Aug 4, 2026 · 33 min · 16 segments
Utilization managers (UMs) have long served as a resource for clinicians determining the appropriate patient status. Key considerations include which clinicians are involved, how communication…
And today I'm gonna talk about how it is time to finally retire the 837 and 835 filing system, that artificial intelligence can replace America's electronic billing maze.
For more than three decades, and I know because I've been doing it, the healthcare industry has vested billions of dollars in electronic billing.
Yet despite these investments, hospitals, physicians, and payers continue to spend extraordinary amounts of money and time simply trying to determine whether a service should be paid and how much should be paid.
The question is whether the architecture we built in the 1990s makes sense in the era of artificial intelligence.
The HIPAA Administrative Simplification Provision of 1996 established national standards for electronic healthcare transactions.
Providers were replacing paper claims, multiple proprietary formats, and manual payments postings with a common electronic language, but there was one significant limitation.
Today's electronic claims often contains hundreds of data elements, modifiers, occurrence codes, condition codes, value codes, diagnosis pointers, service lines, revenue codes, adjustment codes, and payer specific edits, and every payer adds another layer of proprietary requirements.
Clearing houses perform additional edits, and then revenue cycle vendors like me perform yet another round of edits before the claim is ever submitted.
After adjudication, the payer responds with an equally complex 835 remittance advice that must be interpreted, matched to the original claim, posted into the patient accounting system, analyzed for denials, and then is frequently appealed.
Entire industries have developed simply to translate one electronic document into another, and I have financed my children's education through that.
The United States spends more money administering healthcare payments than any other country in the world.
Software vendors have built billion-dollar businesses around claim editing, payment reconciliation, remittance posting, denial prediction, and revenue cycle analytics.
Artificial intelligence fundamentally changes this model, and large language models and reasoning systems no longer require rigid documented structures to understand healthcare information.
An AI system can simultaneously review the patient's medical record, the physician documentation, laboratory and imaging results, coverage policies, medical necessity, prior authorization information, provider contracts, government regulations, and historical payment patterns.
The AI would determine whether the documentation supports the service, whether the medical necessity exists, whether coding is appropriate, which payer rules apply, the appropriate reimbursement, and additional documentation that may be needed before payment.
One of healthcare's greatest inefficiencies is the endless cycle of denials and appeals.
Claims are often denied because one data an- element is missing, one modifier is omitted, one diagnosis pointer is incorrect, or one payer specific edit failed.
And today I'm gonna talk about how it is time to finally retire the 837 and 835 filing system, that artificial intelligence can replace America's electronic billing maze.
For more than three decades, and I know because I've been doing it, the healthcare industry has vested billions of dollars in electronic billing.
Yet despite these investments, hospitals, physicians, and payers continue to spend extraordinary amounts of money and time simply trying to determine whether a service should be paid and how much should be paid.
The question is whether the architecture we built in the 1990s makes sense in the era of artificial intelligence.
The HIPAA Administrative Simplification Provision of 1996 established national standards for electronic healthcare transactions.
Providers were replacing paper claims, multiple proprietary formats, and manual payments postings with a common electronic language, but there was one significant limitation.
Today's electronic claims often contains hundreds of data elements, modifiers, occurrence codes, condition codes, value codes, diagnosis pointers, service lines, revenue codes, adjustment codes, and payer specific edits, and every payer adds another layer of proprietary requirements.
Clearing houses perform additional edits, and then revenue cycle vendors like me perform yet another round of edits before the claim is ever submitted.
After adjudication, the payer responds with an equally complex 835 remittance advice that must be interpreted, matched to the original claim, posted into the patient accounting system, analyzed for denials, and then is frequently appealed.
Entire industries have developed simply to translate one electronic document into another, and I have financed my children's education through that.
The United States spends more money administering healthcare payments than any other country in the world.
Software vendors have built billion-dollar businesses around claim editing, payment reconciliation, remittance posting, denial prediction, and revenue cycle analytics.
Artificial intelligence fundamentally changes this model, and large language models and reasoning systems no longer require rigid documented structures to understand healthcare information.
An AI system can simultaneously review the patient's medical record, the physician documentation, laboratory and imaging results, coverage policies, medical necessity, prior authorization information, provider contracts, government regulations, and historical payment patterns.
The AI would determine whether the documentation supports the service, whether the medical necessity exists, whether coding is appropriate, which payer rules apply, the appropriate reimbursement, and additional documentation that may be needed before payment.
One of healthcare's greatest inefficiencies is the endless cycle of denials and appeals.
Claims are often denied because one data an- element is missing, one modifier is omitted, one diagnosis pointer is incorrect, or one payer specific edit failed.
The rest of this transcript — segmented and speaker-labeled, so you land on the exact moment something was said
Search every transcript — by keyword, by phrase, or by meaning, across every show Radar indexes
Trends — what is surging across podcasts, measured against its own baseline
Alerts — when a name you follow appears in a newly indexed episode
No account is needed to search Radar.
