Observational Health Data Sciences and Informatics

Face-to-Face Oncology Meeting Notes

Next steps

I. Tasks to generate the minimal viable product

1. Diagnosis

a. Vocabulary Mapping

• Rely only on pre-coordination of SNOMED morphology and anatomy corresponding to ICD-O histology and topography. If a pre-coordinated SNOMED diagnosis does not exist, create OMOP-generated one.
• Present proposal for pre-coordinated terms to SNOMED.

b. Diagnosis identification and verification

• Sourced from cancer registry based on ICDO-O – >SNOMED mapping
• Sourced from EMR based on ICD-9/10 → SNOMED mapping or ICDO-O – >ICD-9/10 → SNOMED
• Compare the diagnoses and patient identified from the two sources
• Explore how ICD-9/10 diagnosis changes in EMR, if at all, through the course of the disease

c. Cancer diagnosis representation in OMOP CDM – possible extension to represent recurrences

2. Staging

Choose vocabulary ???

3. Treatments

a. Drugs

• Compare SEER drugs list with RxNorm cancer drug classes and decide which one should be used in the vocabulary. If SEER is more suitable, add SEER classification to the vocabulary.
• Add SEER drug cocktails to the vocabulary (only ingredients, no other specifics) in EMR based on those combinations.
• Identify chemo regimens in EMR using SEER cocktails and validate with chemo orders when available
• Work on cancer Drug Era algorithm and modeling in CDM

b. Overlay cancer registry treatment schema (first occurrence only) with EMR details to construct complete treatment regimen

c. Model representation of the treatment regimen (era) in the CDM

II. Exploratory

1. Discuss with IMO procedure mappings to SNOMED and classification of cancer cases.
2. Explore commercial solutions that claim to identify cancer treatment regimens, recurrences, progression and compare the outcomes with our attempts to identify treatment regimens
3. Invite Andrew Stewart and Mark Denise to present their efforts and results in mapping to CDM and additional modeling
4. Find a geneticist and oncologist to identify cancer-relevant genomic areas
5. Find out if we can get cancer registry data from any of the “SEER” states

III. Operational

Document challenges related to extending OHDSI infrastructure to support cancer research for NCI