Benefits & Results

Identification of unusual and auditable transactions with an accuracy close to 90%.
Improved efficiency and accuracy in audit processes.

High accuracy in detecting unusual procedures, leading to better audit outcomes

Background

APROSS is an autonomous entity with financial individuality, organizing and administering a health care insurance system for the inhabitants of the province. It aims to provide extensive coverage and excellence in medical care administration with a solidarity and proportional system to all affiliates, improving assistance coverage and eliminating additional payment bonuses.

Challenges

Issue Identification: APROSS faced challenges in identifying unusual procedures for patients. There was no mechanism to detect questionable procedures and take preventive actions. The audit lacked data insights to detect anomalies beforehand; Issue Impact: Too many instances of anomalies were detected, leading to audit issues.

Solution

NowVertical assessed existing audit processes and anomaly patterns in patient transactions. They integrated patients' information along with providers and affiliates in GCP, creating a foundation for developing data science models to detect anomaly patterns and raise warnings to the audit team to investigate and prevent questionable medical procedures on patients.

Implementation

Assessed current audit processes for anomaly detection.
Integrated medicine consumption and procedures data into GCP for each patient.
Developed master data for anomalies detection.
Implemented data science models using Isolation Forest P and Neural Networks to identify unusual patterns of procedures.