Authoring a WHO SMART DAK (L2)

On this page

1. The L2 artifacts
2. Workflow
3. A mock session
4. Next

This page is generated from site-docs/source/guides-who-smart-dak/ — each section below links to its own source.

✎ Edit

A Digital Adaptation Kit (DAK) is the L2 — machine-readable but implementation-neutral — representation of a WHO SMART Guideline. This guide covers authoring the L2 artifacts with folio-assistant and an LLM.

Skill package: authoring-who-smart-guidelines

---

The L2 artifacts

✎ Edit

BPMN 2.0 source · full-size SVG

Artifact	Skill	Format
Business processes	bpmn-authoring	BPMN 2.0 XML
Decision logic	dmn-authoring	DMN tables
Data dictionary	l2-dak-authoring	Excel / structured
Terminology	terminology-management	code systems / value sets
Review	content-review	criteria-based

The table is the authoring skills, not the DAK. A DAK has ten components (owner, 2026-09-30): the original eight, plus scheduling logic, split out of decision-support logic, and test scenarios. Against the skills above:

#	DAK component	covered above by
1	Health interventions and recommendations	— (cites L1)
2	Generic personas	—
3	User scenarios	—
4	Business processes and workflows	bpmn-authoring
5	Core data elements	l2-dak-authoring, terminology-management
6	Decision-support logic	dmn-authoring
7	Scheduling logic	dmn-authoring (decision tables)
8	Indicators and monitoring	—
9	Functional and non-functional requirements	—
10	Test scenarios	—

A dash means no skill in this table. It does not mean no skill anywhere. It is where to look before starting a component.

Scheduling logic is authored, but not yet formalized. It is written as DMN decision tables, like decision-support logic. But it may not yet have its own L2 logical-model component or L3 artefact (owner, 2026-09-30). That is why the sources count differently, and the difference is recorded here:

• WHO’s DAK figure (slide 3 of cat-harness/library/kg-folio-asst-2026-09-30) shows nine cards, with scheduling logic as #7, and testing drawn beside them.
• The SMART Base DAK logical model (DAK.fsh) declares nine fields: test scenarios in, and scheduling logic carried inside decision-support logic.
• The speaker notes on that slide said eight.
• WHO’s IG starter kit, L2 DAK authoring (https://smart.who.int/ig-starter-kit/l2_dak_authoring.html; source input/pagecontent/l2_dak_authoring.md in WorldHealthOrganization/smart-ig-starter-kit, read from main on 2026-09-30) disagrees with itself. Its introduction counts “9 interlinked components” with scheduling logic as #7 and no testing. Its own table counts 9 with test scenarios as #9, and scheduling logic inside decision-support logic.

Ten is the owner’s count, and this repository encodes it. DAK_COMPONENTS in the DAK layer’s schemas/dak-kinds.ts lists ten, and names scheduling logic in DAK_UNFORMALIZED_COMPONENTS: it has a field name ready (schedulingLogic), but DAK.fsh does not yet declare that field. The other nine are checked against DAK.fsh field for field.

Workflow

✎ Edit

1. Plan — content-plan: enumerate the processes, decisions, and data elements the guideline implies; identify actors (business analyst, clinical SME, terminologist).
2. Author business processes — ask the agent to draft BPMN for each clinical workflow (bpmn-authoring); it produces valid BPMN 2.0.
3. Author decision logic — capture recommendations as DMN decision tables (dmn-authoring), linked to the data dictionary.
4. Author the data dictionary — define core data elements with types, cardinality, and terminology bindings (l2-dak-authoring).
5. Manage terminology — define code systems / value sets and bind them (terminology-management).
6. Validate & review — content-validate then content-review against the DAK criteria.

A mock session

✎ Edit

You: Plan the L2 DAK for the immunization guideline — I need the “administer vaccine” process and the “is the client due” decision.

Assistant: (loads content-plan) Proposed: 1 BPMN process administer-vaccine, 1 DMN decision client-due-for-vaccine, 4 core data elements (client id, antigen, dose number, date administered), 1 value set (antigen codes). Scaffold it?

You: Yes.

Assistant: (loads bpmn-authoring, dmn-authoring, l2-dak-authoring) Created the BPMN with start → check eligibility → administer → record, the DMN table keyed on age/last-dose, and the data dictionary. (calls content_validate) All artifacts valid.

Next

✎ Edit

Turn the L2 DAK into a computable FHIR Implementation Guide → Authoring a WHO SMART IG (L3).