Defining a Reference Model for AI-Enabled Dementia and Frailty Care

A health professional using a tablet or digital system with an older person in a clinical or home care setting.

Purpose of the Reference Model

The Digital Healthcare Reference Model (DHRM) defines how healthcare processes, data entities, and decision-support components are organized within the COMFORTage platform. It provides the conceptual foundation for integrating AI-enabled interventions, supporting data-driven clinical workflows, and ensuring alignment across clinical, technical, and organizational actors. The model specifies the structural relationships between patients, professionals, digital tools, and system functions, enabling consistency, modularity, and scalability across different care environments involved in dementia and frailty management.

DHRM
Figure 1 The Digital Healthcare Reference Model (DHRM) structure, illustrating the relationship between the Entity Model and the High-Level Framework

Methodology: Combining Clinical Practice and System Architecture

To ensure the reference model reflects both clinical and technical constraints, its development followed an iterative methodology based on Action Research. This process combined problem identification, feedback integration, and incremental refinement through multiple project-wide cycles. Clinical partners contributed domain knowledge through structured interviews and workflow documentation, while technical teams translated these inputs into generalizable model elements. The outcome is a reference model that accommodates domain-specific requirements without locking into any single deployment configuration.

Figure 2 Iterative development process of the Digital Healthcare Reference Model (DHRM) following the Action Research (AR) methodology

Extracting Clinical Logic from Pilot Workflows

The model development process focused on systematically extracting clinical workflows from the thirteen pilot studies involved in COMFORTage. Each pilot site documented its care delivery processes through interviews and workshops with clinical stakeholders, direct on-site observations, and analysis of patient interactions. These heterogeneous inputs were synthesized into a generalized clinical logic that supports both shared and site-specific elements. The extracted processes were then translated into a unified model of care delivery and system interaction.

Figure 3 Indicative ER model of the COMFORTage platform

Model Outputs: Entity Definitions and Layered Architecture

The result of the modelling process is a formal specification consisting of two parts. First, the Entity Model captures the core roles, attributes, and relationships that define dementia and frailty care within COMFORTage. Second, the High-Level Framework structures how these entities interact across system layers—from data ingestion and processing to AI-based decision support. Together, these outputs enable a scalable, modular approach to digital healthcare design, while embedding clinical safety, interoperability and regulatory compliance (e.g., GDPR) into the model logic.

COMFORTage framework
Figure 4 High-Level Architectural Framework of the COMFORTage Platform

Future Work: Supporting Adaptation and Deployment

The reference model will be progressively refined based on feedback received from technical validation, stakeholder input, and pilot site evaluation. As pilot implementations proceed, new insights will inform updates to the entity model, architectural mapping, and role specifications. The future version of the model will also reflect more detailed structuring of clinical and research workflows, in alignment with findings from empirical assessments. This continued development process is defined in the project’s roadmap and will be reported in the second model deliverable.

COMFORTage platform
Figure 5 High-level architectural framework of the COMFORTage platform, mapping system entities, data flows and AI-driven decision support mechanisms

**Article written by Jheronimus Academy of Data Science (JADS),Ā a key partner inĀ theĀ COMFORTage project.

Skip to content
Privacy Overview

This website uses cookies so that we can provide you with the best user experience possible. Cookie information is stored in your browser and performs functions such as recognising you when you return to our website and helping our team to understand which sections of the website you find most interesting and useful.