Invoice Anomaly Detection for Hospital Revenue Recovery

Challenge

Saint-Luc, the largest hospital in Brussels, faced persistent revenue losses caused by missing entries in medical billing. Healthcare professionals must manually encode every procedure and associated material using INAMI codes, but routine omissions—such as failing to bill for blood pouch equipment during a blood draw—resulted in cumulative financial leakage. These small but frequent errors were difficult to trace at scale and remained undetected until month-end reconciliation.

Nature of collaboration

To address this challenge, Saint-Luc partnered with Sagacify to co-develop a solution that could automatically detect anomalies in invoicing data. The collaboration involved the creation of tailored AI models and a dedicated application for the financial team to identify, investigate, and resolve discrepancies with minimal manual effort.

Solution

The solution consisted of two complementary AI models, each addressing a distinct aspect of anomaly detection:

Model 1 evaluates each invoice individually by comparing the billed procedures to a predicted “patient journey,” flagging missing or inconsistent medical acts.
Model 2 performs time-series analysis on INAMI billing codes across the hospital to identify unusual volume patterns that suggest systematic issues.

A web application was co-created to ensure operational integration. It enables the financial team to review flagged anomalies efficiently and explore root causes using an intuitive interface.