NIAR
Back to home

About NIAR-Saúde

Get to know the project — its principles, objectives, workstreams, and trajectory.

The center

Developing artificial intelligence in healthcare also means defining the paths that make it ethical, safe and responsible.

UFMG's Center for Responsible Artificial Intelligence in Health (NIAR-Saúde) is a multidisciplinary initiative dedicated to the responsible development and application of artificial intelligence in healthcare, as well as to building the guidelines that steer this process.

Through the integration of research, technology, and different fields of knowledge, it seeks to transform the use of health data, promoting ethical, safe, transparent, and people-centered solutions.

Concept

Dimensions of Responsible AI

Principles that underpin ethical, secure, transparent, and trustworthy artificial intelligence.

Transparency

Makes the operation, decisions, and limitations of AI understandable and auditable.

Fairness and Bias

Promotes equity, identifies and mitigates bias and discrimination to ensure fair decisions for everyone.

Privacy and Security

Protects personal data and sensitive information, ensuring confidentiality, integrity, and compliance with legislation.

Data Governance

Establishes practices for the collection, storage, quality, sharing, and use of data with reliability, traceability, and compliance.

AI Governance

Defines policies, processes, and responsibilities to ensure ethical and secure use aligned with the organization’s objectives.

Responsible AI

Trust, ethics, and benefits for all

  • AI Governance + Transparency + Fairness and Bias

    Fair, transparent and well-governed decisions.

  • Transparency + Fairness and Bias + Privacy and Security

    Less bias, more trust, security, and inclusion.

  • Fairness and Bias + Privacy and Security + Data Governance

    Equity in decisions, protection of rights, and responsible use of data.

  • Privacy and Security + Data Governance + AI Governance

    Transparent governance over data, AI, and security.

  • Data Governance + AI Governance + Transparency

    Accountability through explainability and AI governance.

  • Transparency + Fairness and Bias

    Transparency for fairer AI

  • Fairness and Bias + Responsible AI

    Less bias, more trust and inclusion

  • Fairness and Bias + Privacy and Security

    Equity with respect for privacy

  • Transparency + AI Governance

    Transparency on data use and protection

  • Privacy and Security + Responsible AI

    Protection of rights and user trust

  • Transparency + Data Governance

    Accountability through explainability

  • AI Governance + Data Governance

    Governance aligns AI and data with its ethical objectives

  • Privacy and Security + Data Governance

    Security enables trust and compliance

Together, these dimensions foster AI systems that are trustworthy, ethical, secure, transparent, and aligned with human values and legal and regulatory requirements.

Specific objectives

Our commitments

  • Responsible data access

    Design, implement, and operate a service that enables responsible access to health data and models, ensuring ethics, security, and legal compliance.

  • Training and education

    Disseminate knowledge and promote training in responsible artificial intelligence applied to healthcare, qualifying professionals in the field.

  • Computational platform

    Design, implement, and validate a computational platform for developing and using responsible AI solutions, focused on transparency and traceability.

  • Applied case studies

    Plan, conduct, and evaluate case studies that demonstrate, in practice, the application of responsible AI in healthcare.

  • Technology transfer

    Promote the transfer of technology and knowledge related to the deployment and operation of NIAR-Saúde, expanding the impact of the solutions developed.

Goals

Workstreams

The project is organized into seven concurrent goals, distributed across four types of activity: processes and best practices, computational platform, pilot projects and dissemination.

Goal 1

Responsible access service

Specification, implementation, and operation of an experimental service for responsible access to health data and models.

Michele Brandão

Coordination

Michele Brandão

Goal 2

Training and education

Knowledge dissemination and courses on ethics and use of the NIAR environment for training in responsible AI.

Ana Paula SilvaZilma Reis

Coordination

Ana Paula Silva and Zilma Reis

Goal 3

Computational platform

Development and validation of a computational platform to support responsible AI in healthcare.

Wagner MeiraDorgival Guedes

Coordination

Wagner Meira and Dorgival Guedes

Goal 4

AI for electrocardiogram

Development of an algorithm for automated ECG diagnosis, expanding access and supporting medical reporting.

Antonio Ribeiro

Coordination

Antonio Ribeiro

Goal 5

Predictive models for NCDs

Prediction of chronic diseases and risk factors based on epidemiological and sociodemographic data.

Deborah Malta

Coordination

Deborah Malta

Goal 6

AI on SUS oncology data

Integration and predictive analysis of oncology patient data from SUS in Belo Horizonte.

ML

Coordination

Mariangela Cherchiglia

Goal 7

Technology transfer

Dissemination and transfer of knowledge and technologies developed in the project.

Wagner Meira

Coordination

Wagner Meira

Journey

NIAR-Saúde milestones

Some of the key moments that shaped the making and development of NIAR-Saúde.

  1. Signing of the TED

  2. Sep 2025

    Project kickoff

  3. NIAR Framework

  4. Mar 2026

    Secure room inauguration