Transforming BPM+ Models to CPG-on-FHIR

This implementation strategy was developed by Mayo Clinical and has been presented by Davide Sottara in several BPM+ community webinars. All of the diagrams below are based on the COVID-19 Severity Classification and Disposition use case, but our project team intends to work on generalizing this guidance and applying the same methodology and toolchain to our Hypertension use case in the near future. Please review the Conceptual Knowledge Architecture as a foundation for this guidance.

This diagram recommends a separation of DMN (Strategy) knowledge from DMN (Inference) knowledge, which aligns with the conceptual knowledge architecture. Strategy knowledge is a component of the Pathway Definition as a means to organize and orchestrate recommendations, whereas the Inference knowledge is included in this architecture as “Derived features” in the Case Summary layer. The Case Model, or BPMN Model, are included here as representation of the decision flow and orchestration knowledge in the pathway.

 

This first DMN decision model represents the Inference Knowledge, which are Derived Features in the Case Summary.

#8 Decision in the following slide corresponds to the #8 shown in the first ED Model Application slide and shows the result of automated transformation from the DMN Inference knowledge to implementation using CQL expressions.

The next decision model represents Strategy knowledge.

This strategy knowledge from a platform-independent DMN decision table is transformed into both FHIR PlanDefinition and CQL logic for implementation.