¹Institute for Medical Informatics and Biometry, Faculty of Medicine Carl Gustav Carus, TUD Dresden University of Technology, Dresden, Germany

Text

Introduction: Clinical decision support systems (CDSS) offer great potential for improving the quality of care but have so far seen only limited adoption in clinical practice [1]. A key barrier lies in their predominantly technology-driven development, which often fails to adequately account for the complex needs and work contexts of end users [1]. To address this gap, the junior research group CDS2USE supports a user-centered design (UCD) approach. The focus is on three design dimensions critical for the usability and acceptance of CDSS: context sensitivity, explainability of AI-based recommendations, and user-centered data visualization. The overarching goal is to support the development of clinically relevant, user-centered CDSS that integrate seamlessly into clinical workflows.

Methods:

1. Context Sensitivity: A scoping review was conducted to identify context factors influencing medical decision-making. These were categorized in a card-sorting workshop, hierarchically structured, and visualized as a graphical context model, which was evaluated by usability experts.
2. Explainability: A modular questionnaire was developed and pretested to guide the selection of appropriate explanation concepts. In addition, UCD recommendations for Explanation User Interfaces (XUIs) were derived based on a scoping review.
3. Visualization: A scoping review identified visualization techniques for single patient data.

Based on these findings, an interactive concept for a design aid to support prototyping was developed and evaluated in a focus group with developers.

Results:

1. Context Sensitivity: A total of 84 publications were analyzed, identifying over 900 context factors [2], which were assigned to six entities (patient, physician, peers, family, treatment, institution) and structured hierarchically. The graphical context model is currently being revised based on evaluation feedback.
2. Explainability: Pretest findings indicate that the questionnaire facilitates the targeted selection of explanation concepts and formats, depending on user needs and system characteristics. A mind map summarizes 240 recommendations from the scoping review into 64 overarching design recommendations [3].
3. Visualization: 28 visualization techniques from 78 publications were identified [4] and synthesized into an interactive concept (visualization technique map). The focus group with developers suggested that the visualization map would benefit from a more hierarchical structure and filtering options. The map is currently being refined.

Discussion: The developed tools seek to bridge the gap between technological innovation and clinical applicability in CDSS design. The context model enables a structured understanding of situational conditions and may serve as a foundation for designing context-sensitive systems. The XUI questionnaire has undergone pretesting and holds promise for informing the development of explainable AI-based decisions. The visualization technique map may support targeted presentation of case data and UCD decisions, and facilitate effective communication. Collectively, these tools address key methodological gaps in CDSS development and contribute conceptually across diverse implementation contexts. They can help overcome limitations of technology-driven processes [1], [5] by promoting clinical integration through systematic, user-centered approaches.

Conclusion: The integrated consideration of context sensitivity, explainability, and visualization provides a conceptual foundation for user-centered CDSS development. The tools may support developers and researchers in implementing user-centered approaches and hold potential to contribute to improved acceptance and usability of CDSS in clinical practice.

Funding: BMBF, grant number 01ZZ2002.

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.

Literatur

[1] Raszke P, Giebel GD, Abels C, Wasem J, Adamzik M, Nowak H, et al. User-oriented requirements for artificial intelligence–based clinical decision support systems in sepsis: protocol for a multimethod research project. JMIR Res Protoc. 2025;14:e62704. DOI: 10.2196/62704
[2] Schuler K, Jung IC, Zerlik M, Hahn W, Sedlmayr M, Sedlmayr B. Context factors in clinical decision-making: a scoping review. BMC Med Inform Decis Mak. 2025;25(1):1–18. DOI: 10.1186/s12911-025-02965-1
[3] Jung IC, Schuler K, Zerlik M, Grummt S, Sedlmayr M, Sedlmayr B. Overview of basic design recommendations for user-centered explanation interfaces for AI-based clinical decision support systems: a scoping review. Digit Health. 2025;11:20552076241308298. DOI: 10.1177/20552076241308298
[4] Zerlik M, Jung IC, Schuler K, Sedlmayr M, Sedlmayr B. Visualization techniques for summarizing single patient health data to support physicians’ clinical decisions – a scoping review. Stud Health Technol Inform. 2024;317:314–23. DOI: 10.3233/SHTI240873?.
[5] Khairat S, Marc D, Crosby W, Al Sanousi A. Reasons for physicians not adopting clinical decision support systems: critical analysis. JMIR Med Inform. 2018;6:e8912. DOI: 10.2196/medinform.8912