Fraud stopped before payment: USD 750k+ saved on fake cardiac ablation claims

Case 1  ·  Claim Level  ·  Latin America

WHERE WE STARTED

One of Latin America’s largest national health insurers was processing high-value cardiac ablation claims from a hospital that had no cardiac ablation facilities. The claims passed standard validation — the procedure codes were legitimate, the amounts plausible. Nothing in the existing rules engine raised a flag. The fraud had been running for less than a year before Kirontech detected it — but in that time, claims were being settled automatically and without challenge.

THE CHALLENGE

Cardiac ablation is a complex, high-cost procedure used to correct abnormal heart rhythms — typically indicated in adults with specific arrhythmia diagnoses, rarely performed in patients under 18. The insurer’s existing rules-based engine checked for obvious red flags — duplicate claims, known exclusions — but had no way to assess whether a specific hospital was clinically capable of performing the procedures it was billing for, or whether the patients themselves were plausible candidates for those procedures.

What had already been tried:

The insurer’s fraud team had no existing workflow to cross-reference patient demographics, facility capability, and procedure frequency in combination. Each element looked normal in isolation — it was the pattern across them that revealed the scheme.

HOW KIRONTECH FOUND IT

• HIP used multiple factors — age, location, time, cost, medical history, entity relationships — to assign a likelihood score to each procedure.
• An ablation claim for a 17-year-old high school student triggered an immediate flag: cardiac ablation is almost never indicated in adolescents without a documented congenital condition, and this patient had no such history. The hospital also lacked the necessary facilities to perform the procedure.
• HIP identified a second identical ablation claim with the same unusual features. Cross-referencing entity relationships revealed the doctor on the first claim was the father of the second patient.
• Further analysis showed the patient in the original claim was actively at work on the date of the supposed treatment — corroborated through secondary data.

WHAT CHANGED

Beyond the financial recovery, this case changed how the insurer’s fraud team operates. For the first time, they had data-led evidence strong enough to suspend a provider and refer findings to their legal team. The detection model trained on this case now runs across all incoming high-value claims in the region — turning a one-off investigation into a standing capability.