Skip to main content
Background Animation

Pattern-Powered Requirements: The LEMON Architecture for Reusable Requirement Patterns in Low-Code/No-Code (LCNC) Software Development for Digital Health

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

How can requirement patterns be reused in low-code/no-code tools without being tied to a specific modelling notation?

Problem description, research question and relevance

LCNC software development allows users to build applications by modelling instead of coding.

 

Communicating and reusing software requirements remains a challenge—especially when requirements are repeated across projects or must comply with regulations, as in digital health.

Requirement patterns provide a reusable way to structure common functional needs—such as visualizing test data or configuring access control. However, current approaches for pattern specification are either textual, lack tooling, or are locked into specific modelling notations like BPMN or UML. This mismatch is problematic for LCNC tools, where visual models vary widely, and non-programmers play a key role in requirement elicitation. Thus, we investigate the following question:

  • RQ: How can we design an architecture supported that allows requirement patterns to be specified, catalogued, and reused across LCNC tools?

Methods and procedures in the project

We design the LEMON—a.k.a. Language for spEcifiying and MOdelling patterNs—architecture to answer RQ1. We designed and evaluated the LEMON architecture, which consists of four coordinated modules:

  • Pattern Specification Module: A DSL for defining requirement patterns including inputs, outputs, and transformations, decoupled from any specific modelling language.
  • Catalogue Module: A semantic repository for storing and indexing patterns, enriched with metadata for intelligent search and reuse.
  • Interaction Module: A guided user interface enabling step-by-step pattern instantiation for non-technical users such as clinicians and business analysts.
  • Integration Module: Interfaces for executing patterns and generating models inside LCNC tools, with metamodel-aware configuration.

We demonstrated LEMON with Whatscount GmbH, a digital health company. Patterns like “Visualize Test Data” were formally specified and reused across multiple projects. An online survey with practitioners confirmed the architecture’s potential while surfacing key challenges for adoption.

Figure 1. The LEMON Architecture. Source: REFSQ24 Posters & Tools track, CEUR [1].

Results and findings

  • The modular LEMON architecture enables tool-agnostic reuse of requirement patterns, validated through integration with a real-world LCNC tool in digital health.
  • The modelling assistant allows stakeholders—such as practitioners, patients, and business analysts—to prototype requirements through guided reuse, reducing ambiguity.
  • Practitioner feedback identified three critical enablers for adoption: clear interaction flows, comprehensive pattern catalogues, and integration with enterprise tooling.
  • The architecture shows promise for improving requirement communication and early prototyping in domains like digital health.

Recommendations for practice

  • For LCNC tool developers: Provide APIs and hooks for embedding tools under the LEMON architecture to support requirement-driven development in LCNC.
  • For requirement engineers: Formalize recurring requirements early and store them in reusable DSL-based patterns tailored to your domain and LCNC tool, following the LEMON architecture.
  • For researchers: Investigate the usability and evolution of requirement patterns within DSL frameworks and architectures as LEMON. Evaluate integration scenarios across LCNC environments to check if the LCNC industry is ready to adopt patterns as a way to develop software.

Literature and other sources

[1] D. Mosquera, O. Pastor, and J. Spielberger, “LEMON: A Tool for Enhancing Software Requirements Communication through Requirements Pattern-based Modelling Assistance,” Posters & Tools REFSQ2024. doi: https://ceur-ws.org/Vol-3672/PT-paper4.pdf

Zitierung des Beitrags