Next Article in Journal
Robot-Assisted Mirror Rehabilitation for Post-Stroke Upper Limbs: A Personalized Control Strategy
Previous Article in Journal
Design and Implementation of a Cascade Control System for a Variable Air Volume in Operating Rooms Based on Pressure and Temperature Feedback
 
 
Search for Articles:
Title / Keyword
Author / Affiliation / Email
Journal
Article Type
 
 
Advanced Search
 
Section
Special Issue
Volume
Issue
Number
Page
 
Journals
Sensors
Volume 25
Issue 18
10.3390/s25185657
sensors-logo
Submit to this Journal Review for this Journal Propose a Special Issue
► Article Menu

Article Menu

share Share announcement Help format_quote Cite question_answer Discuss in SciProfiles
first_page
settings
Order Article Reprints
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

Every Step Counts—How Can We Accurately Count Steps with Wearable Sensors During Activities of Daily Living in Individuals with Neurological Conditions?

by
Florence Crozat
1,†,
Johannes Pohl
2,†,
Chris Easthope Awai
2,
Christoph Michael Bauer
1 and
Roman Peter Kuster
1,*
1
Therapy Science Lab, Lake Lucerne Institute, 6354 Vitznau, Switzerland
2
Data Analytics & Rehabilitation Technology (DART), Lake Lucerne Institute, 6354 Vitznau, Switzerland
*
Author to whom correspondence should be addressed.
†
These authors contributed equally to this work.
Sensors 2025, 25(18), 5657; https://doi.org/10.3390/s25185657
Submission received: 22 July 2025 / Revised: 27 August 2025 / Accepted: 9 September 2025 / Published: 11 September 2025
(This article belongs to the Section Wearables)
Browse Figures
Review Reports Versions Notes

Abstract

Highlights

What are the main findings?
  • In individuals with neurological conditions, steps are most accurately counted with a waist sensor, 0.5–3 Hz filter, 5 s window, and gradient boosting regressor. Sensor location has the largest impact on accuracy, followed by window length, regressor type, and filter range.
  • Algorithms trained on able-bodied data detect only 11–47% of steps taken by individuals with neurological conditions during activities of daily living.
What is the implication of the main finding?
  • Population-specific algorithms are essential for accurate step counting in individuals with neurological conditions.
  • Future algorithm developments should incorporate the sensing and analysis configuration identified in this study.

Abstract

Wearable sensors provide objective, continuous, and non-invasive quantification of physical activity, with step count serving as one of the most intuitive measures. However, significant gait alterations in individuals with neurological conditions limit the accuracy of step-counting algorithms trained on able-bodied individuals. Therefore, this study investigates the accuracy of step counting during activities of daily living (ADL) in a neurological population. Seven individuals with neurological conditions wore seven accelerometers while performing ADL for 30 min. Step events manually annotated from video served as ground truth. An optimal sensing and analysis configuration for machine learning algorithm development (sensor location, filter range, window length, and regressor type) was identified and compared to existing algorithms developed for able-bodied individuals. The most accurate configuration includes a waist-worn sensor, a 0.5–3 Hz bandpass filter, a 5 s window, and gradient boosting regression. The corresponding algorithm showed a significantly lower error rate compared to existing algorithms trained on able-bodied data. Notably, all algorithms undercounted steps. This study identified an optimal sensing and analysis configuration for machine learning-based step counting in a neurological population and highlights the limitations of applying able-bodied-trained algorithms. Future research should focus on developing accurate and robust step-counting algorithms tailored to individuals with neurological conditions.

Graphical Abstract

1. Introduction

The importance of physical activity for quality of life, morbidity, and mortality is well documented not only in the general population but also among individuals living with neurological conditions [1,2,3]. However, individuals with neurological conditions are considerably less active than their healthy peers [3,4]. For example, stroke, the leading reason for neurological rehabilitation and single biggest driver of years of life lived with disability [5], is associated with markedly low physical activity levels. Stroke survivors accumulate only about half the number of daily steps compared to the general population [6,7,8,9,10]. This lack of physical activity is clinically relevant: It is a major lifestyle-related risk factor for stroke recurrence and a strong predictor of long-term disability, while reaching adequate physical activity levels after stroke is associated with substantial reductions in serious adverse events [11,12,13].
Wearable motion sensors have become the preferred method for objectively tracking physical activity, overcoming the recall bias and social desirability effects that limit self-reported questionnaires [14,15,16]. Modern wearable devices for remote monitoring typically incorporate a triaxial accelerometer, often combined with a gyroscope, and are commonly worn on the wrist, waist, chest, thigh, ankle, or lower back [17,18,19,20,21,22,23]. From the raw sensor signal, researchers derive a wide range of physical activity outcomes, including time spent at different intensity levels, energy expenditure, step count, activity recognition, and activity counts [17,18,21,22,24,25].
From a therapeutic perspective, step count is a particularly valuable outcome because it provides a tangible, meaningful, and easily interpretable metric representing a fundamental unit of human activity that is widely used [8,22,26,27,28]. Step count is seen among the top outcomes for assessing physical activity frequency/volume by researchers and clinicians [24]. An accurate step counter could be used to monitor a patient’s progress over time—detecting both improvements and deteriorations—and to identify patients in need of a physical activity intervention. Furthermore, individualized real-time feedback from step counters could be a fundamental part of the physical activity intervention itself [29,30]. However, the gait patterns commonly observed in neurological populations—characterized by slow walking speeds, short stride lengths, brief and irregular walking bouts, asymmetric walking, compensatory upper body movements, and the use of walking aids—differ substantially from those of able-bodied populations, for which most available step-counting algorithms are designed [31,32,33,34]. For this reason, there is a need to evaluate how steps can be accurately counted with wearable sensors during activities of daily living (ADL) in individuals with neurological conditions [22].
The primary aim of the present study is to determine the optimal sensing and analysis configuration for a neurological population-specific machine learning algorithm to count steps during ADL. The secondary aim is to compare the accuracy of such an algorithm with algorithms trained on able-bodied data. The focus on ADL separates the present study from the vast majority of studies investigating the accuracy of step counting in neurological populations during standardized clinical assessments, such as the 2 or 6 min walking test [35,36,37,38,39,40].

2. Materials and Methods

This experimental, cross-sectional study recruited inpatients from the Cereneo Neurorehabilitation Clinic in Hertenstein, Switzerland. This study was reviewed under the Swiss Federal Human Research Act and deemed not to require ethical approval due to its technical focus. All applicable ethical guidelines including the latest version of the Declaration of Helsinki were followed [41].

2.1. Participants

Participants were recruited from the clinic’s inpatient population of neurologically impaired adults. For this purpose, a research assistant visited the patients to explain the study. Each patient was free to consent to the study; participation did not affect their treatment. Eligibility criteria included being at least 18 years old, having the ability to understand written and oral study instructions, and having a Functional Ambulation Category score of ≥2 (walking ability with minimal physical assistance, able to support body weight independently). Medical staff approved participant enrolment. All participants provided informed consent prior to study inclusion, and data were pseudonymized to protect privacy.

2.2. Data Collection

Data collection took place between 5 November and 18 December 2024 at the clinic facility. The protocol of the present study was designed to capture short indoor walking bouts during typical ADL—varying in length, shape, and speed. Walking was performed under both single- and dual-task conditions (e.g., walking with a cup of coffee), including stair use when deemed safe by the therapist. Activities included the morning routine (e.g., getting dressed, brushing teeth), household tasks (e.g., carrying and fetching items, preparing food), and leisure activities (e.g., walking to the couch, drinking coffee). The detailed protocol can be found in Supplementary Material, Table S1. The entire recording lasted approximately 30 min and was supervised by a trained therapist. To reflect ADL, the recorded activities were not standardized. Participants were free to perform them in their own way; only the order was prescribed.
Participants were video recorded (30 Hz) using a handheld Galaxy S23 Ultra (Samsung, Seoul, Republic of Korea) and wore seven triaxial inertial measurement units (AX6, Axivity, Newcastle, UK) recording at 100 Hz (±8 g). Sensors were placed in accordance with previous research at the wrist (left and right, worn with bracelet), chest (medial, taped, between the shoulder blades), lower back (medial, taped, at the level of thoracic vertebra 7), waist (left, clipped to belt or trousers), thigh (left, taped, mid-thigh), and ankle (left, with bracelet, above the malleolus) [17,18,19,20,21,22,23]. Additional sensors not analyzed here were placed on both hands (taped, middle finger axis), both upper arms (taped, mid-biceps), and the right ankle. To synchronize video with sensor data, the left wrist sensor was tapped at both the start and end of the recording while in view of the camera. Video timestamps were subsequently linearly interpolated to synchronize with the sensor timeline [18].

2.3. Ground Truth

An experienced annotator (F.C.) manually labelled steps on the video recording using ELAN 6.8 (The Language Archive, Nijmegen, The Netherlands). Steps as part of walking (defined as transferring weight off from one foot to lift it and subsequently reloading it after displacement in space) and shuffling (defined as transferring weight off from one foot and subsequently reloading it without displacement in space) were labelled, i.e., the moment of reloading was labelled [42]. Data were excluded if the participant’s feet were not visible in the video for more than one second (e.g., toilet visits).
To assess inter-annotator agreement, a second experienced annotator (R.P.K.)—blinded to the first annotator’s labels—independently labelled three participants representing different Functional Ambulation Category scores.

2.4. Data Processing

Data were processed in Python (version 3.9.21). The selection of sensing and analysis parameters followed the review of Boukhennoufa et al. (2022) [18]. This included (1) sensor location, (2) filtering the raw acceleration signal, (3) windowing (i.e., segmenting the signal into windows) for feature extraction, and (4) training various regressor types to count step events. In line with this review, this study is limited to a window-based approach for feature extraction, which is often considered the default procedure, although simpler alternatives such as threshold crossing or peak detection exist. Only accelerometer data were used in this study to minimize power consumption and, therefore, facilitate future long-term field monitoring. Moreover, a previous investigation found that including a gyroscope provided no additional benefit to quantify physical activity [43].

2.4.1. Sensor Location

Accelerometer data from all locations (left and right wrist, chest, lower back, waist, thigh, and ankle) were analyzed. Each sensor location was analyzed independently. While combinations of sensor locations may offer additional insights, they were not considered in this study as the increased complexity and burden of managing multiple sensors limit their feasibility for future long-term field applications.

2.4.2. Filter Range

Human gait generates acceleration components predominantly below 5 Hz, with the fundamental walking frequency typically not exceeding 3 Hz [44]. However, rapid limb movements, sudden changes, or impact peaks might push energy into higher frequency bands. To investigate the impact of signal filtering, the signal was processed using three a priori selected 5th-order Butterworth filters with zero-phase (forward–backward) implementation: wide-band (low-pass, 15 Hz cut-off frequency), medium-band (band-pass, 0.2–5 Hz), and narrow-band (band-pass, 0.5–3 Hz).

2.4.3. Window Length

Signal features were calculated using sliding windows with 50% overlap across three window sizes: short (0.5 s), medium (2 s), and long (5 s). Time-domain (e.g., mean, standard deviation, number of peaks) and frequency-domain features (e.g., energy, dominant frequency, magnitude of dominant frequency) for the vertical and anteroposterior axes and correlation-based features (e.g., cross-correlation, autocorrelation) between the axes were computed (full feature list given in Supplementary Material, Table S2). The vertical and anteroposterior axes have been found to be most reliable in another patient population [45]. Axes were determined according to the SciKit Digital Health method (vertical: highest mean acceleration, anteroposterior: strongest correlation to vertical axis) to ensure signal processing is independent of coordinate system orientation [46].

2.4.4. Regressor Type

Informed by the literature, five methods were evaluated [18]: Gradient Boosting (GB), k-Nearest Neighbors (kNN), Multilayer Perceptron (MLP), Random Forest (RF), and Support Vector Regression (SVR). All regressor types used the signal features as input and step count as output. To reduce the likelihood of overfitting and obtain realistic estimates of generalizability, a leave-one-subject-out (LOSO) cross-validation scheme was employed with three folds. LOSO is well-suited for wearable sensor data, which typically exhibits high inter-subject and low intra-subject variability. It ensures that each validation fold involves a subject not seen during training, thereby simulating deployment to new users [47]. The three test subjects for each parameter combination were randomly drawn from the sample. Each regressor was trained to minimize the root mean squared error (RMSE).

2.5. Data Analysis

2.5.1. Annotator Agreement

Inter-annotator agreement was evaluated using the mean absolute percentage error (MAPE) and Pearson correlation coefficients, calculated for each of the three investigated window lengths (short, medium, long). These metrics compare the number of steps identified by the second annotator to the ground truth established by the first annotator.

2.5.2. Optimal Sensing and Analysis Configuration

To answer the primary research question, a generalized linear model was used to estimate the effect of sensor location (reference: waist), filter range (reference: wide-band), window length (reference: medium), and regressor type (reference: GB) on step-counting accuracy. The outcome variable was the median RMSE scaled to 1 s.
A generalized linear model with a gamma distribution and log-link function was applied to account for the positively skewed RMSE values. Model assumptions were assessed visually and statistically (Kolmogorov–Smirnov test). The short window length condition was excluded due to clear violations of distributional assumptions, including a bimodal distribution with substantially higher RMSE values for the short compared to the other window lengths (see Supplementary Material, Figure S1). If no parameter level significantly outperformed its reference, the reference was retained for the algorithm comparison. Level of significance was set to 0.05. Due to the log-link function, results are directly reported as effect sizes with 95% confidence intervals, p-values, and RMSE values. The predicted RMSE for each parameter level represents the expected RMSE when that level is used in combination with all other parameters set to their reference. Parameter importance was quantified by expressing the RMSE of the worst-performing level relative to that of the best-performing level within each parameter.

2.5.3. Algorithm Comparison

To analyze the secondary research question, the optimal sensing and analysis configuration identified in the previous step was used to train a step-counting algorithm using a LOSO cross-validation approach with seven folds. The performance of this neurologically specific algorithm was subsequently compared to four established algorithms originally developed for able-bodied individuals:
  • Threshold-Crossing Algorithm (TCA) detects a step when the vector magnitude of acceleration exceeds a fixed threshold (here: 0.3 m/s2) [48].
  • Continuous Wavelet Transform (CWT) applies a Morlet wavelet transform to the signal, followed by peak detection. The wavelet scale adapts to walking speed, improving robustness to temporal variation [49].
  • SciKit Digital Health (SKDH) is a pre-trained machine learning algorithm designed for lower back sensor data [46].
  • OxWearables (OxW) is a pre-trained machine learning algorithm designed for wrist worn sensors [50].
TCA and CWT were applied to the optimal sensor location identified in step 1; SKDH and OxW were applied to the original sensor location (lower back and wrist, respectively). Algorithm performance was compared using the mean absolute percentage error (MAPE) between predicted and ground-truth step counts. Normality was assessed visually (Q–Q plot, see Supplementary Material, Figure S3) and statistically (Shapiro–Wilk test). If the assumption of normality was not met, a non-parametric Friedman test was applied. Level of significance was set to 0.05. Results are reported with median and interquartile range.

3. Results

This study included seven neurological patients (6 male, 1 female) with stroke (n = 4), Parkinson’s disease (n = 1), spinal cord injury (n = 1), and hydrocephalus (n = 1). The mean age was 72.7 ± 10.8 years (range: 57–90 years). All participants were right-side dominant and had Functional Ambulation Category scores of 5 (n = 2), 4 (n = 3), 3 (n = 1), and 2 (n = 1). One participant used a walking aid (cane).
The mean recording duration was 34.7 ± 8.8 min (range: 20–46 min), during which participants collected an average of 775 ± 301 steps (range: 443–1156 steps). The measurement of one participant had to be repeated due to video loss caused by a software update of the camera. Inter-annotator agreement was high, with an MAPE of 1.3% and Pearson correlation coefficients exceeding 0.99 for each of the evaluated window lengths (short, medium, long).

3.1. Optimal Sensing and Analysis Configuration

After excluding the short window length (see Supplementary Material, Figure S1), visual inspection of model residuals (Supplementary Material, Figure S2) suggested good model fit. This was confirmed by the Kolmogorov–Smirnov test (p = 0.8275), indicating no violation of distributional assumptions.
Significant variations in step-counting accuracy were observed across all four parameters (Table 1). Sensor location had the largest influence on accuracy: RMSE was 100.2% higher for the worst performing location (right wrist, 0.61) compared to the best (waist, 0.30, p ≤ 0.001). Window length had the second largest effect: RMSE was 29.6% higher for the worst performing length (medium, 0.30) compared to the best (long, 0.23, p ≤ 0.01). Regressor type had the third largest effect: RMSE was 14.1% higher for the worst performing model (kNN, 0.34) compared to the best (RF, 0.30, p ≤ 0.01). Finally, filter range showed the smallest effect: RMSE was 10.9% higher for the worst performing range (wide, 0.30) compared to the best (narrow, 0.27, p ≤ 0.01).
Based on these results, the optimal configuration uses the waist sensor, the narrow filter band (0.5–3 Hz), the long window (5 s), and GB.

3.2. Algorithm Comparison

Compared to the ground-truth step count, the algorithm trained with the optimal sensing and analysis configuration counted a median of 86.4% of total steps [interquartile range: 10.2%]. In contrast, CWT counted 47.4% [21.2%], SKDH counted 39.5% [23.2%], TCA counted 18.1% [33.6%], and OxW counted 11.2% [21.7%] of total steps.
This corresponds to an MAPE of 13.6% [9.3%] for GB, 52.6% [21.3%] for CWT, 60.5% [23.3%] for SKDH, 81.9% [33.6%] for TCA, and 88.8% [21.7%] for OxW (Figure 1). The Friedman test indicated a significant difference in algorithm accuracy (p = 0.002). A non-parametric test was chosen because the Shapiro–Wilk test approached the level of significance (p = 0.053), and the Q–Q plot showed substantial deviations from normality (see Supplementary Material, Figure S3).

4. Discussion

This study investigated the accuracy of step-counting algorithms during ADL in individuals with neurological conditions, who typically exhibit altered gait patterns compared to healthy peers. Impaired gait and limited mobility significantly affect functional independence and quality of life [51,52], and step count serves as a clinically meaningful outcome measure to quantify physical activity [24,26,27]. The results of this study demonstrate that step-counting accuracy varies significantly across sensing and analysis parameters. The optimal configuration—a waist sensor combined with a 0.5–3 Hz filter range, a 5 s window, and a gradient boosting regressor—achieved the lowest error rate (mean MAPE of 13.6%). In contrast, the algorithms developed for able-bodied individuals performed markedly worse in this sample (mean MAPE of 52.6–88.8%).
Using step frequency values for stroke survivors during controlled clinical walking assessments (1.66 steps/s during the 6 min walking test [40]), these error rates translate to 0.23 steps/s for the population-specific algorithm and 0.87–1.48 steps/s for the able-bodied-trained algorithms. These values exceed the error margins typically observed in controlled clinical settings (e.g., 2 or 6 min walking test), where step-counting errors range from 0.04 to 0.45 steps/s [36,37,39,40]. However, similar error margins have been observed in other studies conducted outside standardized test conditions (MAPE: 27.0–320.8% or 0.45–5.34 steps/s) [53]. Furthermore, similar error margins are found when using step frequency values observed in the present study (775 steps in 34.7 min or 0.37 steps/s on average): 0.05 steps/s for the population-specific algorithm and 0.19–0.32 steps/s for the able-bodied-trained algorithms. Whether these error margins are sufficient for rehabilitation monitoring remains unclear, as clinical relevance likely depends on disease-specific factors, mobility levels, and total daily step counts. Of note, all algorithms investigated in the present study underestimate the number of steps compared to the ground truth. We speculate that atypical gait patterns (e.g., shuffling with weak acceleration peaks) and isolated single steps outside a gait sequence are potential reasons for this observation and recommend future studies to investigate the misclassification pattern in greater temporal detail.
Sensor location showed the largest effect on step-counting accuracy among the investigated parameters, with the waist yielding a significantly lower RMSE compared to all other locations. Notably, the lower back and ankle sensors also demonstrated acceptable performance, with effect sizes within 10% of the waist sensor (Table 1). This is in line with a previous study that estimated distance walked in stroke survivors and identified the waist and ankle as the most accurate location [54]. While one could expect the ankle to outperform the waist due to its proximity to the ground, the unilateral placement may have limited the ankle’s ability to detect steps of the contralateral foot. This might be particularly relevant during ADL including steps that are not part of bilateral gait patterns, such as brushing teeth or preparing food. The value of sensors mounted on the core, such as the waist and lower back, has previously been shown with respect to gait asymmetries [55]. The observed difference in performance between the waist and lower back was unexpected and warrants further investigation. We pre-selected the waist as the reference because it is the most frequently used sensor location for measuring physical activity [17,18,21] as well as for practical reasons related to wear comfort and donning/doffing. For long-term field monitoring, a waist-worn sensor clipped to the belt or trousers likely offers superior wear comfort compared to a taped sensor on the lower back, which could interfere with backrests during seated activities and might require assistance for donning/doffing. Conversely, from a data perspective, taping a sensor may improve the accuracy of sensor placement and wear compliance, particularly over long wear periods. When combined with previous research of our group, the ankle might be the preferred choice if step counting is to be combined with a tracking of body posture, while the waist and lower back might be the preferred choices if step counting is to be combined with other physical activity outcomes [43]. For the remaining locations, accuracy declined with increasing distance from the feet (thigh to chest to wrist), a trend consistent with expectations based on biomechanical signal attenuation and movement variability at more cranial/distal locations during ADL. However, when considering wearing comfort, the wrist could become a very interesting alternative. The wrist would allow combining step counting with upper extremity usage, a very well researched topic [17,18]. However, the results observed in this study (Table 1, Figure 1) make the wrist the least preferred choice from an accuracy point of view.
Window length was the second most influential factor, with longer windows (5 s) yielding a lower RMSE rate compared to shorter ones (2 s). The shortest window length (0.5 s) had to be excluded from the analysis due to very poor performance on the RMSE (see Supplementary Material, Figure S1). Window length determines the amount of temporal information captured in each analysis segment (i.e., feature). Very short windows may capture only partial steps, especially in individuals with slow or irregular gait, and are more susceptible to motion artifacts during non-walking activities, making feature extraction less meaningful and stable and, ultimately, negatively impacting step-counting accuracy. In contrast, long windows are more likely to encompass complete or multiple steps, resulting in more robust and representative feature sets. The longest window length investigated in this study yielded the lowest RMSE. As previous research has used even longer windows, up to 10 s [18], future studies should evaluate whether extending window lengths further improves performance during ADL, where many steps may occur outside of structured walking bouts. Conversely, if the goal is to reduce window length to increase temporal resolution for real-time applications and enable the detection of individual steps that are not part of walking bouts, future research should consider alternative feature sets, e.g., without correlation-based features that might be less reliable in very short segments.
Regressor type also influenced performance, with ensemble methods such as GB and RF outperforming simpler models like kNN. This suggests that more complex, non-linear models are better suited to capturing the heterogeneous and irregular movement patterns observed in individuals with neurological conditions and confirms previous research [18]. Although RF yielded a slightly lower RMSE, its performance did not significantly exceed that of GB. In line with our predefined procedure of retaining the reference when no alternative significantly outperforms it, GB was selected for the optimal configuration. Nevertheless, RF demonstrated comparable accuracy and may be considered an equally valid alternative. Ensemble methods aggregate multiple weak learners, enabling them to model subtle, high-dimensional relationships between features and step events, which is particularly valuable when signal characteristics vary between individuals. In contrast, simpler models such as kNN rely heavily on local similarity in feature space and may be more vulnerable to inter-subject variability, which is expected to be pronounced in clinical populations. Interestingly, while neural network models like the MLP offer theoretical flexibility, they did not outperform tree-based methods in this study, potentially due to the limited size of the training data. Our findings support the use of ensemble learning methods in wearable sensor-based gait analysis, particularly in scenarios involving small datasets, which are common in research on clinical populations. Future studies with larger samples should systematically compare ensemble methods with neural network models to better understand their relative strengths.
The filter range, the fourth processing parameter investigated, showed a moderate influence on accuracy. The narrowest band (0.5–3 Hz) yielded the lowest RMSE, likely because it captures the frequency components most relevant to gait while effectively suppressing high-frequency noise and low-frequency drift. The medium band (0.2–5 Hz) is a viable alternative, whereas the wide band (low-pass with a 15 Hz cut-off) does not provide comparable performance.
The algorithm comparison showed that the neurological population-specific step-counting algorithm, trained with the optimal sensing and analysis configuration, achieved the lowest error rate. The able-bodied algorithms reached MAPEs that were 4 to 6.5 times higher. However, it is important to note that this comparison is inherently biased. The population-specific algorithm—despite using LOSO cross-validation—was both trained and validated on the same sample, while the able-bodied algorithms used independent development samples. LOSO cross-validation makes overfitting much more likely and thus limits generalizability of the presented algorithms as compared to independent validation. Consequently, future research should confirm our findings using an independent neurological sample. Such a study should pay close attention to the composition of the sample, including age, diagnosis, Functional Ambulation Category, and use of assistive devices, and, ideally, use an independent sample for validation. Notably, a previous study observed a 6% decrease in performance when moving from LOSO cross-validation in the development sample to validation in an independent sample of healthy participants [56]. A similar or larger reduction in performance should be expected here. To improve the robustness of the population-specific algorithm, we recommend further training on a larger and more diverse neurological cohort prior to independent validation. Interestingly, the ranking of the able-bodied algorithms’ MAPEs in the algorithm comparison (secondary research question, Figure 1) mirrors the ranking of sensor locations in the search for the optimal sensing and analysis configuration (primary research question, Table 1): the waist (CWT) achieved the lowest error, followed by the back (SKDH) and the wrist (OxW) with the substantially largest error. The only exception is the simple TCA algorithm applied to the waist, which had a higher MAPE than SKDH for the back sensor. This aligns with the regressor comparison, where simpler models like kNN and SVR produced the highest error rates.

Methodological Considerations

A key strength of the present study is its focus on ADL, including typical non-walking tasks such as tooth brushing and food preparation. This contrasts with the vast majority of comparable studies, which rely on standardized clinical gait assessments involving continuous, uninterrupted walking, such as the 2 or 6 min walking tests [35,36,37,38,39,40]. The ADL investigated in this study or by Henderson et al. 2022 are far more representative of patients’ activities in their home environment, strengthening the study’s ecological validity [53]. However, this realism comes at the cost of higher error rates [17,22,36,37,39,40,53,57]. Two key factors contribute to this trade-off. First, the average step frequency in our study was around 0.37 steps/s versus 1.66 steps/s observed in the 6 min walking test [40], reflecting the fragmented, irregular nature of daily movements. Second, recent work has shown that non-stepping tasks are frequently misclassified as steps when algorithms are trained or validated exclusively on structured walking tasks, undermining their generalizability to real-life settings [53]. In this context, it should be noted that this study was limited to indoor ADL performed in a rehabilitation clinic, which may not fully capture the variability encountered in home environments.
This study included a diverse neurological sample (Functional Ambulation Category scores ranging from 2 to 5), with four out of seven patients being stroke survivors. Although the sample is relatively small, this study was sufficiently powered to address the primary research question regarding optimal sensing and analysis configuration, as evident from the significant differences observed for all investigated parameters (Table 1). The sample size in combination with the significance level (0.05) rendered effect sizes of ≥9% statistically significant. Whether smaller effect sizes are clinically meaningful remains a question for future research. We consider the sample size to represent a reasonable balance between the need for statistical power and the ethical responsibility to minimize participant burden. However, the small sample limits the generalizability of the results and precludes subgroup analyses. It was also insufficient to perform a post hoc Wilcoxon test for the algorithm comparison (secondary aim). With seven participants, the smallest possible Wilcoxon p-value is 0.0156 (i.e., 2 × 0.57), which cannot reach significance when correcting for multiple comparisons. Therefore, we recommend that the next step involves training an algorithm using the optimal sensing and analysis configuration identified in this study on a larger neurological sample, followed by a validation on an independent dataset. The limited sample size also supported our choice to express the primary results as effect sizes (Table 1) and to constrain the number of signal features to reduce regressor complexity. Future studies with larger samples should consider incorporating more sophisticated features or broader feature sets and employing evidence-based feature selection methods, such as those described previously [58,59]. Additionally, collecting more than 35 min of data or exceeding 775 steps per participant is advisable to further enhance model robustness. Although sensors were placed on both ankles, only the left ankle was analyzed, reflecting our focus on sensor locations most relevant for field-based monitoring (waist and wrists). While no asymmetries were observed in upper-limb movement, lower-limb asymmetries were not explored. Inter-annotator reliability was verified in a small subset of three participants and showed high agreement, indicating that annotation consistency was adequate for the purposes of this study.
The search for optimal sensing and analysis parameters followed a structured approach using generalized linear modelling to obtain results as robust as possible. An alternative would have been to select the single best-performing parameter combination. However, we argue that, in algorithm development, particularly for neurological populations, the primary goal should not be to maximize performance on a specific dataset but rather to prioritize robustness and generalizability. This is especially critical when the intended application involves remote monitoring of ADL. This focus on robustness is also one of the main reasons why the present study incorporated a broad range of ADL not directly related to walking, supporting this study’s ecological validity.
The present study investigated step count as a measure of physical activity volume and frequency [24]. Other gait-related variables, such as gait speed, stride length, or stance duration, may provide even greater clinical insights. Gait speed, for instance, is a strong predictor of functional status, independence, and quality of life and is sometimes referred to as the “sixth vital sign” [22,60]. For assessing other step-based gait characteristics, such as stride length or stance duration, in real-world environments outside standardized clinical assessments, we view accurate step counting as an essential first step. Therefore, we recommend that future algorithm developments for these gait outcomes build on algorithms that can reliably detect steps [61,62]. We fully acknowledge that gait speed has a substantial impact on step detection accuracy [39,40,63,64]. This likely contributes to the poor performance of algorithms developed on able-bodied individuals when applied to neurological populations. Finally, this study did not distinguish between steps taken during walking and those during shuffling. While this distinction may be less critical for general step counting, it could be important in the context of physical activity, where differentiating gait intensity may be relevant.

5. Conclusions

This study is among the first structured investigations of how sensor location, filter settings, window length, and algorithm type affect machine learning-based step detection in neurological populations during ADL. The results demonstrate that population-specific algorithms are essential for accurate step counting. The optimal sensing and analysis configuration includes a waist-mounted sensor, a 0.5–3 Hz filter, a 5 s window, and a gradient boosting regressor. The corresponding algorithm detected, on average, 86% of the steps taken by the individuals with neurological conditions, while the able-bodied trained algorithm detected ≤47% of the steps. Notably, all algorithms tended to underestimate step counts relative to the ground truth. This highlights the inherent challenges of step detection in individuals with neurological conditions, particularly during non-walking activities that are common in daily life but poorly represented in traditional gait assessments such as the 6 min walking test. Future studies are needed to evaluate the generalizability of these findings in larger and independent neurological samples.

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/s25185657/s1, Figure S1: Histogram of RMSE for window length; Figure S2: Visualizations of generalized linear model fit; Figure S3: Assessing normal distribution for algorithm comparison; Table S1: Detailed study protocol with all tasks; Table S2: Feature list.

Author Contributions

Conceptualization, J.P., C.M.B. and R.P.K.; methodology, F.C., J.P. and R.P.K.; software, F.C. and R.P.K.; validation, F.C. and R.P.K.; formal analysis, F.C. and R.P.K.; investigation, F.C. and J.P.; resources, J.P., C.E.A. and C.M.B.; data curation, F.C., J.P. and R.P.K.; writing—original draft preparation, F.C. and R.P.K.; writing—review and editing, F.C., J.P., C.E.A., C.M.B. and R.P.K.; visualization, R.P.K.; supervision, J.P., C.E.A., C.M.B. and R.P.K.; project administration, J.P. and R.P.K.; funding acquisition, C.M.B. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

The study was approved by Ethikkommission Nordwest- und Zentralschweiz (EKNZ) (Req-2023-00539).

Informed Consent Statement

Informed consent was obtained from all subjects involved in this study.

Data Availability Statement

The raw data supporting the conclusions of this article will be made available by the authors on reasonable request.

Conflicts of Interest

The authors declare no conflicts of interest.

Abbreviations

The following abbreviations are used in this manuscript:
ADLActivities of Daily Living
RMSERoot Mean Square Error
MAPEMean Absolute Percentage Error
GBGradient Boosting
kNNk-Nearest Neighbors
MLPMultilayer Perceptron
RFRandom Forest
SVRSupport Vector Regression
LOSOLeave-one-subject-out
TCAThreshold-Crossing Algorithm
CWTContinuous Wavelet Transform
SKDHSciKit Digital Health
OxWOxWearables

References

  1. Lear, S.A.; Hu, W.; Rangarajan, S.; Gasevic, D.; Leong, D.; Iqbal, R.; Casanova, A.; Swaminathan, S.; Anjana, R.M.; Kumar, R.; et al. The effect of physical activity on mortality and cardiovascular disease in 130,000 people from 17 high-income, middle-income, and low-income countries: The PURE study. Lancet 2017, 390, 2643–2654. [Google Scholar] [CrossRef]
  2. Morley, J.F.; Subramanian, I.; Farahnik, J.; Grout, L.; Salcido, C.; Kurtzer, J.; Mischley, L.K. Physical Activity, Patient-Reported Outcomes, and Quality of Life in Parkinson’s Disease. J. Geriatr. Psychiatry Neurol. 2025. [Google Scholar] [CrossRef]
  3. Yu, C.; Cao, Y.; Liu, Q.; Tan, H.; Xia, G.; Chen, B.; Du, F.; Lu, K.; Saposnik, G. Sitting Time, Leisure-Time Physical Activity, and Risk of Mortality Among US Stroke Survivors: A Prospective Cohort Study from the NHANES 2007 to 2018. Stroke 2025, 56, 1738–1747. [Google Scholar] [CrossRef] [PubMed]
  4. Andersen, T.M.; Andersen, A.M.; Riemenschneider, M.; Taul-Madsen, L.; Diechmann, M.; Gaemelke, T.; Dalgas, U.; Brønd, J.C.; Hvid, L.G. Comprehensive evaluation of accelerometer-based physical activity in persons with multiple sclerosis—The influence of disability status and its impact on walking capacity. Mult. Scler. Relat. Disord. 2025, 93, 106243. [Google Scholar] [CrossRef]
  5. Cieza, A.; Causey, K.; Kamenov, K.; Wulf Hanson, S.; Chatterji, S.; Vos, T. Global estimates of the need for rehabilitation based on the Global Burden of Disease study 2019: A systematic analysis for the Global Burden of Disease Study 2019. Lancet 2020, 396, 2006–2017. [Google Scholar] [CrossRef]
  6. Butler, E.N.; Evenson, K.R. Prevalence of Physical Activity and Sedentary Behavior Among Stroke Survivors in the United States. Top. Stroke Rehabil. 2014, 21, 246–555. [Google Scholar] [CrossRef]
  7. English, C.; Manns, P.J.; Tucak, C.; Bernhardt, J. Physical Activity and Sedentary Behaviors in People with Stroke Living in the Community: A Systematic Review. Phys. Ther. 2014, 94, 185–196. [Google Scholar] [CrossRef] [PubMed]
  8. Fini, N.A.; Holland, A.E.; Keating, J.; Simek, J.; Bernhardt, J. How Physically Active Are People Following Stroke? Systematic Review and Quantitative Synthesis. Phys. Ther. 2017, 97, 707–717. [Google Scholar] [CrossRef] [PubMed]
  9. Fini, N.A.; Bernhardt, J.; Churilov, L.; Clark, R.; Holland, A.E. Adherence to physical activity and cardiovascular recommendations during the 2 years after stroke rehabilitation discharge. Ann. Phys. Rehabil. Med. 2021, 64, 101455. [Google Scholar] [CrossRef]
  10. Gebruers, N.; Vanroy, C.; Truijen, S.; Engelborghs, S.; Deyn, P.P.D. Monitoring of Physical Activity After Stroke: A Systematic Review of Accelerometry-Based Measures. Arch. Phys. Med. Rehabil. 2010, 91, 288–297. [Google Scholar] [CrossRef]
  11. Hobeanu, C.; Lavallée, P.C.; Charles, H.; Labreuche, J.; Albers, G.W.; Caplan, L.R.; A Donnan, G.; Ferro, J.M.; Hennerici, M.G.; A Molina, C.; et al. Risk of subsequent disabling or fatal stroke in patients with transient ischaemic attack or minor ischaemic stroke: An international, prospective cohort study. Lancet Neurol. 2022, 21, 889–898. [Google Scholar] [CrossRef]
  12. Kang, S.M.; Kim, S.H.; Han, K.D.; Paik, N.J.; Kim, W.S. Physical activity after ischemic stroke and its association with adverse outcomes: A nationwide population-based cohort study. Top. Stroke Rehabil. 2021, 28, 170–180. [Google Scholar] [CrossRef]
  13. Turan, T.N.; Nizam, A.; Lynn, M.J.; Egan, B.M.; Le, N.-A.; Lopes-Virella, M.F.; Hermayer, K.L.; Harrell, J.; Derdeyn, C.P.; Fiorella, D.; et al. Relationship between risk factor control and vascular events in the SAMMPRIS trial. Neurology 2017, 88, 379–385. [Google Scholar] [CrossRef]
  14. Adams, S.A.; Matthews, C.E.; Ebbeling, C.B.; Moore, C.G.; Cunningham, J.E.; Fulton, J.; Hebert, J.R. The Effect of Social Desirability and Social Approval on Self-Reports of Physical Activity. Am. J. Epidemiol. 2005, 161, 389–398. [Google Scholar] [CrossRef]
  15. Prince, S.A.; Adamo, K.B.; Hamel, M.E.; Hardt, J.; Gorber, S.C.; Tremblay, M. A comparison of direct versus self-report measures for assessing physical activity in adults: A systematic review. Int. J. Behav. Nutr. Phys. Act. 2008, 5, 56. [Google Scholar] [CrossRef]
  16. Troiano, R.P.; McClain, J.J.; Brychta, R.J.; Chen, K.Y. Evolution of accelerometer methods for physical activity research. Br. J. Sports Med. 2014, 48, 1019–1023. [Google Scholar] [CrossRef]
  17. de Quirós, B.M.; Douma, E.H.; van den Akker-Scheek, I.; Lamoth, C.J.C.; Maurits, N.M. Quantification of Movement in Stroke Patients under Free Living Conditions Using Wearable Sensors: A Systematic Review. Sensors 2022, 22, 1050. [Google Scholar] [CrossRef]
  18. Boukhennoufa, I.; Zhai, X.; Utti, V.; Jackson, J.; McDonald-Maier, K.D. Wearable sensors and machine learning in post-stroke rehabilitation assessment: A systematic review. Biomed. Signal Process. Control. 2022, 71, 103197. [Google Scholar] [CrossRef]
  19. Godfrey, A.; Hetherington, V.; Shum, H.; Bonato, P.; Lovell, N.H.; Stuart, S. From A to Z: Wearable technology explained. Maturitas 2018, 113, 40–47. [Google Scholar] [CrossRef] [PubMed]
  20. Johansson, D.; Malmgren, K.; Alt Murphy, M. Wearable sensors for clinical applications in epilepsy, Parkinson’s disease, and stroke: A mixed-methods systematic review. J. Neurol. 2018, 265, 1740–1752. [Google Scholar] [CrossRef]
  21. Letts, E.; Jakubowski, J.S.; King-Dowling, S.; Clevenger, K.; Kobsar, D.; Obeid, J. Accelerometer techniques for capturing human movement validated against direct observation: A scoping review. Physiol. Meas. 2024, 45, 07TR01. [Google Scholar] [CrossRef]
  22. Peters, D.M.; O’bRien, E.S.; Kamrud, K.E.; Roberts, S.M.; Rooney, T.A.; Thibodeau, K.P.; Balakrishnan, S.; Gell, N.; Mohapatra, S. Utilization of wearable technology to assess gait and mobility post-stroke: A systematic review. J. Neuroeng. Rehabil. 2021, 18, 67. [Google Scholar] [CrossRef]
  23. Sun, Y.; Chen, J.; Ji, M.; Li, X. Wearable Technologies for Health Promotion and Disease Prevention in Older Adults: Systematic Scoping Review and Evidence Map. J. Med. Internet Res. 2025, 27, e69077. [Google Scholar] [CrossRef]
  24. Fini, N.A.; Simpson, D.; Moore, S.A.; Mahendran, N.; Eng, J.J.; Borschmann, K.; Conradsson, D.M.; Chastin, S.; Churilov, L.; English, C. How should we measure physical activity after stroke? An international consensus. Int. J. Stroke 2023, 18, 1132–1142. [Google Scholar] [CrossRef]
  25. Pohl, J.; Ryser, A.; Veerbeek, J.M.; Verheyden, G.; Vogt, J.E.; Luft, A.R.; Easthope, C.A. Accuracy of gait and posture classification using movement sensors in individuals with mobility impairment after stroke. Front. Physiol. 2022, 13, 933987. Available online: https://www.frontiersin.org/journals/physiology/articles/10.3389/fphys.2022.933987/full (accessed on 14 July 2025). [CrossRef] [PubMed]
  26. Bassett, D.R.; Toth, L.P.; LaMunion, S.R.; Crouter, S.E. Step Counting: A Review of Measurement Considerations and Health-Related Applications. Sports Med. 2017, 47, 1303–1315. [Google Scholar] [CrossRef] [PubMed]
  27. Brandenbarg, P.; Hoekstra, F.; Barakou, I.; Seves, B.L.; Hettinga, F.J.; Hoekstra, T.; van der Woude, L.H.V.; Dekker, R.; Krops, L.A. Measurement properties of device-based physical activity instruments in ambulatory adults with physical disabilities and/or chronic diseases: A scoping review. BMC Sports Sci. Med. Rehabil. 2023, 15, 115. [Google Scholar] [CrossRef]
  28. Fini, N.A.; Holland, A.E.; Keating, J.; Simek, J.; Bernhardt, J. How is physical activity monitored in people following stroke? Disabil. Rehabil. 2015, 37, 1717–1731. [Google Scholar] [CrossRef] [PubMed]
  29. Larsen, R.T.; Wagner, V.; Korfitsen, C.B.; Keller, C.; Juhl, C.B.; Langberg, H.; Christensen, J. Effectiveness of physical activity monitors in adults: Systematic review and meta-analysis. BMJ 2022, 376, e068047. [Google Scholar] [CrossRef]
  30. Neumann, S.; Ducrot, M.; Vavanan, M.; Naef, A.C.; Easthope Awai, C. Providing Personalized Gait Feedback in Daily Life. In Converging Clinical and Engineering Research on Neurorehabilitation V; Pons, J.L., Tornero, J., Akay, M., Eds.; Springer Nature: Cham, Switzerland, 2025; pp. 632–635. [Google Scholar]
  31. Coca-Tapia, M.; Cuesta-Gómez, A.; Molina-Rueda, F.; Carratalá-Tejada, M. Gait Pattern in People with Multiple Sclerosis: A Systematic Review. Diagnostics 2021, 11, 584. [Google Scholar] [CrossRef]
  32. Lavelle, G.; Norris, M.; Flemming, J.; Harper, J.; Bradley, J.; Johnston, H.; Fortune, J.; Stennett, A.; Kilbride, C.; Ryan, J.M. Validity and Acceptability of Wearable Devices for Monitoring Step-Count and Activity Minutes Among People with Multiple Sclerosis. Front. Rehabil Sci. 2022, 2, 737384. Available online: https://www.frontiersin.org/journals/rehabilitation-sciences/articles/10.3389/fresc.2021.737384/full (accessed on 25 June 2025). [CrossRef]
  33. Xu, J.; Witchalls, J.; Preston, E.; Pan, L.; Waddington, G.; Adams, R.; Han, J. Stroke-related factors associated with gait asymmetry in ambulatory stroke survivors: A systematic review and meta-analysis. Gait Posture 2025, 121, 173–181. [Google Scholar] [CrossRef]
  34. Zanardi, A.P.J.; da Silva, E.S.; Costa, R.R.; Passos-Monteiro, E.; dos Santos, I.O.; Kruel, L.F.M.; Peyré-Tartaruga, L.A. Gait parameters of Parkinson’s disease compared with healthy controls: A systematic review and meta-analysis. Sci. Rep. 2021, 11, 752. [Google Scholar] [CrossRef]
  35. Costa, P.H.V.; De Jesus, T.P.D.; Winstein, C.; Torriani-Pasin, C.; Polese, J.C. An investigation into the validity and reliability of mHealth devices for counting steps in chronic stroke survivors. Clin. Rehabil. 2020, 34, 394–403. [Google Scholar] [CrossRef] [PubMed]
  36. Fulk, G.D.; Combs, S.A.; Danks, K.A.; Nirider, C.D.; Raja, B.; Reisman, D.S. Accuracy of 2 activity monitors in detecting steps in people with stroke and traumatic brain injury. Phys. Ther. 2014, 94, 222–229. [Google Scholar] [CrossRef]
  37. Katzan, I.; Schuster, A.; Kinzy, T. Physical Activity Monitoring Using a Fitbit Device in Ischemic Stroke Patients: Prospective Cohort Feasibility Study. JMIR mHealth uHealth 2021, 9, e14494. [Google Scholar] [CrossRef]
  38. Macko, R.F.; Haeuber, E.; Shaughnessy, M.; Coleman, K.L.; Boone, D.A.; Smith, G.V.; Silver, K.H. Microprocessor-based ambulatory activity monitoring in stroke patients. Med. Sci. Sports Exerc. 2002, 34, 394. [Google Scholar] [CrossRef]
  39. Mahendran, N.; Kuys, S.S.; Downie, E.; Ng, P.; Brauer, S.G. Are Accelerometers and GPS Devices Valid, Reliable and Feasible Tools for Measurement of Community Ambulation After Stroke? Brain Impair. 2016, 17, 151–161. [Google Scholar] [CrossRef]
  40. Schaffer, S.D.; Holzapfel; Simon, D.; Fulk; George Bosch, P.R. Step count accuracy and reliability of two activity tracking devices in people after stroke. Physiother. Theory Pract. 2017, 33, 788–796. [Google Scholar] [CrossRef]
  41. World Medical Association. World Medical Association Declaration of Helsinki: Ethical Principles for Medical Research Involving Human Participants. JAMA 2025, 333, 71–74. [Google Scholar] [CrossRef]
  42. Ustad, A.; Logacjov, A.; Trollebø, S.Ø.; Thingstad, P.; Vereijken, B.; Bach, K.; Maroni, N.S. Validation of an Activity Type Recognition Model Classifying Daily Physical Behavior in Older Adults: The HAR70+ Model. Sensors 2023, 23, 2368. [Google Scholar] [CrossRef] [PubMed]
  43. Kuster, R.P.; Baumgartner, D.; Hagströmer, M.; Grooten, W.J.A. Where to Place Which Sensor to Measure Sedentary Behavior? A Method Development and Comparison Among Various Sensor Placements and Signal Types. J. Meas. Phys. Behav. 2020, 3, 274–284. [Google Scholar] [CrossRef]
  44. Xu, Y.; Li, G.; Li, Z.; Yu, H.; Cui, J.; Wang, J.; Chen, Y. Smartphone-Based Unconstrained Step Detection Fusing a Variable Sliding Window and an Adaptive Threshold. Remote Sens. 2022, 14, 2926. [Google Scholar] [CrossRef]
  45. Kobsar, D.; Osis, S.T.; Phinyomark, A.; Boyd, J.E.; Ferber, R. Reliability of gait analysis using wearable sensors in patients with knee osteoarthritis. J. Biomech. 2016, 49, 3977–3982. [Google Scholar] [CrossRef]
  46. Adamowicz, L.; Christakis, Y.; Czech, M.D.; Adamusiak, T. SciKit Digital Health: Python Package for Streamlined Wearable Inertial Sensor Data Processing. JMIR mHealth uHealth 2022, 10, e36762. [Google Scholar] [CrossRef]
  47. Kuster, R.P.; Huber, M.; Hirschi, S.; Siegl, W.; Baumgartner, D.; Hagströmer, M.; Grooten, W. Measuring Sedentary Behavior by Means of Muscular Activity and Accelerometry. Sensors 2018, 18, 4010. [Google Scholar] [CrossRef]
  48. Ducharme, S.W.; Lim, J.; Busa, M.A.; Aguiar, E.J.; Moore, C.C.; Schuna, J.M.; Barreira, T.V.; Staudenmayer, J.; Chipkin, S.R.; Tudor-Locke, C. A Transparent Method for Step Detection using an Acceleration Threshold. J. Meas. Phys. Behav. 2021, 4, 311–320. [Google Scholar] [CrossRef]
  49. Brajdic, A.; Harle, R. Walk detection and Step Counting on Unconstrained Smartphones. In Proceedings of the 2013 ACM International Joint Conference on Pervasive and Ubiquitous Computing, Zurich, Switzerland, 8–12 September 2013; Association for Computing Machinery: New York, NY, USA, 2013; pp. 225–234. [Google Scholar] [CrossRef]
  50. Small, S.R.; Chan, S.; Walmsley, R.; von Fritsch, L.; Acquah, A.; Mertes, G.; Feakins, B.G.; Creagh, A.; Strange, A.; Matthews, C.E.; et al. Development and Validation of a Machine Learning Wrist-worn Step Detection Algorithm with Deployment in the UK Biobank. medRxiv 2023, medRxiv:23285750. [Google Scholar] [CrossRef]
  51. Kim, K.; Kim, Y.M.; Kim, E.K. Correlation between the Activities of Daily Living of Stroke Patients in a Community Setting and Their Quality of Life. J. Phys. Ther. Sci. 2014, 26, 417–419. [Google Scholar] [CrossRef]
  52. Ramos-Lima, M.J.M.; Brasileiro Ide, C.; de Lima, T.L.; Braga-Neto, P. Quality of life after stroke: Impact of clinical and sociodemographic factors. Clinics 2018, 73, e418. [Google Scholar] [CrossRef] [PubMed]
  53. Henderson, C.E.; Toth, L.; Kaplan, A.; Hornby, T.G. Step Monitor Accuracy During PostStroke Physical Therapy and Simulated Activities. Transl. J. Am. Coll. Sports Med. 2022, 7, e000186. [Google Scholar] [CrossRef]
  54. Compagnat, M.; Batcho, C.S.; David, R.; Vuillerme, N.; Salle, J.Y.; Daviet, J.C.; Mandigout, S. Validity of the Walked Distance Estimated by Wearable Devices in Stroke Individuals. Sensors 2019, 19, 2497. [Google Scholar] [CrossRef]
  55. Zhang, W.; Smuck, M.; Legault, C.; Ith, M.A.; Muaremi, A.; Aminian, K. Gait Symmetry Assessment with a Low Back 3D Accelerometer in Post-Stroke Patients. Sensors 2018, 18, 3322. [Google Scholar] [CrossRef]
  56. Kuster, R.P. Advancing the Measurement of Sedentary Behaviour: Classifying Posture and Physical (in-)Activity. Ph.D. Thesis, Karolinska Institutet, Stockholm, Sweden, 2021; p. 80. [Google Scholar]
  57. Andrews, K.; Stewart, J. Stroke Recovery: He can but does he? Rheumatol. Rehabil. 1979, 18, 43–48. [Google Scholar] [CrossRef]
  58. Kuster, R.P.; Grooten, W.J.A.; Baumgartner, D.; Blom, V.; Hagströmer, M.; Ekblom, Ö. Detecting prolonged sitting bouts with the ActiGraph GT3X. Scand. J. Med. Sci. Sports 2020, 30, 572–582. [Google Scholar] [CrossRef]
  59. Xu, D.; Zhou, H.; Quan, W.; Jiang, X.; Liang, M.; Li, S.; Ugbolue, U.; Baker, S.; Gusztav, F.; Ma, X.; et al. A new method proposed for realizing human gait pattern recognition: Inspirations for the application of sports and clinical gait analysis. Gait Posture 2024, 107, 293–305. [Google Scholar] [CrossRef]
  60. Schmid, A.; Duncan, P.W.; Studenski, S.; Lai, S.M.; Richards, L.; Perera, S.; Wu, S.S. Improvements in Speed-Based Gait Classifications Are Meaningful. Stroke 2007, 38, 2096–2100. [Google Scholar] [CrossRef]
  61. Neumann, S.; Bauer, C.M.; Nastasi, L.; Läderach, J.; Thürlimann, E.; Schwarz, A.; Held, J.P.O.; Easthope, C.A. Accuracy, concurrent validity, and test–retest reliability of pressure-based insoles for gait measurement in chronic stroke patients. Front. Digit. Health 2024, 6, 1359771. [Google Scholar] [CrossRef]
  62. Werner, C.; Awai Easthope, C.; Curt, A.; Demkó, L. Towards a Mobile Gait Analysis for Patients with a Spinal Cord Injury: A Robust Algorithm Validated for Slow Walking Speeds. Sensors 2021, 21, 7381. [Google Scholar] [CrossRef]
  63. Clay, L.; Webb, M.; Hargest, C.; Adhia, D.B. Gait quality and velocity influences activity tracker accuracy in individuals post-stroke. Top. Stroke Rehabil. 2019, 26, 412–417. [Google Scholar] [CrossRef]
  64. Taraldsen, K.; Askim, T.; Sletvold, O.; Einarsen, E.K.; Bjåstad, K.G.; Indredavik, B.; Helbostad, J.L. Evaluation of a Body-Worn Sensor System to Measure Physical Activity in Older People With Impaired Function. Phys. Ther. 2011, 91, 277–285. [Google Scholar] [CrossRef] [PubMed]
Sensors 25 05657 g001
Figure 1. Mean absolute percentage error (MAPE) of the investigated algorithms in n = 7 participants: Gradient Boosting (GB, waist sensor), Continuous Wavelet Transform (CWT, waist sensor), SciKit Digital Health (SKDH, lower back sensor), Threshold-Crossing Algorithm (TCA, waist sensor), and OxWearables (OxW, wrist sensor). Only the GB algorithm was developed in this study using data from individuals with neurological conditions; the remaining algorithms were developed in other studies using data from able-bodied individuals.
Figure 1. Mean absolute percentage error (MAPE) of the investigated algorithms in n = 7 participants: Gradient Boosting (GB, waist sensor), Continuous Wavelet Transform (CWT, waist sensor), SciKit Digital Health (SKDH, lower back sensor), Threshold-Crossing Algorithm (TCA, waist sensor), and OxWearables (OxW, wrist sensor). Only the GB algorithm was developed in this study using data from individuals with neurological conditions; the remaining algorithms were developed in other studies using data from able-bodied individuals.
Sensors 25 05657 g001
Table 1. Results of the generalized linear model assessing the effects of sensor location, filter range, window size, and regressor type on the root mean squared error (RMSE) of step counting. Significant differences compared to the reference are indicated with an asterisk (*). Effect sizes and their 95% confidence intervals (95% CI) are reported as exp (β) and 95% CI (exp (β)), respectively, with exp (β) representing the multiplicative change in RMSE relative to the reference. Data for each level is shown relative to their parameter reference, with all other parameters held at their reference.
Table 1. Results of the generalized linear model assessing the effects of sensor location, filter range, window size, and regressor type on the root mean squared error (RMSE) of step counting. Significant differences compared to the reference are indicated with an asterisk (*). Effect sizes and their 95% confidence intervals (95% CI) are reported as exp (β) and 95% CI (exp (β)), respectively, with exp (β) representing the multiplicative change in RMSE relative to the reference. Data for each level is shown relative to their parameter reference, with all other parameters held at their reference.
ParameterLevelEffect Size95% CIp-ValueRMSE
Reference (Ref) 0.30
Location
(Ref = Waist)		Lower Back *1.100[1.002, 1.206]0.0440.33
Ankle *1.101[1.004, 1.208]0.0410.33
Thigh *1.195[1.089, 1.310]<0.0010.36
Chest *1.276[1.164, 1.400]<0.0010.39
Wrist (l) *1.848[1.685, 2.027]<0.0010.56
Wrist (r) *2.002[1.825, 2.196]<0.0010.61
Filter
(Ref = wide band)		narrow band *0.902[0.849, 0.958]0.0010.27
medium band *0.911[0.857, 0.968]0.0030.28
Window
(Ref = medium)		long *0.772[0.734, 0.811]0.0000.23
Regressor
(Ref = GB)		RF0.989[0.914, 1.069]0.7760.30
MLP1.063[0.983, 1.150]0.1250.32
SVR *1.089[1.007, 1.178]0.0320.33
kNN *1.128[1.043, 1.219]0.0030.34
Abbreviations: left (l), right (r), Gradient Boosting (GB), Random Forest (RF), Multilayer Perceptron (MLP), Support Vector Regression (SVR), k-Nearest Neighbors (kNN). Note that the short window condition was excluded due to modelling constraints (see Supplementary Material, Figure S1).
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Crozat, F.; Pohl, J.; Easthope Awai, C.; Bauer, C.M.; Kuster, R.P. Every Step Counts—How Can We Accurately Count Steps with Wearable Sensors During Activities of Daily Living in Individuals with Neurological Conditions? Sensors 2025, 25, 5657. https://doi.org/10.3390/s25185657

AMA Style

Crozat F, Pohl J, Easthope Awai C, Bauer CM, Kuster RP. Every Step Counts—How Can We Accurately Count Steps with Wearable Sensors During Activities of Daily Living in Individuals with Neurological Conditions? Sensors. 2025; 25(18):5657. https://doi.org/10.3390/s25185657

Chicago/Turabian Style

Crozat, Florence, Johannes Pohl, Chris Easthope Awai, Christoph Michael Bauer, and Roman Peter Kuster. 2025. "Every Step Counts—How Can We Accurately Count Steps with Wearable Sensors During Activities of Daily Living in Individuals with Neurological Conditions?" Sensors 25, no. 18: 5657. https://doi.org/10.3390/s25185657

APA Style

Crozat, F., Pohl, J., Easthope Awai, C., Bauer, C. M., & Kuster, R. P. (2025). Every Step Counts—How Can We Accurately Count Steps with Wearable Sensors During Activities of Daily Living in Individuals with Neurological Conditions? Sensors, 25(18), 5657. https://doi.org/10.3390/s25185657

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop