This Request for Information (RFI) seeks input on how Generative Artificial Intelligence (AI) can be employed to enhance the use and integration of big data (e.g., omics data) in clinical measurements and high-throughput omics screening in heart, lung, blood, and sleep (HLBS) research.
Generative AI models learn the patterns and structure of their input training data, and then generate new content that has similar characteristics. The generative AI model uses neural networks to identify the patterns and structures within existing data in order to generate new content, including text, imagery, audio, and synthetic data.
The biological language coded inside human genomes regulates gene expression (e.g., RNAs and proteins) in response to environmental perturbations that can lead to changes in clinical measurements. The question arises: How can generative AI help human experts recognize patterns of dynamic gene-environment interactions underlying health and disease?
NHLBI's TOPMed program has produced 200,000 whole-genome sequences and is on track to amass multi-omics data (including RNA-seq, methylome, metabolome, and proteome) for nearly 270,000 samples. Despite the enormous amount of data generated by high-throughput screening with new technologies and omics platforms, only a fraction of these data are being optimally assessed and incorporated into practice.
NHLBI seeks to develop innovative approaches for the integration, analysis, and interpretation of these multi-dimensional data. The ultimate goal is to deepen our understanding of how these diverse data contribute to the health and disease states of heart, lung, blood, and sleep (HLBS) systems. By achieving this, we can enhance diagnosis, treatment, and prevention strategies, ultimately improving patient outcomes.
Through this RFI, our objective is to understand how Generative AI can help in these endeavors, effectively serving as a tool for integrating and analyzing these vast and complex datasets.
NHLBI is specifically seeking input on how Generative Artificial Intelligence (AI) can be employed to enhance the use and integration of big data (e.g., omics data) in clinical measurements and high-throughput omics screening in HLBS research.
Topics of interest include, but are not limited to, the following:
- Current biomedical research use-cases for integrating and interpreting clinical measurements and high-throughput omics data
- Resources needed for the HLBS research community to effectively use Generative AI
- Strategies for employing "foundation" models in the analysis of omics data and high-throughput screening data
- Challenges and barriers to incorporating Generative AI into omics research and other big data, and ways that these challenges might be mitigated/addressed
- Best practices for using Generative AI to develop hypotheses from existing data
- Types of training needed for the HLBS research community to effectively incorporate Generative AI in research
- Specific HLBS research questions that could be addressed using Generative AI
- Safeguards or protocols necessary to ensure the ethical use of Generative AI in patient/participant data analysis and for HLBS research
- Validation and verification of Generative AI insights
- Identification and mitigation of bias in Generative AI models for HLBS research
- Any publications or examples that showcase the use of Generative AI models or Large Language Models in harmonizing the phenotypes across various cohorts