starling - Record Linkage for Public Health Surveillance

Record linkage for public health surveillance datasets using either the Fellegi-Sunter probabilistic framework (via reclin2) or deterministic exact-key matching. Provides pre-linkage data quality auditing (preflight()), Medicare number checksum validation (check_medicare()), blocking variable construction (flock()), and the murmuration() linkage engine. murmuration() expresses linkage as a single cohort-to-event concept (linking a linelist of people to a dated stream of vaccination, hospitalisation, or case events within a before/during/after time window), with the historical four-code taxonomy (case-to-hospitalisation, vaccination-to-case, vaccination-to-hospitalisation, vaccination-to-event) retained as deprecated aliases. Also provides linkage weight distribution visualisation (murmuration_plot()) and threshold sensitivity analysis (perch()) with annotated AIHW, WA Data Linkage Unit, and PHRN reference benchmarks. perch() can be called standalone or triggered automatically mid-linkage via murmuration(perch_before_linking = TRUE). Includes synthetic datasets (cases_notifiable, vax_air) and three vignettes demonstrating the complete linkage workflow. Part of the aviary public health surveillance ecosystem.