aiExpert In Practice – #4 – Healthcare Data Model
Shows a summary of the feedback on some parts of a large, public domain healthcare model from Canada
Source model
This data was taken from the CIHI model, available here.
We used exactly the same settings as example #1, and show just a few highlights from a few of the many diagrams in this model.
As with the other example, we make no comment about whether the feedback form aiExpert is good or bad, relevant or stupid. But it’s mostly interesting. Nothing here should be taken as a criticism of this model, which is clearly comprehensive and well thought-out.
Allergy and Intolerance
This is a reat example of where the AI can provide domain-specific feedback. We don’t know if that feedback is exactly correct in this instance, but it might be.
Issue: The model requires each Patient Allergy or Intolerance to be ‘documented in one Encounter Patient’. In practice, allergies/intolerances are often recorded at the patient-level outside of a specific encounter (e.g., allergy list maintained in a longitudinal record). Forcing a required relation to an Encounter may prevent representing patient-level allergy records.
Suggested fix: Patient Allergy or Intolerance documented in zero or one Encounter Patient (optional). Allow allergies to exist as patient-level records independent of an encounter.
Patients and Party Role
This is a model idea which many data modelers will have grappled with: the idea of a person/organization, and the roles they they can take in the world. We have no idea if is this feedback is valid for this domain, but it sure looks interesting.
Issue: Patient is modeled as a subclass of Party Role but the ‘Patient’ class description reads as a person (entity). This mixes the concept of a party (the person/organization) with the role (the fact they are a patient). The model lacks an explicit Party/Person/NonPersonEntity class, which can lead to confusion about which relationships are to the person vs. to the role.
Suggested fix: Introduce a clear Party (or Person) entity class. Make ‘Patient’ a Party Role that references a Party/Person. Update relationships to distinguish between the person (Party) and the role (Patient).
More Insights
aiExpert In Practice - #3 - BPMN Process Models
7 November 2025
This example gets aiExpert to provide feedback on an BPMN 2 process model diagram
Learn MoreaiExpert In Practice - #2 - Archimate 3
6 November 2025
This example gets aiExpert to provide feedback on an Archimate3 Motivation Model.
Learn MoreaiExpert In Practice - #1 - Aeronautical Data model
5 November 2025
This worked example shows how aiExpert was applied to a public-domain data model, and shows the results which the LLM returned
Learn MoreA model without validation is just some pictures
22 June 2023
Why you need your models to be validated.
Learn MoreUsing Derived Connectors to simplify your models
7 June 2023
..and make them more valuable
Learn MoreTechnical and Content-quality
1 June 2023
Maintain quality with tools to find and fix mechanical errors in your modeling
Learn MoreTop 11 model quality tips for Sparx EA modelers
3 November 2022
Discover new insights into your model, and spot and resolve problems.
Learn MoreHow to get your EA models ready for Prolaborate
6 July 2022
To make the most of Prolaborate it's a good idea to start by doing some housekeeping
Learn MoreStep by step to a tailored UI for Enterprise Architect
28 March 2022
The 'out-of-the-box' EA element properties window can be intimidating for a new user, and it's easy to get overwhelmed, miss critical data, or more...
Learn MoreOnly Connect - Exploring traceability in Sparx EA models
1 November 2021
This webinar looks at the importance of consistency to get value from traceability.
Learn MoreWebinar - How to improve model quality and consistency
16 June 2020
This webinar from the EA Global Summit 2020 provides hints and tips for managing models as they grow.
Learn MoreHard and Soft
11 October 2017
No, not about Brexit… More about styles of Business Analyst.
Learn MoreModel curation techniques for EA
6 September 2016
How cleaning, navigating and validating your EA model makes sharing and collaborating much more effective.
Learn More