Clinical protocol
The protocol JSON defines a study's annotation lanes, labels, relationships and rules, and its available measurement outcomes. It is shared across workspaces and retained with each annotation set. No separate outcome file is needed.
What it controls
- Annotation lanes and labels on the timeline.
- Hierarchy, automatic effects and violation checks.
- Named outcomes: their calculation, event selection and optional eligible-time selection.
Load a protocol JSON during workspace setup. Click a measurement value in the workspace to inspect its calculation and contributing annotations.
Minimal example
{
"name": "Walking study",
"version": "1",
"lanes": [
{"name": "Tasks", "level": 1, "color": "#4a90a4", "labels": ["Walk"], "allow_overlap": false},
{"name": "FOG", "level": 2, "color": "#d36161", "labels": ["FOG"], "allow_overlap": false}
],
"groups": [],
"rules": [],
"measurements": [
{
"id": "fog_count",
"name": "FOG episode count",
"version": "1",
"calculation": "count",
"calculation_version": "1",
"events": {"lane": "FOG", "label": "FOG"},
"scope": {"lane": "Tasks", "label": "Walk"}
}
]
}
The implemented calculations are covered_duration, percentage_coverage, and count.
label: null selects all labels in a lane. scope: null uses the entire observation period. Outcome IDs must be unique within the protocol, selectors must name existing
lanes, and duration/percentage and scope selections require interval lanes. Count can also
select point lanes. An omitted or empty measurements array leaves annotation work available
without adding implicit outcomes.
Outcome version identifies the study's definition; calculation_version identifies the
backend arithmetic contract. Both default to "1". The calculation returns its own unit;
protocol authors do not supply a unit or executable formula. Update the outcome version when
changing its meaning or selections. Saved records retain their effective definitions.
Use a study protocol
Select a JSON file in the Workspace Wizard. RIME retains its contents with the annotation set,
so reopening does not require the original file. The bundled gpfog_schema.json is a complete
working example with all three outcomes. Agree on annotation meanings with the study team;
for an episode count, one retained annotation must represent one episode.