Analysis of Data Availability Statements in Journal Articles Published by HKU Scholars

<table><tr><td><p dir="ltr">This study assesses the prevalence and quality of Data Availability Statements (DAS) in 7,301 journal articles published by scholars at the University of Hong Kong (HKU) between 2015 and 2024, with a focus on Medicine and Engineering. Using an AI-powered framework (DeepSeek-V3-0324), we automate DAS detection and classification to assess compliance with open science principles and FAIR data standards. Results reveal a rise in DAS adoption post-2020, but persistent gaps remain: only 11–19% of statements include direct repository links, while vague phrases such as 'available upon request' predominate. Medical articles frequently cite ethical restrictions, whereas Engineering papers more often reference repositories (e.g., GitHub, DOIs). Publisher policies significantly influence prevalence (e.g., PLOS: 93% vs. Elsevier: 9%), but enforcement of actionable data sharing remains inconsistent. The study provides a scalable method for institutional DAS auditing and recommends reforms to enhance transparency, reproducibility, and alignment with global open science goals.</p></td></tr></table><p></p>