Studyflow vs BPMN

What BPMN provides, what Studyflow adds, what it leaves alone

Studyflow extends BPMN 2.1 rather than replacing it. If you already know BPMN, this page tells you what is standard process notation and what is domain-specific.

What Studyflow inherits from BPMN

Studyflow uses BPMN’s full vocabulary for control flow and structure:

  • Events: start, end, intermediate, boundary, timer, error.
  • Activities: tasks, sub-processes, call activities.
  • Gateways: exclusive, parallel, inclusive, event-based, complex.
  • Sequence flows and message flows.
  • Data objects and stores: with data associations.
  • Pools and lanes.
  • Choreography for modeling multi-agent interactions.

These elements behave exactly as in BPMN and render with standard BPMN shapes. A BPMN-only tool can open the BPMN 2.0 XML export of a .studyflow file and see a well-formed process – the Studyflow-specific elements and attributes degrade to BPMN without breaking the diagram.

What Studyflow adds

Studyflow extends BPMN in four areas where experimental research needs more vocabulary.

Study. A top-level container element that extends bpmn:Process and groups everything in a single research project. It carries study-level metadata (name, version, authors, etc) that doesn’t fit naturally on the process itself.

Domain-specific activity types. BPMN has generic Task, ScriptTask, and UserTask. Studyflow specializes Task into domain-specific types (i.e., cognitive sciences) so the inspector can show the right attributes:

  • CognitiveTask, Questionnaire, Instruction, Rest, VideoGame
  • BehaverseTask (standardized cognitive test from the Behaverse catalog)
  • Script and Manual for codes and manual procedures

Each renders with a distinctive icon while remaining a normal BPMN task underneath.

Research-specific data elements. BPMN’s DataObject and DataStore are generic. Studyflow adds:

  • Dataset – a logical collection (potentially multi-table, multi-modal).
  • Table – a tabular component of a dataset, linked to a schema.
  • Array – a multi-dimensional structure for tensors, images, fMRI, video.
  • Schema – formal column/dimension definitions (CSVW-compatible).
  • DataCatalog – a registry of datasets, potentially across stores.
  • Snapshot – an immutable version of a dataset.

Data operations as task markers. Pipelines need to express how data is transformed, not just that it flows through a task. Studyflow adds a small data operations vocabulary (Map, Filter, Reduce, Group, FlatMap, Transform, Compose) that renders as a small marker on an otherwise normal BPMN task. The control flow stays BPMN; the data semantics become explicit.

Random gateway. Studyflow adds RandomGateway for stochastic branching (random assignment, counterbalancing), with branch probabilities as attributes.

Temporal attributes. BPMN has timer events but no native notion of activity duration, progress, or scheduled start/end. Studyflow attaches these as element attributes so trial timelines and Gantt-style views can be derived from the same diagram. See Views.

What Studyflow deliberately leaves alone

Studyflow does not redefine BPMN’s semantics or its serialization conventions (namespaces, element IDs, extensionElements).

The native .studyflow format is YAML, but it is a one-to-one mapping of the BPMN 2.0 XML serialization, and the modeler exports that XML on demand (File → Save As → BPMN 2.0 XML). The export is standard BPMN XML with dedicated namespaces for the extensions (i.e., elements inside the <extensionElements> blocks, plus extension attributes on standard elements). Any conforming BPMN renderer or analyzer can process it and produce something meaningful, even if it ignores the Studyflow extensions. This design choice maximizes compatibility with existing BPMN tools and libraries.

When you should use plain BPMN instead

If your study is well-served by generic BPMN – for example, a clinical operations workflow with no cognitive tasks, no specialized data structures, and no random assignment – you don’t need Studyflow. Studyflow helps if:

  • The study involves cognitive tests, questionnaires, or other research activities.
  • The diagram needs to describe data flow with explicit transformation semantics.
  • The study includes checklists.
  • The diagram should drive both a participant-facing runtime and an analysis pipeline.
  • The study includes randomization or counterbalancing.
  • The diagram targets a reproducibility or reporting standard (CONSORT, SPIRIT, etc.).