Metrics Report
From IMU signals to clinically useful gait indicators.
The metrics layer transforms synchronized sensor streams into interpretable measures of rhythm, timing, symmetry, stability, and progress across rehabilitation sessions.
Collect synchronized IMU streams from wearable nodes.
Identify heel strike, toe-off, stance, and swing phases.
Extract cadence, timing, symmetry, and variability metrics.
Summarize results for clinical interpretation and follow-up.
Step & Stride Detection
Identifies complete gait cycles and separates each step into clinically meaningful events.
- Heel-strike and toe-off events
- Step time and stride time
- Cadence and step frequency
Gait-Phase Estimation
Segments the walking cycle so clinicians can inspect how much time is spent in each movement phase.
- Initial contact and loading response
- Mid-stance and terminal stance
- Swing phase duration
Symmetry Analysis
Compares left and right movement patterns to quantify imbalance and track rehabilitation progress.
- Step-time symmetry
- Stride-length symmetry when available
- Session-to-session comparison
Progress Tracking
Aggregates session metrics into trends that make improvements, fatigue, or instability easier to document.
- Baseline comparison
- Temporal variability
- Graphical phase and metric plots
Symmetry Index
A compact measure of left-right imbalance.
Symmetry is calculated by comparing the difference between left and right values against their average. Lower values indicate more balanced movement.
SI = |Left - Right| / (0.5 × (Left + Right)) × 100%
CadenceWalking rhythm and fatigue indicator
Double supportBalance, confidence, and stability cue
Stride variabilityConsistency across repeated gait cycles
SymmetryQuantified left-right movement imbalance
Performance Targets
Engineering thresholds that support clinical usefulness.
These targets guide validation and help the team decide when a metric is reliable enough to present in the dashboard.
- Event detection latency: below 100 ms for real-time feedback.
- Synchronization skew: below 5 ms across wearable nodes.
- Step detection accuracy: above 95% in controlled walking trials.