estimated_loss_of_exclusivity
"estimated_loss_of_exclusivity" can be a fact but not going to come from drug approval. For example some companies in their annual report show these date by region. When this data is not available from the annual report then we could also look at the US patent database (we haven't tried it because it is complicate) or use a standard timeline.
drug_approval_organisations
drug_approval_organization is NOT necessarily the same as drug ownership. When one company buys another the marketing authorization owner often stays the original even the ownership of the drug changes.
breakthrough_therapy
"breakthrough_therapy", "fast_track", "orphan_drug", "prime_eu" and "rmat" can happen independently of drug approval. These regulatory designations are give during development and not at approval.
drug_target_actionsSection titled “drug_target_actions”
There are additional values are missing here "Mutation(s)" (FK), "HLA Allele" (FK), "Action Type" (FK), "Target - Action Specificity" (FK), "Binding Site Specificity" (FK)
discontinuation_reason
We also need "discontinuation date" and more free text describing the reasons beyond the categorical values.
cell_therapy_metadatajsonbCell source, cell type, genetic modification, target antigen (cell/gene therapies only)
For each "technology_id" we would need to develop a "technology_metadata". Currently we have this for few eg. ADCs or Cell Therapies but I would like to revise the existing ones and create it for those technologies where it doesn't exists.
drug_type
Isn't "drug_type" the same as "technology_id"?
When to add a new matview
Should we have a matview for all UI tables?
Alternative considered: Compute status from drug_approvals (approved = marketed, no approval = in_development). Rejected because it loses curated knowledge that cannot be recovered from structured sources.
In this specific examples we should make changes in the "Drug Approvals" facts and "drug status" should be a computed uneditable data. Simply we should plug in more data sources: - for example we could extract the China approval from press release - we could get additional EU approvals from other data sources
scraped
Could we avoid using the word "scrape"? These authoritative resources are ofte APIs. We could use "Ingest".