Signal acquisition
01 · Capture
Acquire signals with traceable placement, sampling and clock synchronisation. Preserve wear-time and signal-quality information so that missingness, activity exposure and sensor artefact remain distinguishable.
SWISSTEP / KNOWLEDGE LIBRARY
The science of movement monitoring
Gait as a longitudinal sensorimotor phenotype: mechanisms, digital mobility outcomes and clinically interpretable change.
Gait as an integrated marker of functional state
Walking feels automatic, yet it relies on close coordination between the nervous system, sensory feedback and the muscles and joints. Injury, pain, fatigue, cognitive demands and emotional state can all influence how we move. Changes in pace, rhythm or coordination can therefore offer clues to a person’s functional state, especially when followed over time. The challenge is to distinguish a meaningful change from the effects of context and measurement. Research on locomotor automaticity explains why walking may demand more attention when this coordination is impaired.
- 1. Heel contact
- 2. Mid-stance
- 3. Heel rise
- 4. Push-off
- 5. Early swing
- 6. Leg advances
The clinical value lies in the pattern of change: pace with preserved rhythm differs from increasing temporal variability; reduced walking volume with stable tested capacity raises different questions from deterioration in both. Repeated observations can characterise trajectories, fluctuations and responses that a cross-sectional score cannot resolve. They support mechanistic hypotheses and targeted assessment, provided acquisition effects and environmental exposure are separated from physiological change.
Observed clinic walking may not represent habitual performance. Background monitoring adds the temporal coverage needed to examine bout distributions, within-day fluctuation and within-person trajectories in ordinary life. Its advantage is ecological and longitudinal: it samples how mobility is expressed repeatedly, not only what a person can demonstrate during an appointment. Parkinson@Home illustrates the importance of context-sensitive interpretation.
From a sensor signal to a clinical decision
Editorial relevance
Evidence
02 · Validate
Establish verification, analytical validity and clinical validity for the intended metric, population and context of use. Examine absolute agreement and responsiveness rather than relying on correlation or a device-wide accuracy claim.
Analysis
03 · Analyse
Segment locomotor bouts and gait events; derive prespecified spatial, temporal and coordinative features. Model within-person change alongside bout composition, speed dependence, uncertainty and missing-data mechanisms.
Interpretation
04 · Get insight
Integrate the estimated change with symptoms, treatment timing and clinical findings. Determine whether it is reproducible, clinically relevant and sufficiently specific to alter a working hypothesis or assessment priority.
Clinical response
05 · Act and reassess
Link the interpretation to a defined review or intervention pathway. Evaluate subsequent functional change, patient-reported outcomes and unintended effects; retain the distinction between detecting deterioration and establishing its cause.
Follow one foot through the gait cycle
One cycle, two alternating limbs
Follow the dark-purple leg from one ground contact to its next. During stance it supports progression; during swing it advances for the next contact.
Dark purple: reference leg · Lavender: opposite leg
Conceptual walking sequence, not measured kinematics. Phase timing varies with speed and gait pattern. Gait-analysis background and terminology.
From a movement signature to a clinical hypothesis
A gait endpoint is generated by a chain of decisions: sensor placement, event detection, bout segmentation, exclusion rules and aggregation. Each changes the estimand. A mean computed across all detected steps weights long bouts differently from a mean of bout-level summaries; changing the proportion of short indoor bouts can shift an outcome without a corresponding change in locomotor capacity. Interpretation therefore starts with the acquisition and analysis model, not the visual appearance of a trend. The V3 framework separates the relevant layers of measurement evidence.
Pace, rhythm, variability, asymmetry and postural control should be treated as related domains rather than interchangeable proxies. Speed couples spatial progression with cadence; temporal variability concerns cycle-to-cycle consistency; interlimb asymmetry concerns a bilateral relationship. Both variability and asymmetry can change with speed and task composition. Their clinical meaning depends on whether those influences were standardised, stratified or explicitly modelled.
The most informative signal may be a dissociation. Improving clinic speed with persistently fragmented daily walking suggests a different problem from concurrent deterioration in capacity and participation. The former directs attention towards symptoms, confidence, opportunity and task demands; the latter strengthens the case for reassessing physiological or functional limitation.
Automaticity provides a mechanistic link between gait and higher-order function. When locomotion competes for executive resources, a secondary task can expose instability that is not apparent during unchallenged walking. In a study of 30 participants with Parkinson’s disease and 28 controls, dual tasking reduced speed in both groups, whereas increased gait variability was observed in the Parkinson’s group. That dissociation is more informative than describing all dual-task effects as general slowing. Yogev and colleagues, 2005.
Affective state is another plausible modifier, but the evidence does not support a universal mapping from movement to emotion. In a 36-participant experimental study, induced sadness reduced step length and speed; a separate observational study of 90 young adults found no significant association between the measured gait parameters and affective or depressive symptoms. These different designs illustrate why emotional modulation and diagnostic discrimination must not be conflated. Experimental emotion induction; observational null findings.
OneStep is a prominent contributor to clinical translation of smartphone gait measurement. Its research collaborations connect analytical comparison with motion capture, postoperative feasibility and associations with patient-reported lower-limb function. These are complementary contributions to the evidence chain. Teams evaluating phone-based monitoring can use OneStep’s research resources to identify relevant protocols, then assess the original study against the intended population, endpoint and clinical workflow.
“The smartphone application can be a valid, reliable and feasible alternative to motion laboratories”Christensen et al., 2022 · conclusion excerpt
The practical question is parameter-specific transportability: whether the reported agreement, reliability and sensitivity to change remain adequate for the target decision. The paper’s relative underestimation of spatiotemporal outputs and small postoperative cohort should inform that judgement, particularly where an absolute threshold or individual change estimate will influence care.
Read the parameter, then its clinical meaning
PaceFunctional capacity
Walking speed
Pace and global functional performance. Speed integrates spatial progression and stepping frequency and is responsive to several forms of impairment. Its breadth gives it prognostic utility but limited aetiological specificity. Distinguish usual, fast and free-living speed, and report the aggregation window and bout-selection rules.
The pooled survival analysis of 34,485 older adults reported a hazard ratio of 0.88 per 0.1 m/s higher baseline speed (95% CI 0.87–0.90). This population association is not an individual treatment-response function.
RhythmStep frequency
Cadence
Temporal organisation of progression. Interpret cadence jointly with step length and speed. A frequency-driven increase in pace and a length-driven increase represent different adaptations. Verify that the system reports steps rather than strides and that pauses and turns are handled consistently.
In the rehabilitation cohort, cadence means increased across LEFS strata from 90.06 to 99.97 and 109.48 steps/min. The finding describes concurrent functional association, not a target cadence for treatment.
Spatial parameters
Stride length
Spatial scaling and compensatory strategy. Stride length characterises progression over a complete ipsilateral cycle. Interpret absolute and body-size-normalised values separately; speed dependence and compensatory strategies can confound between-person comparison. A spatial asymmetry measure adds information that the bilateral mean omits.
Rozanski and colleagues reported mean stride lengths of 93.17, 107.78 and 121.51 cm across low, medium and high LEFS strata. These are cohort means, not normative boundaries.
Support timingBalance
Double support
Temporal support allocation. Double-support duration reflects the distribution of support across the cycle and commonly covaries with pace. Specify whether the endpoint is total double support per stride, a component interval or a percentage of cycle time. It is not a direct measurement of dynamic stability.
The same LEFS strata had mean double-support values of 35.38%, 31.73% and 28.87%. Cross-sectional separation does not establish a fall-risk threshold.
Interlimb coordination
Asymmetry
Interlimb coordination and load-related adaptation. Specify the bilateral variable, equation, denominator and sign convention. A signed difference preserves laterality; an absolute index can conceal reversal of the affected side. Symmetry can improve through compensation or bilateral deterioration as well as recovery.
For longitudinal interpretation, retain side-specific values and their uncertainty alongside the index. Clinical attribution requires examination of pain, strength, range of motion and task strategy.
Step consistencyMotor control
Variability
Consistency of locomotor control. Variability estimates depend on the number of observations, speed, bout length and whether transitions are included. Separate biological fluctuation from segmentation and event-detection error; report the dispersion statistic and unit of analysis.
In Parkinson’s disease, dual-task-related changes in variability can differ from changes in mean speed. Mean and dispersion therefore should not be treated as redundant endpoints.
Daily activityParticipation
Walking volume
Exposure, participation and behavioural adaptation. Volume, bout duration and fragmentation describe enacted mobility. Their interpretation requires a valid observation denominator: wear time, monitored days and bout-detection rules. A shift in participation is clinically informative even when standardised capacity is preserved.
Analyse coverage and missingness with the outcome. Informative nonwear and changing daily routines can otherwise mimic deterioration or recovery.
Postural controlTurning
Turning and balance
Transitions and postural control. Turning involves reorientation and intersegmental coordination; quiet-standing sway probes a different control problem. Specify task, duration, sensor location and output domain rather than combining these under an undifferentiated balance score.
Select a reference assessment aligned to the construct. A temporospatial walking parameter, a postural-sway endpoint and a multifactorial falls evaluation are complementary, not equivalent.
A published example: gait and perceived lower-limb function
The LEFS-stratified differences support convergent clinical interpretation while preserving the distinction between perceived function and observed performance. Substantial within-group dispersion limits individual classification from group means. Longitudinal responsiveness and clinical utility require separate evaluation from this cross-sectional association.
Explore the knowledge library
Applications
Translate measurements into clinical and research questions.
Research
Examine original studies, cohort sizes and measurement limitations.
Reports & white papers
Find practical assessment and implementation guidance.
Websites & data sources
Locate reusable datasets, methods and research tools.
Regulation & reimbursement
Understand market access and payment pathways.
Methods & glossary
Select a protocol and define each measurement.