Event-Sourced Governance OfData Quality Rules For Real-Time, Auditable Healthcare Data Platforms

Rule-based checks are central to data quality (DQ) assessment on healthcare data platforms, and their parameters change often. In regulated settings each change must be approved, promoted through staged environments, and later explained to auditors, yet DQ tools rarely govern the rule set itself. This study asks whether event sourcing can make such changes both fast and verifiable. We propose an architecture in which every rule-lifecycle action is an immutable, hash-chained event in a log partitioned by rule; separation of duties and risk-tiered approval are enforced before an event is written; idempotent projections drive gated promotion; rule parameters are bound at runtime; and the log head is anchored outside the platform. We evaluated the design in a single-process simulation on synthetic patient data against three baselines: a dbt rebuild pipeline, a mutable parameter table with an audit trigger, and a ledger-style table with a hash-chained audit. The simulation reproduces the relevant Kafka log semantics but is not a production deployment. Approved changes reached production in a medianof 0.092 s, against tens of seconds to minutes for the rebuild pipeline; the two table baselines were faster still, so the speed comes from runtime parameter binding rather than from event sourcing. With every record anchored, the hash-chained designs detected all seven tested attack types, whereas the unchained audit table missed every edit of attribution or value. Historical states were reconstructed exactly, an independent verifier found no constraint violations in normal runs and flagged every injectedviolation, and ablations identified the components that correctness depends on. The main limitations are synthetic data, a simulated log, and modeled rather than measured approval delays

Authors

Publication Details

Journal
International journal of data science and machine learning.
Published
2026-11-16
DOI
https://doi.org/10.55640/ijdsml-06-02-15
Primary Topic
Data Quality and Management
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Event-Sourced Governance OfData Quality Rules For Real-Time, Auditable Healthcare Data Platforms

Ritulkumar Patel
International journal of data science and machine learning.
Data Quality and Management
article

Event-Sourced Governance OfData Quality Rules For Real-Time, Auditable Healthcare Data Platforms

Ritulkumar Patel
article en

Abstract

Rule-based checks are central to data quality (DQ) assessment on healthcare data platforms, and their parameters change often. In regulated settings each change must be approved, promoted through staged environments, and later explained to auditors, yet DQ tools rarely govern the rule set itself. This study asks whether event sourcing can make such changes both fast and verifiable. We propose an architecture in which every rule-lifecycle action is an immutable, hash-chained event in a log partitioned by rule; separation of duties and risk-tiered approval are enforced before an event is written; idempotent projections drive gated promotion; rule parameters are bound at runtime; and the log head is anchored outside the platform. We evaluated the design in a single-process simulation on synthetic patient data against three baselines: a dbt rebuild pipeline, a mutable parameter table with an audit trigger, and a ledger-style table with a hash-chained audit. The simulation reproduces the relevant Kafka log semantics but is not a production deployment. Approved changes reached production in a medianof 0.092 s, against tens of seconds to minutes for the rebuild pipeline; the two table baselines were faster still, so the speed comes from runtime parameter binding rather than from event sourcing. With every record anchored, the hash-chained designs detected all seven tested attack types, whereas the unchained audit table missed every edit of attribution or value. Historical states were reconstructed exactly, an independent verifier found no constraint violations in normal runs and flagged every injectedviolation, and ablations identified the components that correctness depends on. The main limitations are synthetic data, a simulated log, and modeled rather than measured approval delays

International journal of data science and machine learning.Vol. 6(02)
Data Quality and Management
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon