How mobilise-d/mobgap works
Mobgap is a Python toolbox for estimating gait and mobility outcomes from lower-back IMU data. Its principal workflow runs sensor data through gait-sequence detection and per-bout gait analysis, then assembles walking-bout results and aggregates digital mobility outcomes. Users can also run individual algorithms and evaluation pipelines independently. The tree and README establish these capabilities; sampled source confirms the configurable full-pipeline stages and several evaluation wrappers, but not every internal call, so uncertain wiring is omitted.
Main entry points: _mobilised_pipeline.py Full pipeline · base.py Gait dataset · data Dataset loaders · aggregation DMO aggregation
Components 20
Data input
- Dataset loaders
src/mobgap/data - Gait dataset
src/mobgap/data/base.py
Gait analysis
- Full pipeline
src/mobgap/pipeline/_mobilised_pipeline.py - Gait sequences
src/mobgap/gait_sequences - Reorientation
src/mobgap/re_orientation - Initial contacts
src/mobgap/initial_contacts - Laterality
src/mobgap/laterality - Cadence
src/mobgap/cadence - Stride length
src/mobgap/stride_length - Walking speed
src/mobgap/walking_speed - Turn detection
src/mobgap/turning
Outcome assembly
- Stride selection
src/mobgap/wba/_stride_selection.py - Bout assembly
src/mobgap/wba/_wb_assembly.py - DMO aggregation
src/mobgap/aggregation - Gait outcomes
Evaluation
- Evaluation tools
src/mobgap/pipeline/evaluation.py - Validation results
src/mobgap/data/validation_results.py - Algorithm wrappers
src/mobgap/stride_length/pipeline.py
Other components
- Researcher
- IMU data
Connections 21
- Researcher → Dataset loaders: selects data
- IMU data → Dataset loaders: provides recordings
- Dataset loaders → Gait dataset: loads
- Researcher → Full pipeline: runs
- Gait dataset → Full pipeline: provides datapoint
- Full pipeline → Gait sequences: detects sequences
- Full pipeline → Reorientation: configures correction
- Full pipeline → Initial contacts: configures detection
- Full pipeline → Laterality: configures classification
- Full pipeline → Cadence: configures calculation
- Full pipeline → Stride length: configures calculation
- Full pipeline → Walking speed: configures calculation
- Full pipeline → Turn detection: configures detection
- Full pipeline → Stride selection: selects strides
- Full pipeline → Bout assembly: assembles bouts
- Full pipeline → DMO aggregation: aggregates outcomes
- DMO aggregation → Gait outcomes: produces
- Researcher → Algorithm wrappers: evaluates algorithm
- Gait dataset → Algorithm wrappers: provides reference data
- Researcher → Evaluation tools: analyzes performance
- Evaluation tools → Validation results: reads results