Wolff Miriam K, Royston Sam, Fougner Anders L, Schaathun Hans Georg, Steinert Martin, Volden Rune
Norwegian University of Science and Technology, Department of ICT and Natural Sciences, Larsgårdsvegen 2, 6009 Ålesund, Norway.
Replica Health, 249 Willoughby Ave 4A, Brooklyn NY, United States.
Data Brief. 2025 Feb 17;59:111399. doi: 10.1016/j.dib.2025.111399. eCollection 2025 Apr.
Diabetes management datasets are often compiled from various sensors and devices, including diabetes technology, activity trackers, and other health-related equipment, resulting in heterogeneous data formats. Despite the abundance of available data, inconsistencies in dataset formats and data-sharing practices limit the ability to build on prior work and compare results across studies. Standardizing data-sharing formats can improve consistency, facilitate dataset consolidation, and reduce the data processing burden for researchers. This article explores the current state of data-sharing practices in diabetes management research and proposes guidelines for harmonizing datasets using a unified time-aligned tabular format. We demonstrate the application of these guidelines on three widely used datasets and highlight key challenges in achieving data harmonization. We call on the broader research community to develop and adopt detailed recommendations for standardized data-sharing practices.
糖尿病管理数据集通常由各种传感器和设备汇编而成，包括糖尿病技术、活动追踪器及其他与健康相关的设备，从而产生异构数据格式。尽管有大量可用数据，但数据集格式和数据共享做法的不一致限制了在先前工作基础上开展研究以及比较不同研究结果的能力。标准化数据共享格式可提高一致性、促进数据集整合并减轻研究人员的数据处理负担。本文探讨了糖尿病管理研究中数据共享做法的现状，并提出了使用统一的时间对齐表格格式来协调数据集的指南。我们展示了这些指南在三个广泛使用的数据集上的应用，并突出了实现数据协调方面的关键挑战。我们呼吁更广泛的研究界制定并采用有关标准化数据共享做法的详细建议。
