Scholarlink

<p>The potential of artificial intelligence (AI) to improve the quality of care in vascular surgery is widely discussed; however, a&#xa0;high quality of data is a&#xa0;fundamental prerequisite for reliable AI applications. Routinely collected clinical data are often fragmented, heterogeneous and incomplete, which complicates the analysis and interpretation. Guidelines, such as the Good Clinical Data Management Practices (GCDMP), STROBE and its extension RECORD provide key recommendations for handling clinical data. In particular, the process of data cleaning plays a&#xa0;central role when working with routinely collected data, which are likely to form the basis for many future AI developments.</p><p>This article presents the modelling of clinical data in established, so-called relational databases as a&#xa0;solution approach. By implementing a&#xa0;universal data model spanning diagnosis and procedure, complex vascular surgical treatment pathways can be represented consistently, unambiguously and in an extendable manner.</p><p>The systematic improvement of the quality of routinely collected data is an essential prerequisite for the development and safe implementation of AI-based analyses in vascular surgery. Achievement of this requires close interdisciplinary collaboration between clinicians and database developers.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Datenmodellierung in der Gefäßchirurgie – zwischen Anforderung und Wirklichkeit

  • 期刊论文

  • 发表日期：

    2025年11月26日

查看全文
  • Johannes Hatzl,
  • Oana Bartos,
  • Christian-Alexander Behrendt,
  • Bernhard Dorweiler,
  • Jörg Heckenkamp,
  • Grischa Hoffmann,
  • Benedikt Reutersberg,
  • Christian Uhl

摘要

The potential of artificial intelligence (AI) to improve the quality of care in vascular surgery is widely discussed; however, a high quality of data is a fundamental prerequisite for reliable AI applications. Routinely collected clinical data are often fragmented, heterogeneous and incomplete, which complicates the analysis and interpretation. Guidelines, such as the Good Clinical Data Management Practices (GCDMP), STROBE and its extension RECORD provide key recommendations for handling clinical data. In particular, the process of data cleaning plays a central role when working with routinely collected data, which are likely to form the basis for many future AI developments.

This article presents the modelling of clinical data in established, so-called relational databases as a solution approach. By implementing a universal data model spanning diagnosis and procedure, complex vascular surgical treatment pathways can be represented consistently, unambiguously and in an extendable manner.

The systematic improvement of the quality of routinely collected data is an essential prerequisite for the development and safe implementation of AI-based analyses in vascular surgery. Achievement of this requires close interdisciplinary collaboration between clinicians and database developers.