Aug 18, 2026 · 32 min · 17 segments
Like the virtual line between actors and the audience, an invisible line exists between doctors and patients. Yet, Dr. Juliet B. Ugarte Hopkins will explore the sharp contrast in how pediatric and…
And today I'm going to talk about how AI needs claims data, but it can't just take it.
So we all may remember how ChatGPT started by reading everything known to mankind in order to learn what it needed to know.
But artificial intelligence companies developing revenue cycle tools need large volumes of real claims data.
Two of the most valuable sources are 837 electronic claims and 835 electronic remittance files.
The 837 reports is what providers have billed and the 835 reports are how the payers adjudicated the claim.
including payments, payment details, adjustments, deductibles, coinsurance.
When properly matched, these transactions help train AI to identify underpayments, predict denials, and estimate collectability and recognize payer behavior.
But where can AI companies obtain the data? The most practical sources are organizations already authorized to receive and maintain transactions like hospitals and physician groups, health plans, healthcare clearinghouses, revenue cycle management companies, and companies purchasing or financing healthcare accounts receivable.
All AI companies can contract with one of these organizations to develop or operate defined applications because the work may involve creating, receiving, or maintaining or transmitting protected information.
The parties must exercise a business associates agreement or BAA specifying the permitted use of the data requiring safeguards.
Signing a BAA does not give the A company unlimited authority to use a hospital's claims.
The use must be authorized by the agreement and permitted by the HIPAA privacy rule.
Contracts should address model training directly, including whether the client data may be used for training, whether the information from different clients may be combined, who owns the trained model and the derived data, whether the vendor may retain information may retain information after the termination and how the vendor will prevent PHI from appearing in model outputs.
HIPAA recognized two identification methods, the safe harbor and expert determination.
Safe harbor requires the removal of specified identifiers Expert determination permits a qualified expert to determine that the risk of identifying the individuals is very small.
Claims may contain medical record numbers, claim control numbers, subscriber identifiers, service dates, addresses, free text fields, rare diagnosis, unusual combinations, or procedures.
Safe harbor may also eliminate dates needed to obtain timely filing or payment delay models.
An expert determination may preserve more useful relationships while still reducing re-identification risk.
And today I'm going to talk about how AI needs claims data, but it can't just take it.
So we all may remember how ChatGPT started by reading everything known to mankind in order to learn what it needed to know.
But artificial intelligence companies developing revenue cycle tools need large volumes of real claims data.
Two of the most valuable sources are 837 electronic claims and 835 electronic remittance files.
The 837 reports is what providers have billed and the 835 reports are how the payers adjudicated the claim.
including payments, payment details, adjustments, deductibles, coinsurance.
When properly matched, these transactions help train AI to identify underpayments, predict denials, and estimate collectability and recognize payer behavior.
But where can AI companies obtain the data? The most practical sources are organizations already authorized to receive and maintain transactions like hospitals and physician groups, health plans, healthcare clearinghouses, revenue cycle management companies, and companies purchasing or financing healthcare accounts receivable.
All AI companies can contract with one of these organizations to develop or operate defined applications because the work may involve creating, receiving, or maintaining or transmitting protected information.
The parties must exercise a business associates agreement or BAA specifying the permitted use of the data requiring safeguards.
Signing a BAA does not give the A company unlimited authority to use a hospital's claims.
The use must be authorized by the agreement and permitted by the HIPAA privacy rule.
Contracts should address model training directly, including whether the client data may be used for training, whether the information from different clients may be combined, who owns the trained model and the derived data, whether the vendor may retain information may retain information after the termination and how the vendor will prevent PHI from appearing in model outputs.
HIPAA recognized two identification methods, the safe harbor and expert determination.
Safe harbor requires the removal of specified identifiers Expert determination permits a qualified expert to determine that the risk of identifying the individuals is very small.
Claims may contain medical record numbers, claim control numbers, subscriber identifiers, service dates, addresses, free text fields, rare diagnosis, unusual combinations, or procedures.
Safe harbor may also eliminate dates needed to obtain timely filing or payment delay models.
An expert determination may preserve more useful relationships while still reducing re-identification risk.
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.
