Cases/ Hospital Sírio-Libanês + RIVIO
A benchmark in hospital technology, Sírio-Libanês recovers R$ 7.5 million in a single month without seeing a single additional patient.
Among the first institutions in the country to instrument the revenue cycle with artificial intelligence, the hospital began measuring, month by month, how much the technology returns on each claim.
- Period measured
- July 2026
- Claims
- 25,736
- Findings
- 107,732
- Results
- R$ 7.70 million
Space for a testimonial from a spokesperson. It can come from the institution itself or from another one that already runs on Rivio, as Palantir does on the Airbus page, where the quote comes from an airline customer rather than the main client. No quote has been drafted here: it needs to be said by a real person and approved by them.
Antonio Pereira General Administrative Director at Sírio‑Libanês
01/ The challenge
Sírio-Libanês runs two high-complexity units, in São Paulo and Brasília. Each claim carries thousands of items, checked against dozens of payer contracts whose rules change every week. In the audit stage alone, more than 25,000 claims are finalized every month.
Between patient care and cash in the bank there are eight stages, and claim auditing is only one of them. Within that stage there are three distinct review phases. At high-complexity volume, reviewing item by item, across every claim, every month, stopped being achievable by human work a long time ago.
What the hospital brought to the table was a list of questions the operation needed to answer every month, with a number attached:
How many items are provided to the patient but never reach the invoice, and what is that worth per month?- Which claims will come back denied by the payer, and for what specific reason?
- How many claims could be closed without human intervention?
- How much does each real that goes through an automated audit return to the hospital?
- And does that result improve from one month to the next, or does it stand still?
02/ The solution
Rivio began operating the claim audit on a defined slice of the two units’ billing, with the rest following the traditional process, and put measurement on top of every indicator.
Each claim is read against a base of more than 2,500 rules, spread across professional fees, supplies, OPME (implants and special materials), medications, packages, fees, procedures, daily rates, exams, medical gases, dressings and nutrition. In July, 206 of those rules were triggered across the two units.
When a rule finds a discrepancy between what was performed, what is documented and what is contracted, it generates a finding on the claim, with the evidence attached. A claim with no discrepancy is closed automatically. A claim with a finding goes to the auditor, who decides.
Where the technology operates today
-
01
Verification and authorization
-
02
Item entry
-
03
Claim audit
-
04
Closing
-
05
Filing and billing
-
06
Submission to the payer
-
07
Denial analysis
-
08
Appeals and reconciliation
One of the eight stages of the cycle, in the retrospective phase of the audit. The other seven follow the traditional process and are not part of this assessment.
What changes in the auditor’s day
The auditor is still the one who decides. What changed is how the day starts: when the auditor opens the claim, the issue has already been flagged, with the evidence attached and the amount calculated. Clinical judgment goes to the exception that needs a person, and the rest arrives already resolved.
At a human pace, reviewing July’s 25,736 claims item by item would take more than 10,000 hours of work. That is a full month for more than 60 people dedicated solely to rereading claims already signed off as ready.
03/ The result
-
R$ 7.50M
in results measured in July 2026, across the two units
-
+30.7%
growth in results from June to July
-
50.5%
of claims closed without an auditor touching them
-
33,035
items added or corrected in the month’s claims
The result per finalized claim is derived. See the methodology note.
How the result breaks down
Additions are items the hospital provided to the patient that had not reached the invoice: medication administered, supplies used in the operating room, an item billed below the contracted amount.
are items adjusted before the claim leaves the building. Each of them would have been sent to the payer, come back rejected weeks later and opened a cycle of appeal, resubmission and waiting. The claim goes out clean on the first attempt, which shortens the road to payment.
Additions and removals by unit, July 2026
Values measured on the ERP database. The total height of each bar is the unit’s result for the month.
The result improves from one month to the next
From June to July, the consolidated result rose 30.7%, from R$ 5.74 million to R$ 7.50 million, with the same team and on the same portfolio of claims. What changed was the maturity of the rules and the share of the operation covered by the audit.
Consolidated result by month, April to July 2026
June and July are measured values. April and May were read from the monthly tracking charts and are marked as estimates.
Half the claims the auditor doesn’t need to open
In Brasília, automatic closure went from 13.0% in June to 58.1% in July, the fastest jump of the two units, driven by the local operation reaching steady state. In São Paulo there was a 4.0-point dip, from 54.1% to 50.1%, in the same month that the volume of findings nearly doubled, from 52,892 to 102,240. More findings send more claims to a human decision, and that is what the indicator records.
Automatic closure by unit, April to July 2026
Percentage of claims finalized without an auditor touching them. The dashed segment covers the estimated months.
Brasília, the unit ramping up
Brasília’s calculation base shows the ramp-up curve month by month. The volume of finalized claims more than doubled between April and July. The drop in the average claim value in July follows the opening of the scope to lower-value claims, which previously did not go through the audit.
Calculation base for the Brasília unit
Values measured on the ERP database, April to July 2026.
| Month | Claims finalized | Average claim | Change in average |
|---|---|---|---|
| April | 525 | R$ 15,875.02 | --- |
| May | 693 | R$ 16,848.03 |  +6.1% |
| June | 779 | R$ 19,181.89 | +13.9% |
| July | 1,149 | R$ 11,906.01 |  -37.9% |
Average impact per audited claim
This indicator divides the month’s result by the number of finalized claims, and answers a practical question: how much the audit returns, on average, on each claim that goes through it. Any hospital can multiply it by its own claim volume to estimate the order of magnitude of its case.
It is a gross figure, an impact on revenue. The net return depends on the cost of the service, which is outside this assessment. The difference between the two units comes from the claim profile. Brasília has a heavier weight of inpatient stays, with longer claims, more items charged and a much higher average value, so each audited claim carries more opportunity.
-
São Paulo
R$ 252
average impact · 24,587 claims finalized in July, mostly outpatient and emergency.
-
Brasília
R$ 1,137
average impact · 1,149 claims finalized in July, with a heavier weight of inpatient stays.
-
Consolidated
R$ 291
average impact · 25,736 claims finalized across the two units.
Gross impact on revenue, with no deduction for the cost of the service. Calculation: result measured in the month divided by the number of finalized claims.
04/ What comes next
Claim auditing is the first of the eight stages between care and cash.
The hospital’s plan is to instrument each of the other seven with the same measuring stick already applied here.
That is why this revenue stayed where it was: the math of human labor never added up, and the technology able to do this at scale has only now arrived. Today the audit reads every claim, every month. What was unfeasible became operating routine.
And the effect doesn’t end at the bank account. Every real recovered is a real the hospital doesn’t have to find elsewhere. It becomes an on-call physician hired, a bed opened, new equipment, a test that comes back the same day. At a leading hospital, recovered revenue goes back to the patient.
R$ 7.5 million in one month is what the first stage returned.
Methodology note
- Source
- Direct extraction from the ERP database of the São Paulo and Brasília units, measured monthly.
- Period
- July 2026 for the month’s indicators; April to July 2026 for the series.
- Scope
- Claim audit, one of the eight stages of the revenue cycle, in the retrospective phase. The concurrent and consensus phases were not active during the period.
- Coverage
- A defined slice of each unit’s billing, being expanded progressively. The rest follows the traditional process and is not measured by this assessment. Coverage percentages and audited financial volumes are the institution’s information and are not disclosed.
- The result includes
- Additions, items provided but not billed that started being charged; and removals, items adjusted before submission to the payer.
- Not included
- The effect of the other seven stages of the cycle, changes in the average time to payment, and recoveries obtained through denial appeals with payers.
- Derived values
- The average impact per claim comes from the measured result divided by the number of claims finalized in the month. It is a gross figure on revenue, with no deduction for the cost of the service, and therefore does not represent return on investment. The more than 10,000 hours come from the benchmark of 40 hours per 100 claims in the manual process, applied to July’s volume.
- Estimated values
- April and May 2026 come from reading the monthly tracking charts, without direct extraction from the database, and are marked as estimates in every chart. June is derived from the published percentage changes and matches the sum of the month’s additions and removals.
Indicator definitions
-
Addition
An item identified as provided to the patient and missing from the claim, or billed below the contracted amount, that starts being billed.
-
Removal
An item removed or corrected on the claim before submission to the payer, because of a discrepancy between what was performed, the documentation and the contract.
-
Finding
An alert generated by a rule in the base on a specific claim. It can become an addition, a removal or be dismissed by the auditor.
-
Automatic closure
The percentage of claims finalized in the audit without any intervention from a human auditor.
-
Rule effectiveness rate
The share of generated findings that were actually acted on. In July: 41.5% in São Paulo and 23.8% in Brasília.
-
Coverage
The share of the unit’s billing that goes through Rivio’s audit in the period.
