Harness AI (artificial intelligence) to Beat Health Insurance Company Claim Denials

Let's face it: U.S. commercial healthcare insurance companies are incredibly greedy and they don't fundamentally give a $#!+ about patients' health. Their only concern is collecting as much money as possible in premiums while minimizing the amount of money in claims they pay, screwing the Federal government out of billions of dollars by administering Medicare with so-called "Medicare Advantage" plans known as Medicare Part C (for which they make money on/by denying care that regular Medicare would have paid without question), and collecting legally-exempted rebate kickbacks from drug companies which are paid in cash to their PBM units for preferred drug formulary placement or to prevent coverage of cheaper alternative drugs (like generics and biosimilars) via inappropriate, kickback-driven "formulary exclusions". 

As healthcare insurance company whistle-blower Wendell Potter (see https://healthcareuncovered.substack.com/p/unitedhealth-group-the-biggest-private for his write-up) reported:

UnitedHealth's total revenues for the first three months of 2024 were just shy of $100 billion, an increase of 8.6% from the $91.9 billion the company reported for the first quarter of 2023. Revenues have more than tripled over the past decade as the company has moved aggressively into health care delivery, government programs and the pharmacy benefits [management] space. During the same months in 2014, revenues totaled just $31.7 billion. Back then UnitedHealth was the 14th largest U.S. company. It is now the fifth largest, having leapfrogged over companies like Ford and General Motors.

A big driver of United Health's growth has been the taxpayer-supported Medicaid and Medicare programs. Enrollment in the Medicaid plans managed by UnitedHealth has increased nearly 80% since 2014, from 4.3 million to 7.7 million. Enrollment in its Medicare Advantage plans has more than doubled over the same period, from 2.99 million in 2014 to 7.76 million. While the growth in Medicare Advantage slowed this year, UnitedHealth, along with Humana and CVS/Aetna, have captured 86% of new enrollees in the privatized version of Medicare so far this year. Medicaid enrollment declined by 165,000 as states began eliminating coverage for many low-income families that had been covered temporarily with increased federal funding during the pandemic.

In a different article, (see https://healthcareuncovered.substack.com/p/unitedhealths-self-dealing-is-accelerating for that write-up), Wendell Potter added:

Optum is a dominant provider of [which includes Optum's physicians, PBM, claims integrity processing, and revenue cycle management] all these other services. One subdivision, Optum Health, acquired or hired 20,000 physicians [practices] over the past year and now has 90,000 on its roster, or about 10% of all doctors in the U.S. Another subdivision, OptumRx, is the third-largest PBM, a kind of company that comes up with and administers drug benefit plans for insurers, or about 22% of the market. It's also the fourth-largest pharmacy, with about 7% of the market.

Of course, the preceding three paragraphs are exclusively about UnitedHealth, but its big rivals include CVS Health's retail pharmacy chain (the country's biggest) and its Aetna insurance business as well as its Caremark PBM unit plus ancillary businesses such as CVS Healthspire and Oak Street Health, while Cigna's business unit branded as Evernorth operates both the PBM Express Scripts as well as an entity known as MD Live. 

My point is that these insurance companies are giants and they don't really care about whether you receive care your employer pays premiums for; in fact, they would prefer you did not receive any care. They make more money by denying care.

An excellent overview of the insurance company appeals process can be read at https://t1dexchange.org/denied-by-insurance-a-pharmacist-tells-you-how-to-appeal/ and is worth a read. In spite of the ability to appeal insurance company denials, it's still a hassle to deal with. That's where some emerging innovation can make the task a lot easier.

As a point of reference, the nonprofit journalism wire service known as ProPublica has built a tool to help patients to find out the reason a health insurance has been denied a particular claim, and for the patient to see the actual reason which is listed in the insurance company's denial notes. To use this tool, visit https://projects.propublica.org/claimfile/ and follow the instructions. Most of these bogus claims denials can be attributed to a third-party company EviCore, which is a business unit of Cigna's PBM unit known broadly as Evernorth (which also owns/operates Express Scripts) which often denies prior authorization requests written by doctors. To read ProPublica's coverage of how EviCore operates, visit https://www.propublica.org/article/evicore-health-insurance-denials-cigna-unitedhealthcare-aetna-prior-authorizations.

Personally, I have only occasionally had routine care denied for vague and non-specific reasons such as "Not Medically Necessary" (especially when a doctor prescribed it and no one from the insurance company examined me, so who are they to question my own doctor?). Fortunately, such vague denial reasons tend to be fairly easy to dispute via appealing those decisions, but appeals are a time-consuming hassle to deal with and many people do not bother.

What if there was a way to make that process easier?

Meet the "Fight Health Insurance" AI Tool

Well, now there is something which harnesses the power of artificial intelligence (AI) which is still in beta, but may be poised to make the lives of covered individuals easier.

On August 23, 2024, The San Francisco Standard reported in an article entitled "'Make your health insurance company cry': One woman's fight to turn the tables on insurers" (see https://sfstandard.com/2024/08/23/holden-karau-fight-health-insurance-appeal-claims-denials/ for the article) about a San Francisco tech worker named Holden Karau's efforts to harness AI to manage the laborious task of writing appeals letters.

"Instead of passively accepting the providers' decisions, she'd spend hours writing letters and filling out forms to appeal. It usually worked: Out of roughly 40 denials, she won more than 90% of her appeals, she estimates.

She began helping friends file appeals, too, then asked herself a question that's typical for engineers: Could she figure out a way to automate the process?"

After a year of tinkering, she just launched her answer: Fight Health Insurance https://fighthealthinsurance.com/, an open-source platform that takes advantage of large language models to help users generate health insurance appeals with AI.

 

Today, Fight Health Insurance's "About Us" page describes itself this way:

"We at Totally Legit Co (of Delaware) decided to create 'Fight Health Insurance' (sometimes called by...more colorful names) because we're tired of dealing with health insurance denials. In our opinion, many health insurance denials are bogus, and the insurance companies depend on us not appealing them. We hope that by making it easier to appeal medical health insurance denials, we can change that calculus on insurance companies' part."

It added: 

"Fight Health Insurance is currently a passion project. We have day jobs, but most of us have had terrible experiences with health insurance and we're tired of watching health insurance companies grind down our friends (and tired of being ground down by them ourselves)."

"Fight Health Insurance" even references Breakthrough T1D (fka JDRF) on the website under the category of "Other Resources" specifically for diabetes. I wonder if Breakthrough T1D knows they are referenced here? If so, it would seem to be a good talking point for the organization.

In order to use the "Fight Health Insurance" AI tool, the first step involves scanning your denial letter. This needs to be done using optical character recognition (OCR) software either on your device/phone (recommended for increased privacy) or on "Fight Health Insurance" servers. OCR makes the text in the denial letters understandable to machines, allowing its generative AI model to produce potential appeals letters for the patient to choose from.

As for OCR software applications, there are quite a number of those out there. My preference is to use free tools when possible. Google has several OCR tools of its own (Google Workspace has a fee, although it has others at no charge), but I also have Microsoft Office (these days, Microsoft sells a subscription-based software model for Office software [Word, Excel, PowerPoint]), but somehow I managed to find a one-time purchase version which I prefer), hence I went with the Microsoft OCR tool which is known as "Microsoft Lens" which I downloaded on my smartphone because it was free with MS Office (and I trust Microsoft slightly more than I trust Google with such info), hence I downloaded that on my iPhone which is where I typically take pictures of printed denial letters from my insurance.

"Fight Health Insurance" Outlines Steps to Overturn Insurance Denial

1. After scanning the denial letter using OCR software, load the OCR-scanned denial letter into the Fight Health Insurance website
   
   
2. Answer questions based on health conditions
   
   
3. Within seconds, the Fight Health Insurance AI program will generate an appeal letter with accurate information and circumstances
   
   
4. Then, you would submit the appeal letter along with the compiled documents to the insurance company
   
   
5. Follow up with the insurance company regularly to track the status of your appeal. Often, they're required [under state insurance laws] to respond in reasonable [sometimes as defined by state law; in the state where I now live, healthcare insurance companies are mandated to issue a decision within 60 days of the receipt of all necessary information].

I recently discovered the creator Holden Karau was interviewed by the news startup NewsNation, and the interview with her can be found below, or at https://youtu.be/lI26LrDU2dg?si=wdyV5UPHiRuNUnze:

To be sure, these are six steps which still involve a fair amount of work, although the AI part makes it faster, easier and more efficient for patients. Plus, the use of AI in dealing with commercial healthcare insurance company coverage [or denial] decisions is still very new. But perhaps it is not ironic that the commercial healthcare insurance companies are reportedly rendering decisions on denials using AI themselves, hence it seems to me that patients being able to deploy similar technology in order to fight seemingly baseless denials (like "Not Medically Necessary") is an appropriate use for this new technology.