Machine intelligence in healthcare-perspectives on trustworthiness, explainability, usability, and transparency.

Area of Science:

• Biomedical research
• Clinical applications
• Healthcare innovation

Background:

• Machine Intelligence (MI) is increasingly vital in biomedical discovery, clinical research, diagnostics, and precision medicine.
• MI tools enhance decision-making for researchers, physicians, and patients, improving health outcomes.
• The integration of MI in healthcare settings promises to boost efficiency and patient care quality.

Purpose of the Study:

• To address challenges arising from the growing use of MI in clinical settings.
• To identify key issues and propose solutions for advancing MI in healthcare.
• To foster effective, transparent, and ethical progress in healthcare MI.

Main Methods:

• A workshop co-hosted by NIH, NCATS, NCI, and NIBIB on July 12, 2019.
• Discussions involved researchers, clinicians, patient advocates, industry, academia, and federal agencies.
• Key topics included data quality, EHR access, system transparency, explainability, and bias.

Main Results:

• Identified critical issues in applying MI to healthcare.
• Highlighted the need for improved data quality, accessibility, and EHR integration.
• Emphasized the importance of transparency, explainability, and mitigating bias in MI systems.

Conclusions:

• Addressing identified challenges is essential for accelerating MI progress in healthcare.
• Proposed avenues and solutions aim to guide the effective, transparent, and ethical deployment of MI.
• Collaboration and focused improvements can unlock MI's full potential in the health ecosystem.