Problem: The rapid adoption of AI in healthcare is hindered by a lack of stable funding, limited digital infrastructure, and a lack of data standardization.

Solution: Leverage collaboration to develop country-specific solutions, prioritizing data governance and clear regulatory frameworks.

Introduction

Artificial Intelligence (AI), including generative Artificial Intelligence (Gen AI), is increasingly being used in health systems, but its adoption raises important challenges for public policy.

Health professionals recognize both the potential benefits and risks associated with the use of AI. While they emphasize its capacity to reduce administrative burdens, they also express concerns regarding inadequate training and the existence of policy gaps.

This study examines the development and use of AI in Brazil’s Public Health System (SUS). By analysing recent scientific publications and interviewing experts, we found growing interest in AI alongside persistent obstacles.

Persistent obstacles and data challenges

These include unstable funding, unclear rules, poor data quality, and limited digital infrastructure.

Persistent problems also exist in the standardisation, integration, and accessibility of clinical data, including the need to unify electronic health records. One researcher explained, the data «is not curated, it is not processed in a way that ensures high quality».

Algorithmic bias was a concern, given Brazil's diverse racial, ethnic, and regional composition: «if there is bias in the data, then that same bias is reflected in the results of the AI model», making it a barrier to ethical and equitable implementation.

The public ecosystem as a driver

SUS constitutes a unique model within Latin America, inasmuch as universities and Fiocruz collaborate in the development of domestically-tailored solutions that are deployed nationally.

This stands in contrast to other countries, which lack comparable structures and are, therefore, more reliant on externally developed technologies. The adaptation of algorithms to Brazilian diversity and the danger of deeper technological dependence on Big Tech firms were also cited.

Regarding human resources, interviewees noted that while initial resistance from health professionals to AI has decreased, inadequate training in digital literacy remains a barrier.

Policy Implications

• Stable and long-term funding mechanisms are important to support the adoption of Generative Artificial Intelligence in public health systems.

• Clear and adaptive regulatory frameworks can reduce uncertainty and facilitate responsible AI deployment in healthcare.

• Investments in data quality and digital infrastructure are preconditions for AI implementation.

• Capacity-building strategies targeting health professionals and public sector institutions can enhance the equitable use of AI, including Gen AI technologies.

This article was co-authored by Itala Laurente, a researcher and Ph.D. candidate in Science and Technology Policy at the State University of Campinas (Unicamp), and her doctoral supervisor, Dr. André Sica de Campos (linkedin.com/in/andré-s-de-campos-73573326). Their research examines the development, obstacles, and drivers of artificial intelligence in public health systems.

How is your department addressing the challenges for AI in health? Leave us a comment below to share experiences among public servants.

This article is based on the manuscript Generative Artificial Intelligence in Public Health: Recent Evidence on Barriers and Drivers in Brazil. During its preparation, Gemini was used solely for language refinement and stylistic improvements. The authors reviewed and edited the final content and assumed full responsibility.

This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) Finance Code 001