readmit 1.1.0

Canonical production now rejects discharge episodes with discharge_time > as_of_time before eligibility, risk-provider calls, or history writes. A discharge exactly at as_of_time remains valid.
Canonical terminal events now require only terminal_event_id, episode_id, event_type, and occurred_at. Additional columns remain permitted. Eligibility uses occurred_at <= as_of_time; an event-availability timestamp is no longer required or interpreted.
Newly initialized assessment history and rrp_inspect_history() retain selected terminal-event identity, type, and occurrence time without terminal_available_at.

readmit 1.0.0

readmit retains its hospital-readmissions focus and evolves from the exploratory utilities in 0.0.1 into the Readmission Risk Pool (RRP), a project-based framework for repeated readmission risk assessment. This revision intentionally changes the public API and moves from MIT to Apache License 2.0.
rrp_project_init() creates a user-owned project with editable configuration, source-mapping and risk-provider scripts, branding, and a Shiny application. Project authors supply their own clinical data mapping and risk logic; rrp_project_validate() checks the authoring structure, and rrp_state_init() establishes persistent history.
rrp_run() assesses the remaining risk of a first readmission through 30 elapsed days after discharge at a chosen analytical time. It applies shared eligibility rules to project-supplied canonical data and retains accepted estimates, provider failures, and ineligible outcomes. Use rrp_project_status() and rrp_inspect_history() to examine the results.
rrp_refresh_products() derives current remaining risk, recorded risk trajectories, and an operational scope summary as CSV files from retained history. The project-owned Shiny app opens with empty views before products exist and displays them after refresh.
A short installed getting-started vignette and a reproducible website implementation guide demonstrate the project workflow with illustrative data and risk logic. No clinically validated prediction model is included.