News Center

Originally published in The Tennessean, January 2026

After you get care, providers use codes to describe diagnoses and services. These codes tell health plans what care was delivered and determine how much is paid. They can also affect what patients and employers pay through premiums and out-of-pocket costs. 

Many large hospitals and provider groups now use AI tools to help manage their revenue cycles. These AI tools read clinical notes and suggest to the provider or the coder other diagnoses and codes that could be submitted on a claim. Used appropriately, this technology can improve efficiency and accuracy. But, it can also drive costs higher, without improving patient care. 

One example is the concerning pattern known as “discordant treatment.” These are cases where a diagnosis is billed, but the medical record does not show treatment that would normally accompany that condition.  

At BlueCross BlueShield of Tennessee, we saw costs for a specific blood problem in new mothers go up by more than $2.6 million between 2021 and 2024.

Yet during that same period, treatments expected for this blood problem remained relatively flat. This means more money was paid for the diagnosis, even though the treatments typically expected for that condition did not increase. 

This is not an isolated example. Nationally, similar patterns are emerging with sepsis, a serious and life-threatening condition. More patients are being coded with sepsis at billing, which triggers higher payments. But, many of these patients didn’t receive treatment consistent with that diagnosis and often had shorter hospital stays.  

Some AI tools also rely on “ambient listening” to record conversations during clinical visits with a health provider and flag certain words or phrases to suggest diagnoses. Once the audio is deleted, it becomes difficult to verify what was actually said or whether a diagnosis was clinically appropriate.  

These trends raise an important question: are higher costs being driven by sicker patients, or because AI is finding ways to bill more without providing care that matches those charges? And how can we be sure? 

Why it matters and what we’re doing about it 

Health insurance works by spreading costs across members. When diagnostic coding gets more intense without more care, those added costs don’t disappear. They show up in higher premiums for employers and individuals. Depending on health coverage, patients may also face higher out of pocket costs that don’t reflect the care they actually received. 

While most providers use billing tools appropriately, we’re working to make sure payments for health care services are fair for both providers and members.  

For example, we look at providers who submit more than half of their claims with more complicated or intensive codes. Since those claims would be reimbursed at higher rates, we review them to make sure the diagnoses and treatment levels submitted are aligned. Only a small fraction of claims end up getting adjusted under this process, and each is reviewed by a clinician.  

We support members upfront by offering price transparency tools so they can estimate expected costs before receiving care. We also provide education and support to help members understand their bills and speak up when something doesn’t look right.  

Technology should serve patients and affordability – not quietly inflate costs.

By recognizing how advanced billing tools can contribute to rising expenses, and by working together on thoughtful safeguards, we can ensure innovation improves care without making it less affordable for everyone.  

As a health insurer, our responsibility is to strike a careful balance: ensuring providers are paid fairly for the care they deliver, while protecting our members from unnecessary costs.