Poster Presentation Clinical Oncology Society of Australia Annual Scientific Meeting 2026

Synthetic Creation of OMOP Oncology Patients (SCOOP) (146587)

Brianna Argall 1 , Ghazaleh Niknam Shirvan 1 2 3 , Tim Churches 1 2 , Frank Lin 4 5 , Jeremy L Warner 6 7 , Georgina Kennedy 1 2 3
  1. South Western Sydney Clinical School, University of NSW, Sydney, NSW, Australia
  2. Ingham Institute, Liverpool, NSW, Australia
  3. Maridulu Budyari Gumal (SPHERE) Cancer Clinical Academic Group, Sydney, NSW, Australia
  4. St Vincent's Clinical School, UNSW, St Vincent's Clinical School , Sydney, NSW, Australia
  5. Garvan Institute, Sydney, NSW, Australia
  6. Brown University, Providence, Rhode Island, USA
  7. Rhode Island Hospital, Providence, Rhode Island, USA

Background

The Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM) oncology extension facilitates privacy-preserving federated studies. In these studies “the code goes to the data”, avoiding the need for sharing data outside of summary results through rigorous data harmonisation. However, adoption of the OMOP CDM for oncology studies is limited by lack of access to sample data in this format, which would facilitate training, code development outside of trusted research environments and demonstration of software without risking exposure of real clinical data.

Methods

We developed Synthetic Creation of OMOP Oncology Patients (SCOOP), as an open-source Python package for generation of data compliant with the OMOP CDM oncology extension. SCOOP uses YAML files allowing treatment logic and statistical distributions to be modified without writing code. Each configuration specifies properties of rules-based synthetic patient creation, treatment allocation, and operationalisation of treatment delivery into groups using HemOncKB data curated from clinical trials. SCOOP interprets these intended treatment descriptions into dated treatment delivery events, applying probabilistic variation in timing, dosing and occurrence to components marked as ongoing or optional. A baseline configuration is provided for non-small cell lung cancer cohort, with treatment allocation derived from ESMO guidelines and patient attributes informed by published literature. SCOOP also includes helper functions that visualise engine interpretation of these the configuration, alongside guidance for authoring profiles for additional diseases.

Results

SCOOP demonstrates the use of a curated oncology knowledge base generatively rather than descriptively. Clinical realism is inherited from expert curation rather than fitted to a source cohort, ensuring data remain fit for intended use cases in a strictly privacy-preserving manner.

Conclusions

SCOOP produces synthetic oncology data in the OMOP CDM format, with treatments grounded in the HemOncKB and extension to new diseases groups facilitated by a structured yet flexible YAML configuration paradigm.

  1. Warner, J. L., Dymshyts, D., Reich, C. G., Gurley, M. J., Hochheiser, H., Moldwin, Z. H., ... & Yang, P. C. (2019). HemOnc: A new standard vocabulary for chemotherapy regimen representation in the OMOP common data model. Journal of biomedical informatics, 96, 103239.
  2. Belenkaya, R., Gurley, M. J., Golozar, A., Dymshyts, D., Miller, R. T., Williams, A. E., ... & Reich, C. (2021). Extending the OMOP common data model and standardized vocabularies to support observational cancer research. JCO Clinical Cancer Informatics, 5, 12-20.