📄 bioengineering

StaBLE: digital metrics capture balance performance across a wide spectrum of balance tasks and abilities

The Stanford Balance Level Evaluation (StaBLE) is a validated digital assessment tool comprising 18 video-based tasks that effectively quantifies a wide spectrum of balance abilities, overcoming the limitations of existing clinical scales by demonstrating strong correlations with standard measures and the ability to distinguish fall risk across diverse populations.

Original authors: Heigold, H., Ruth, P. S., Muccini, J., Lee, D., Barta, S., Rodrigues, A., Hastie, T., Steenerson, K. K., Delp, S. L.

Published 2026-09-23

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Heigold, H., Ruth, P. S., Muccini, J., Lee, D., Barta, S., Rodrigues, A., Hastie, T., Steenerson, K. K., Delp, S. L.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

Balance is the quiet engine of human movement, the invisible skill that allows us to walk, reach, and turn without toppling. It is so fundamental that we rarely notice it until it fails, at which point the consequences can be severe, especially for older adults. When stability is lost, the result is often a fall, a leading cause of injury and death in the elderly. While doctors have long relied on simple questions about past falls or basic physical tests to gauge a person's stability, these methods often miss the nuances of how well someone actually moves. They struggle to distinguish between a person who is merely steady and one who is exceptionally agile, and they often hit a "ceiling" where even the most skilled individuals receive the same top score as those who are merely adequate. To move beyond these limitations, researchers need a way to measure balance that is as precise as it is accessible, capturing the full spectrum of human stability from the impaired to the elite.

A team of researchers at Stanford University has developed a new approach called the Stanford Balance Level Evaluation, or StaBLE, designed to measure balance with digital precision using nothing more than a smartphone. Instead of relying on a clinician's subjective judgment or a coarse scoring system, this method records a person performing a series of challenging physical tasks and uses computer vision to analyze their movements in detail. The team recruited 180 participants with vastly different abilities, ranging from individuals with medical conditions affecting their balance to professional ballerinas and gymnasts. They asked these participants to perform 18 different tasks, such as standing on one leg with eyes closed, jumping, and walking backwards, all while being filmed by three smartphones set up around them. The system then converted the video footage into a single, continuous score that reflects how well the person completed each objective, from the speed of their movement to the steadiness of their posture.

The results of this new evaluation show that it can capture a much wider range of ability than traditional clinical tests. When the researchers compared the new digital score to existing methods, they found that the old tests often failed to distinguish between high performers. For instance, in standard assessments, nearly all the elite athletes in the study received the maximum possible score, making it impossible to tell who was truly the most balanced. In contrast, the StaBLE score spread these athletes out, revealing subtle differences in their performance that the older tests missed. The new score also correlated strongly with age, showing a clear decline in balance ability as people get older, and it aligned well with how confident people felt about their own balance. Crucially, the digital score was able to separate people who were at risk of falling from those who were not, a distinction that some traditional tests failed to make when looking at people with a history of falls.

To make this assessment practical for everyday use, the researchers investigated whether they could shorten the testing time without losing accuracy. They used statistical methods to identify which specific tasks provided the most valuable information. They found that a small subset of five tasks, including a dynamic balance test where a person reaches in different directions while standing on one leg, could predict the full score with high accuracy. This shorter version could be completed in under ten minutes, making it feasible for use in clinics or community centers. The study demonstrates that by combining a diverse set of physical challenges with digital analysis, it is possible to create a balance assessment that is both objective and sensitive enough to detect improvements or declines in performance. This work suggests that the future of balance testing may lie in accessible, video-based tools that can measure human movement with a clarity that was previously only possible in expensive, specialized laboratories.

Technical Summary: StaBLE – Digital Metrics Capture Balance Performance

Problem Statement
Balance is fundamental to human mobility and independence, yet its assessment remains a significant challenge in clinical and research settings. Current methods suffer from a dichotomy: clinical screening tools (e.g., Mini-BESTest, 4-Stage Balance Test, SPPB) are feasible for widespread use but rely on coarse, ordinal scoring systems that exhibit "ceiling effects," limiting their sensitivity to differentiate high-performing individuals or detect subtle improvements. Conversely, laboratory-based kinematic measures offer high precision but are impractical for broad adoption due to cost, equipment requirements, and the need for specialized operators. Furthermore, retrospective fall history is often used as a proxy for balance ability, but this is susceptible to reporting bias and fails to capture the multi-dimensional nature of balance control. There is a critical need for a quantitative, objective, and accessible balance assessment that overcomes the limitations of existing scales while remaining feasible for diverse populations.

Methodology
The authors developed the Stanford Balance Level Evaluation (StaBLE), a comprehensive protocol designed to capture a wide spectrum of balance abilities using digital motion capture.

Participants: The study recruited 180 participants (ages 18–87) with diverse balance capabilities, ranging from individuals with impairments (vestibulopathy, Parkinson's disease, peripheral neuropathy) to elite balancers (professional ballerinas, acrobats, collegiate gymnasts).
Protocol: The assessment consists of 18 balance-challenging tasks, including static standing, dynamic movements (jumping, reaching), and gait tasks. Five tasks include an eyes-closed component to challenge sensory integration.
Data Collection: Data were captured using OpenCap, an open-source, markerless motion capture system utilizing three smartphone cameras and a laptop. The setup was standardized with specific camera heights, orientations, and background drop cloths to minimize visual noise. A single experimenter could administer the protocol, which took approximately 25–35 minutes to complete the movement trials.
Feature Engineering: From the video data, 76 digital features were extracted to quantify task performance. These features measured biomechanical objectives such as peak vertical center of mass speed, peak plantar flexion, sway deviation, and hold times.
Scoring Algorithm:
1. Normalization: Each feature was normalized to a [0, 1] scale based on the minimum and maximum values observed across the entire dataset. Maximization goals (e.g., speed) and minimization goals (e.g., sway) were scaled accordingly.
2. Aggregation: Normalized features were averaged to create 18 task scores. These were further aggregated into a single StaBLE score, where a higher score indicates better performance.
3. Handling Failure: Participants unable to perform a task or achieve specific objectives were assigned a score of 0 for that feature.

Key Results

Distribution and Ceiling Effects: The StaBLE score exhibited a left-tailed distribution (mean: 0.64, range: 0.08–0.84) across the cohort. Crucially, the score demonstrated no ceiling effects; no participant achieved a perfect score of 1.0, even among elite balancers. In contrast, existing scales showed significant ceiling effects (e.g., 24/26 elite balancers scored the maximum on the Mini-BESTest Rise-to-Toes task).
Validity and Correlations:
- Age: Strong negative correlation with age ( $\rho = -0.76, p < 0.001$ ).
- Self-Efficacy: Positive correlation with the Activities-Specific Balance Confidence (ABC) Scale ( $\rho = 0.60, p < 0.001$ ).
- Clinical Scales: The StaBLE score correlated significantly with estimated scores from the Mini-BESTest ( $\tau = 0.55$ ), 4-Stage Balance Test ( $\tau = 0.55$ ), and Short Physical Performance Battery (SPPB) ( $\tau = 0.57$ ).
Fall Risk Differentiation: The underlying distribution of StaBLE scores differed significantly between participants identified as "at risk" for falling (via CDC STEADI criteria) and those not at risk (Kolmogorov-Smirnov test: $KS = 0.67, p < 0.001$ ). Furthermore, StaBLE scores distinguished between those with a history of falls and those without, a distinction that existing scales (4-Stage Balance Test, SPPB) failed to make statistically significant in this dataset.
Task Subset Selection: Using forward subset selection, the authors identified that a subset of five tasks (Y-Balance, Single-Leg-Stance Eyes-Closed, Gait-Backwards, Gait-Pivot, and Single-Leg-Stance Rise-to-Toes Eyes-Closed) could explain 88% of the variance in the full StaBLE score. This suggests the potential for a reduced-time assessment (under 10 minutes) while maintaining predictive power.

Significance and Claims
The paper claims that StaBLE represents a significant advancement in digital health metrics for balance by:

Overcoming Ceiling Effects: By utilizing continuous kinematic features rather than ordinal clinical scoring, StaBLE differentiates performance across the entire spectrum of ability, from impaired to elite, without hitting a performance ceiling.
Accessibility and Scalability: The protocol leverages ubiquitous smartphone technology and markerless motion capture (OpenCap), allowing for high-quality data collection in non-laboratory settings (e.g., dance studios, community centers) by a single operator.
Objective Quantification: The score provides an objective, data-driven measure that avoids the subjectivity of clinician observation and the limitations of retrospective self-reporting.
Differentiation of Risk: The metric successfully identifies statistical differences in balance ability between fall-risk and non-risk groups, as well as those with and without fall history, where traditional scales showed limitations.

The authors maintain a modest stance, noting that the score is a "snapshot" of performance at a single point in time and requires further validation regarding test-retest reliability, longitudinal response to training, and prospective fall prediction. They emphasize that while the score identifies balance deficits, it does not yet pinpoint the specific underlying subsystem causes (e.g., reactive balance, which was not included due to equipment constraints). The work serves as a validated baseline for future clinical translation and the development of more specific diagnostic metrics.

Technical Summary: StaBLE – Digital Metrics Capture Balance Performance

More like this