Introduction: Amid ongoing digitalization in healthcare, physicians face increasing challenges in deriving evidence-based recommendations from growing, complex data volumes. Clinical Decision Support Systems (CDSS) offer crucial support by providing tailored recommendations for diagnosis and therapy [1]. However, CDSS often fail to adapt adequately to specific contexts, as evidenced by irrelevant information and warning alerts [2]. This not only disrupts workflows but also reduces system utilization [3]. To improve this, future CDSS must be user-friendly and context-sensitive to ensure appropriate adaptation to environmental requirements. A major challenge in the development is identifying relevant contextual factors in the complex medical environment. Therefore, the aim of this work is to present a concept that aims to develop a tool for identifying these factors to aid in the conceptual design of context-sensitive CDSS, enhancing their effectiveness in clinical settings.

Method: The work comprises two phases: Phase I focuses on developing a methodology to capture contextual elements crucial for medical decision-making, while Phase II evaluates the methodology's practical use. In Phase I, a scoping review following PRISMA-ScR guidelines [4] gathers evidence-based contextual factors. These factors are organized using card sorting and formed into a context model, which is rigorously evaluated, refined for clarity and quality, and validated by physicians and developers for practical utility. In Phase II, this refined model identifies key contextual factors for integration into a context-sensitive CDSS prototype, which undergoes experimental testing. The main goal of this phase is to assess how these factors affect medical decision-making effectiveness to ensure the CDSS enhances decision-making and meets the specific needs of the healthcare environment.

Results: As first results of Phase I, a total of N = 84 relevant articles were identified through a scoping review, yielding N = 774 different context factors. These were systematically categorized and hierarchically structured by n = 4 human-computer interaction experts during a card-sorting workshop. The context factors were assigned to n = 6 entities (attending physician, patient, patient's family, disease treatment, physician's institution, and peers) and incorporated into an initial context model. This model will be iteratively refined and evaluated in the next step of Phase I to develop a method for integrating context factors into future CDSS.

Discussion: The integration of context-based information into CDSS is crucial for ensuring they are appropriately adapted to the environment and the user. As a first step, an initial context model was developed based on the entities influencing the medical decision-making process. This model will serve as the basis for a structured communication tool designed to aid the development of context-sensitive CDSS, enhancing the identification of relevant factors and improving system functionality in clinical settings.

Conclusion: This work emphasizes the critical need to develop context-sensitive CDSS that are tailored to the specific circumstances of the medical environment. The concept of this work provides a framework for the integration of relevant contextual factors into CDSS, which can improve their effectiveness and applicability in the clinical setting to support medical decision-making and thus improve patient care.