Digital Health Technology
Analytical and clinical validation of step counting method in people living with amyotrophic lateral sclerosis
This paper introduces and validates a novel step-counting algorithm for ankle-worn accelerometers, specifically designed to accurately quantify mobility in individuals with Amyotrophic Lateral Sclerosis (ALS). Addressing the limitations of existing methods in handling pathological gait, the algorithm leverages wavelet-based decomposition for robust heel strike detection. Rigorously validated across multiple datasets and real-world conditions, it demonstrates high accuracy in recognizing walking activity and counting steps, offering a scalable, low-burden digital endpoint for ALS research and clinical trials.
Executive Impact: Quantifying Mobility in ALS with Precision
Our advanced AI model extracts key performance indicators, offering unprecedented insights into mobility and disease progression in ALS.
0
Walking Recognition Sensitivity
0
Walking Recognition Specificity
0
Mean Step Counting Bias
0
Steps/Day (Normal Gait)
0
Steps/Day (Walks with Assist)
0
95% LoA for Step Count
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Wavelet-based Step Detection
The proposed algorithm leverages Continuous Wavelet Transform (CWT) on ankle-worn accelerometer data to robustly identify heel strikes. This method is specifically designed to accommodate the slow, variable, and asymmetric gait patterns characteristic of Amyotrophic Lateral Sclerosis (ALS), overcoming limitations of algorithms designed for healthy populations.
Analytical & Clinical Validation
The method underwent extensive validation across multiple datasets, including both healthy individuals and ALS patients, and various activity settings (controlled, semi-controlled, free-living). This robust framework ensures the algorithm's accuracy in diverse real-world conditions and its clinical relevance for monitoring ALS progression.
Scalable Digital Endpoint
By providing accurate and interpretable quantification of walking activity, the ankle-worn accelerometer-based step counting method serves as a promising digital endpoint for ALS research and clinical trials. It offers a scalable, low-burden alternative to traditional, coarse-grained mobility assessments, enabling continuous, high-frequency monitoring.
98%
Walking Recognition Sensitivity (Peak)
The algorithm demonstrated high sensitivity in recognizing walking activities, with peak sensitivity reaching 98% across various walking conditions and body types, confirming its ability to reliably detect movement relevant to mobility assessment.
Enterprise Process Flow
Raw Accelerometer Data (Ankle-Worn)
→
Continuous Wavelet Transform (CWT) on Vertical Axis
→
Frequency-Wise Summation of Wavelet Coefficients
→
Local Peak Detection (Heel Strikes)
→
Temporal Constraint Filtering
→
Total Step Count Calculation
The core of our method lies in its robust signal processing pipeline. By transforming raw accelerometer data into a time-frequency representation, we can accurately pinpoint heel strikes, even amidst the complex and heterogeneous gait patterns of individuals with ALS. This process ensures high precision and reduces the risk of misclassification.
| Feature | Our Method | Traditional ALS Assessment (ALSFRS-R) |
|---|---|---|
| Data Collection |
|
|
| Gait Pattern Adaptability |
|
|
| Objectivity |
|
|
| Sensitivity to Change |
|
|
| Scalability/Burden |
|
|
Our method addresses critical shortcomings of traditional ALS mobility assessments like the ALSFRS-R. By leveraging continuous, objective sensor data and a specialized algorithm for pathological gait, it offers superior sensitivity to subtle changes, scalability, and reduced patient burden, making it a powerful tool for longitudinal monitoring and treatment evaluation.
Impact on ALS Clinical Trials: Quantifying Real-World Mobility
In a clinical validation cohort of 305 people with ALS, the step-counting method successfully differentiated between various levels of gait impairment. Participants reporting 'walks with assist' on the ALSFRS-R averaged 1283 steps/day, significantly lower than those with 'normal' walking (3984 steps/day). Crucially, it registered almost no steps for 'nonambulatory functional movement' (81 steps/day), demonstrating its robustness in severe impairment. This provides a quantifiable, objective measure of daily mobility that can serve as a sensitive digital endpoint, capturing treatment effects or disease progression with greater resolution than existing scales.
Calculate Your Potential ROI
This AI-powered step counting algorithm can significantly reduce the manual effort and subjectivity associated with traditional gait assessment in clinical research and patient monitoring for neurological conditions like ALS. By automating step detection from wearable sensors, it frees up clinician time, improves data accuracy, and enables continuous, real-world data collection, leading to more efficient trials and personalized patient care.
Estimated Annual Savings
$0
Reclaimed Human Hours Annually
0
Implementation Timeline & Roadmap
A phased approach ensures seamless integration and maximum impact for your enterprise.
01: Sensor Integration & Data Stream Setup
Integrate ankle-worn accelerometer devices (e.g., ActiGraph GT3X+, SHIMMER) with existing data capture infrastructure. Establish secure and compliant data streams for continuous raw accelerometer data ingestion. This phase focuses on hardware deployment and secure initial data pipelines.
02: Algorithm Deployment & Backtesting
Deploy the wavelet-based step counting algorithm into your data processing environment. Perform extensive backtesting using historical, annotated datasets (PedEval, PFALS, etc.) to confirm performance metrics (sensitivity, specificity, bias) against ground truth, ensuring robust operation in diverse gait patterns.
03: Pilot Clinical Validation & Optimization
Conduct a pilot study with a small cohort of ALS patients to validate the algorithm's performance in a real-world clinical setting. Gather feedback from clinicians and patients. Optimize algorithm parameters and refine integration based on pilot results to ensure clinical utility and seamless user experience.
04: Scaled Deployment & Longitudinal Monitoring
Roll out the validated step counting solution to a larger patient population or for wider clinical trial deployment. Implement robust monitoring and maintenance protocols. Begin collecting and analyzing longitudinal step count data to track disease progression and evaluate intervention efficacy at scale.
Ready to Transform Your ALS Research?
Accelerate your clinical research with objective mobility insights. Schedule a personalized consultation to explore how our validated step-counting AI can enhance your ALS studies and patient monitoring initiatives.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat