TY  - JOUR
T1  - The Pharmacy Data Commons Needed for Safe and Equitable Algorithm Development
A1  - Kristian Eriksen
A1  - Solveig Berg
A1  - Magnus Dahl
JF  - Archives of Pharmacy Practice
JO  - Arch Pharm Pract
SN  - 2320-5210
Y1  - 2025
VL  - 16
IS  - 3
DO  - 10.51847/Zuc4NBkGdt
SP  - 54
EP  - 62
N2  - Pharmacy algorithms are increasingly developed from data dispersed across dispensing systems, electronic health records, claims, laboratories, pharmacovigilance repositories, and patient-facing technologies. Fragmentation is not only a technical inconvenience: inconsistent semantics, incomplete provenance, selective representation, unclear permissions, and weak accountability can propagate into algorithm development and medication-related decisions. This article proposes a pharmacy data commons as a governed, distributed capability for responsible data access, documentation, harmonization, analysis, evaluation, and benefit sharing. The commons is not defined as a single pooled database. Its proposed architecture links participating data assets, semantic and provenance services, computable use permissions, representation profiles, privacy-preserving analysis environments, institutional decision rights, community participation, and explicit boundaries between model output and professional medication authority. Access is treated as a conditional lifecycle rather than a one-time approval, with stewardship, representation review, safety and equity evaluation, output review, monitoring, incident learning, and retirement. Commons performance should therefore be assessed across data fitness, interoperability, provenance, access consistency, privacy, representation, human–AI workflow, medication-safety risk, community legitimacy, distributive benefit, and sustainability. Validation would require multisite technical audits, external and subgroup evaluation, human-factors studies, safety assessment, governance-process evaluation, privacy testing, and community-defined measures of legitimacy and benefit. The proposal remains non-empirical and context-dependent. It does not establish clinical benefit, regulatory acceptability, fairness, lawful use, or deployment readiness. Its original contribution is an integrated governance architecture that makes the conditions for safe and equitable pharmacy-algorithm development visible, testable, and contestable.
UR  - https://archivepp.com/article/the-pharmacy-data-commons-needed-for-safe-and-equitable-algorithm-development-ksbrnpuzhj9oobs
ER  -