Artificial Intelligence

For decades, Artificial Intelligence (AI) has been a valuable tool in healthcare, recognizing patterns that help guide care delivery and research. With the emergence of more sophisticated machine-learning algorithms and generative AI, there is an opportunity to augment the human elements of healthcare even further. AI-enabled capabilities have the potential to sharpen the accuracy of diagnoses, improve clinical decision-making, and enable earlier disease intervention. The scope for applications ranges beyond immediate patient care from administrative tasks to improving efficiencies to predictive analytics to help forecast disease outbreaks.

Considerable challenges in its adoption include:

• Disparities in the ability of less-resourced healthcare systems limit opportunities to leverage emerging AI technology.
• Privacy and security risks associated with manipulating sensitive patient information and data.
• Incomplete or inaccurate data sources compromise the quality of AI-generated findings.
• Fear that data used to train technology may result in discriminatory treatment, diagnosis, or misdiagnosis of patients from underrepresented demographics.
• Resistance to adoption by healthcare professionals and the public due to lack of trust.
• Complex and potentially conflicting payment, patent, and liability issues across regulatory and legal frameworks.
• Uncertainty regarding best practices and the risk of a state-by-state patchwork emerging without federal guidance.
• Patient concerns about personal data and health information being included in an AI algorithm.

To realize AI’s potential in healthcare, policymakers must work closely with private sector leaders to address these ethical, technical, financial, and regulatory issues.

Solutions

Specific areas for policy action include:

• Determining the role of human oversight based on AI’s decision-making risk and ensuring the explainability of AI-generated recommendations.
• Ensuring that AI leverages high-quality, accurate, and relevant data and that the industry implements robust data governance practices to maintain data integrity and prevent biases.
• Developing and applying voluntary federal guidelines that focus on best practices and accreditation.
• Protecting against overly restrictive regulations, which would impede innovative use and the pioneering adoption of AI to attain the greatest potential.