Skip to main content
SHIFT | Smart Hospital – Integrated Framework, Tools & SolutionsSHIFT | Smart Hospital – Integrated Framework, Tools & Solutions
Background Animation

Modelling Assistance in LCNC Software Development: Towards a User-Centric Framework for Digital Health and Beyond

Knowledge database Organisation Structures & processes Data management & digitalisation Integration & interoperability Training & digital expertise C.3: Artificial intelligence-based software factory for MedTech applications

Low-Code/No-Code (LCNC) tools empower non-programmers, such as practitioners, innovation managers, and patients, to build software through software modelling.

Problem description, research question, relevance

Low-Code/No-Code (LCNC) software development has emerged as a powerful approach to reduce time-to-market by enabling software generation from models rather than traditional code.

 

However, LCNC tools remain underutilized in practice. A key reason is that these tools often lack intelligent modelling assistance—leaving users, especially in domains like digital health, without adequate support for navigating modelling complexity, constraints, and software quality demands.

To better understand how to improve modelling assistance in LCNC tools, we conducted two focus groups targeting LCNC users: one with non-programmers, the other with professional software developers. Through these sessions, we addressed the following questions:

  • What challenges do LCNC users face during modelling?
  • Which features of current modelling assistants are liked or disliked by them?
  • What unmet needs do users have that are not satisfied by current assistants?

Methods and procedures in the project

We conducted two focus groups using the World Café method. The first group included 11 undergraduate engineering students with recent LCNC training. The second group involved three experienced developers from Whatscount, a Swiss company that develops LCNC-based applications for digital health.

Each group discussed three guiding questions focused on challenges, liked/disliked features, and unmet needs related to modelling assistants. Data were categorized and prioritized using the MoSCoW method to derive practical requirements.

Results and findings

Twelve major challenges were identified, including model complexity, tool interoperability, usability, and lack of guidance for less experienced users. Participants also expressed concerns about runtime performance, reusability, and domain-specific support. Key liked features included debuggers, error hints, and graphic model aids. Disliked aspects involved overly technical interfaces and poor dialog design in assistants.

Ten unmet needs were also identified—six considered “must-haves,” such as undo/redo, clearer assistant interaction, and better documentation. These insights informed a set of 12 user-cantered requirements and the creation of an emerging framework for modelling assistance.

The emerging framework proposes three assistance modules:

  • A) Data Gathering Assistance: Helps users create models from unstructured sources or existing models, boosting modelling speed and clarity.
  • B) Model Refinement Assistance: Enhances traceability, debugging, and error correction.
  • C) Model Maintainability Assistance: Promotes model reuse and long-term consistency.
Figure 1: User-centirc emerging framework for modelling assistance in LCNC tools. Source: Springer Nature Research Challenges in Information Science [1].

Recommendations for practices

  • Centre modelling assistance around user interaction, not just automation. Effective assistants simplify complex tasks and provide meaningful, context-aware guidance—especially for non-programmers such as healthcare practitioners and patients involved in digital health development.
  • Ensure support for model reuse and maintainability. Modelling assistants should not only help create models but also support long-term quality and scalability. This is particularly important in digital health, where non-functional requirements—like safety, reliability, and auditability—are essential.
  • Design with domain-specific support in mind. Modelling assistants should adapt to the domain they serve. In sectors like digital health, assistants must interpret context appropriately and offer guidance aligned with regulatory requirements.

Literature and other sources

[1] Mosquera, D., Ruiz, M., Pastor, O., Spielberger, J. (2022). Assisted-Modeling Requirements for Model-Driven Development Tools. In: Guizzardi, R., Ralyté, J., Franch, X. (eds) Research Challenges in Information Science. RCIS 2022. Lecture Notes in Business Information Processing, vol 446. Springer, Cham. https://doi.org/10.1007/978-3-031-05760-1_27

Zitierung des Beitrags

Back