Enabling the clinical applications of artificial intelligence in genomics: a perspective of the AMIA Genomics and Translational Bioinformatics Workgroup

December 18, 2023

Our summary

Artificial Intelligence (AI) is reshaping diverse fields, including genomics, with immense potential for discovery and improved clinical care despite the challenges in handling complex genomic data. In this article, the American Medical Informations Association Genomics and Translational Bioinformatics Workgroup outlines key factors for the successful integration of AI in genomics covering data availability, technical aspects, ethics, engagement, and regulations.

Researchers discuss that while the landscape of genomic data is expanding, hurdles in data democratization remain, emphasizing the urgency for global sharing facilitated by government-backed structures. Clinical adoption calls for transformative changes, advocating for a division between sequencing and interpretation, embracing next-generation sequencing data, and highlighting the role of functional genomics and standards development. Ethical dimensions stress privacy, accuracy, and bias mitigation, emphasizing advanced techniques like federated learning for patient data protection. Stakeholder education, particularly for clinicians, becomes pivotal for unlocking the benefits of AI, with a focus on explainability, transparency, and interpretability in AI models to foster trust. 

While acknowledging existing FDA guidelines, this article calls for a nuanced regulatory approach that balances safety and innovation in the evolving landscape of AI in clinical genomics, advocating for regulatory frameworks that adapt to the dynamic nature of AI technologies.

Why this matters

This article highlights the pivotal role of AI in clinical genomics, emphasizing responsible application through best-practice guidelines to ensure ethical integration. It prioritizes transparency, reproducibility, and the prevention of biases while addressing legal and regulatory aspects and safeguarding patient privacy. Active collaboration with clinical and informatics communities is essential for successful AI validation and deployment, emphasizing documentation, human oversight, and rigorous real-world validation.

Read the research