Grandes modelos de linguagem aplicados à análise de precificação de plano de saúde

Abstract

This study aims to test the performance of Large Language Models (LLMs) in conducting regression analyses, with a focus on practical applications for health plan pricing. To achieve this, open data provided by the National Supplementary Health Agency (ANS) will be used, ensuring a reliable foundation for model evaluation. The core approach focuses on assessing the ability of LLMs to perform actuarial calculations accurately and support decision-making in the context of supplementary health. In addition to analyzing technical performance, the research also evaluates the validity and reliability of LLMs, identifying their limitations and areas for improvement. As a result, the study aims to propose alternatives that contribute to enhancing actuarial practices in the healthcare market. It considers the potential of these models to provide innovative solutions for challenges related to health product pricing and rating, as well as for optimizing traditional processes. Ultimately, this research seeks not only to assess the effectiveness of LLMs in the current landscape but also to foster the development of new actuarial techniques based on artificial intelligence, with a positive impact on the supplementary healthcare market.