Artificial Intelligence Applied to the Triage of Requests with DUT: An Experience Report in Regulatory Technical Auditing in the Brazilian Supplementary Health System

Abstract

Introduction: The Technical Utilization Guidelines (Diretrizes de Utilização Técnica – DUT) established by the Brazilian National Supplementary Health Agency (ANS) define mandatory regulatory criteria for the coverage of procedures within the supplementary health system. Manual analysis of requests is often time-consuming and subject to interpretative variability.

Objective: To report the practical experience of applying an artificial intelligence (AI) tool in regulatory technical auditing, designed to support AI-assisted triage of medical requests according to the normative criteria established by the DUT guidelines.

Methods: A classifier based on natural language processing (NLP) was developed and trained using real medical requests previously analyzed by auditors. The algorithm searched for specific clinical terms and compared them with the criteria established in the DUT guidelines through a parameterized approach.

Results: The tool proved effective in the early identification of inconsistencies, prevention of inappropriate approvals, and improvement of traceability and standardization. Its use reduced interpretative variability and strengthened regulatory compliance and oversight.

Conclusion: The use of AI in this context is feasible and replicable, providing greater agility, traceability, and regulatory compliance when supervised by experienced reviewers. Although initially applied to events regulated by DUT guidelines, AI demonstrates broader potential for use in medical auditing, provided that well-defined clinical and regulatory criteria are available for parameterization.