# NIH grant: GENESIS: Generative Ecosystem for Navigating Efficient Scientific Information Sharing (1R01CA312900-01) — retained record

> Retained us-gov-api record from us-gov-api:nih-reporter.

Canonical: https://highsignal.app/data/records/d314ef39171c79ed

[← sources](https://highsignal.app/data)


source: [us-gov-api](https://highsignal.app/data/us-gov-api) (us-gov-api:nih-reporter)  published 2026-08-01 00:00 ingested 2026-08-03 09:17

[original source ↗](https://reporter.nih.gov/project-details/1R01CA312900-01)

## retained text



```
NIH grant 1R01CA312900-01: GENESIS: Generative Ecosystem for Navigating Efficient Scientific Information Sharing
Institute: {'code': 'CA', 'abbreviation': 'NCI', 'name': 'National Cancer Institute', 'admin_org_id': '12500008', 'admin_funding_url': ''}. PI: Judy  Gichoya. Award: $660586.
Start: 2026-08-01T00:00:00.
PROJECT SUMMARY
Biomedical research faces a critical paradox: while data sharing is essential for accelerating discoveries, privacy
concerns create substantial barriers to collaboration and reproducibility. This challenge is particularly acute in
breast cancer research, where complex disease heterogeneity necessitates large-scale, integrated multimodal
datasets for developing personalized therapeutic strategies. Despite new NIH data sharing mandates, genuine
open data sharing remains scarce. Systematic reviews suggest that only 14.3% of mammography datasets
are practically available to researchers, providing open data in appearance only. Current approaches provide
incomplete solutions to fundamental barriers: lack of computing infrastructure for data management and inability
to sufficiently de-identify sensitive health information.
This project proposes to develop GENESIS (Generative Ecosystem to NavigatE Scientific Information Sharing),
a synthetic data generation platform that addresses privacy concerns while expanding data accessibility. The
approach leverages advanced diffusion models and large language models to generate realistic synthetic datasets
that preserve statistical properties and clinical utility while eliminating re-identification risks. The project employs
a systematic approach combining machine learning innovation with rigorous clinical validation through three
complementary aims. First, the project will develop a comprehensive health-specific evaluation framework that
synthesizes existing metrics with newly proposed health-specific measures to ensure clinical relevance and
diagnostic utility preservation. Second, new multimodal synthetic data generation models will be created that
capture interdependencies across diverse data types, including electronic health records, imaging repositories,
and clinical trial data, while accommodating incomplete datasets. Third, novel causal synthetic generation
processes will be derived that explicitly encode dependencies of covariates on treatment and outcome variables,
preserving causal relationships essential for clinical decision-making.
The project directly addresses a fundamental bottleneck in precision oncology by enabling secure data sharing that
maintains privacy while supporting reproducible science. The expected outcome is critical software infrastructure
that allows any researcher to generate and share realistic synthetic data without specialized computing expertise or
privacy risks. GENESIS will transform biomedical data sharing by enabling researchers to combine datasets across
institutions, validate findings across various populations, and develop more generalizable artificial intelligence
models while maintaining the highest standards of privacy protection and data stewardship.
```



document fetched 2026-08-09 06:52