answer. Provide a direct link to the original document. Do not guess. If you cannot verify a requirement, tell me that the information could not be confirmed."
You should also ask AI to distinguish between information it actually retrieved from a document and information it found in a search result but could not independently review.
Remember that a medical necessity policy, a reimbursement policy, a provider contract, and the patient's benefit plan are not interchangeable.
Each may have a different role in determining whether the insurance company will pay the claim.
What About HIPAA?
In most cases, you should be able to research insurance company policies without providing any protected health information.
You do not need to provide a patient's name, date of birth, medical record number, or insurance identification number to ask whether an insurer has a medical necessity policy for a particular procedure.
You may need to identify the insurance company, insurance product, CPT code, diagnosis, and date of service. This information can often be provided without identifying the patient.
If you intend to have AI review an actual patient's medical record or claim, you need to consider the Health Insurance Portability and Accountability Act (HIPAA).
A publicly available AI application should not automatically be considered appropriate for processing protected health information. The physician must determine whether the service has the appropriate security safeguards and whether a Business Associate Agreement is required.
For researching publicly available insurance policies, there is generally no reason to expose patient information.
The Practical Benefit for Physicians
Physicians already spend too much time dealing with administrative requirements.
Understanding insurance company policies is necessary to submit claims properly, avoid preventable denials, and comply with billing and coding requirements.
The information we need is often available. The problem is finding it and determining which requirements apply.
AI provides another way to conduct this research.
Instead of searching through multiple websites and lengthy documents, you can ask AI to find the relevant policies, explain their requirements, compare insurance companies, and provide direct links to the original sources.
You still need to review the information and make sure it applies to the specific insurance plan, procedure, and date of service.
But you no longer need to do all the searching manually.
AI does not replace your responsibility to understand billing and coding requirements. It can make finding and understanding those requirements much easier.
For physicians looking for practical ways to incorporate AI into their practices, researching insurance company policies is a useful place to start.
Disclosure
I am providing consulting services to many companies who are building AI tools to assist with billing and coding.
Read Comments