Performance Measurements from split Timed Up and Down Stairs test in Cerebral Palsy with and without Ankle Foot Orthoses

@article{PhysioNet-cerebral-palsy-timed-stairs-1.0.0,
  author = {Hopkins, Samuel and Gaudette, Connor and {Silveti Rodriguez}, Maria Jose and Lerner, Zach},
  title = {{Performance Measurements from split Timed Up and Down Stairs test in Cerebral Palsy with and without Ankle Foot Orthoses}},
  journal = {{PhysioNet}},
  year = {2026},
  month = nov,
  note = {Version 1.0.0},
  doi = {10.13026/fy65-t837},
  url = {https://doi.org/10.13026/fy65-t837}
}

@article{pollard_physionet_2026, title = {{PhysioNet} as a global platform for biomedical research}, volume = {1}, issn = {3005-0693}, url = {https://doi.org/10.1038/s44360-026-00096-z}, doi = {10.1038/s44360-026-00096-z}, abstract = {Twenty-five years after its first public repository was established, the PhysioNet resource for biomedical and clinical data has enabled tens of thousands of research papers. Further efforts are ongoing to advance secure data sharing, community engagement and scalable research infrastructure.}, number = {8}, journal = {Nature Health}, author = {Pollard, Tom and Moody, Benjamin E. and Lehman, Li-wei H. and Gow, Brian J. and Fernandes, Chrystinne and Xie, Chen and Johnson, Alistair and Mark, Roger G. and Heldt, Thomas}, month = aug, year = {2026}, pages = {792--795},}

Abstract

The dataset presented was collected during split Timed Up and Down Stairs tests of a novel, robotic ankle foot orthosis (AFO) with adjustable stiffness and manually variable neutral angle for seven individuals with spastic cerebral palsy (CP). Tests were conducted between May 2025 – October 2025 at Northern Arizona University. Test conditions included nine combinations of stiffness and neutral angle for the robotic AFO, a solid AFO, and shoes alone. Outcomes included TUDs time, peak ankle dorsiflexion and plantarflexion, knee range of motion, soleus and vastus lateralis integrated electromyography, perceived exertion, balance confidence, 3-axis trunk acceleration, and robotic AFO power and torque. These outcomes, conditions, and participant characteristics are reported in this dataset. These data may contribute to 1) further development of robotic ankle foot orthoses and 2) understanding stair ambulation biomechanics in CP.

Background

Individuals with CP walk five times slower than their peers [@wright2020]. Ankle Foot Orthoses (AFOs) are commonly prescribed to improve level walking ability [@bayon2023], but individuals express a desire for adaptability and improved stair ambulation from their AFOs [@sienkothomas2002]. Current evidence indicate that AFOs may not be a hinderance to stair mobility, but conflict regarding their benefit [@ozturk2015, @hopkins2026]. Surveys indicate the widespread desire for adaptability as most individuals who wear AFOs describe adjustable stiffness (72.2%) and range of motion (86.1%) as important or highly important considerations for AFO design [@sienkothomas2002]. Improved stair performance was a concern for 12.6% of these survey respondents. Robotic AFOs offer the ability to adapt stiffness and neutral angle, which the typical solid AFOs cannot do, but require continued testing in new contexts where adjustability may be relevant like stair ascent and descent.

The purpose of this dataset was to:

1. Quantify how stiffness and neutral angle of a robotic AFO affect clinical, biomechanical, and user experience outcomes for individuals with cerebral palsy
2. Compare stair ambulation while wearing shoes alone, a solid AFO, and the robotic AFO

This resource provides all the data for the 2026 report in the Annals of Biomedical Engineering titled “Exploratory Assessment of Robotic Ankle Foot Orthosis Stiffness and Neutral Angle on Stair Performance in Individuals with Cerebral Palsy” [@barhaim2004]. These data were recorded at Northern Arizona University from seven male individuals with spastic CP throughout May 2025 to October 2025. Participants ascended and descended a 12-step flight of stairs (tread depth: 30.5 cm, riser height: 17.8 cm) while being timed, in the various robotic AFO configurations, with solid AFOs, and with shoes alone.

These data are being provided to increase transparency of the original publication, and further development of the evolving field of robotics ankle foot orthoses and other exoskeletons. It can be used to benchmark other robotic AFOs or wearable interventions against the tested device, standard-of-care rigid AFOs, and shoes alone conditions in the context of stair climbing with CP. Other robotic AFOs devices, including the inGAIT-VSO [@noortlvan2024], ADJUST [@miedema2023], and quasi-passive and electromechanical AFOs [@kumar2020, @diller2016] would benefit from these data as they provide a better understanding of how AFO stiffness and neutral angle may impact ambulatory activities. Findings of the attached study could enable further development of novel mechanisms and control strategies for robotic AFOs and other exoskeletons that adapt across activities, like stair climbing. Finally, these data can provide information on functional mobility, muscle activity, balance (trunk instability and confidence), exertion, and device function (peak torque and power). These are important factors for assessing stair ambulation ability in CP. This dataset may contribute to a better biomechanical understanding of the challenges individuals with CP have when ascending and descending stairs.

Methods

Participants

The Institutional Review Board of Northern Arizona University approved this study (protocol #2137266). This study was performed in line with the principles of the Declaration of Helsinki. Clinical Trial Number: not applicable. The inclusion criteria for this study were age 10 years old or older, spastic cerebral palsy, the ability to follow written and verbal instructions, walk for at least 6 minutes and up and down at least ten flights of stairs. Exclusion criteria included any lower limb surgery or botulinum toxin injections within the last 6 months. Participants under the age of 18 provided verbal assent, while their parents and legal guardian provided written consent. Participants older than 18 provided written consent. Seven male participants (average age 21.6 years) were recruited (Table 1).

Orthotic Conditions

• Bilateral, Robotic AFOs
  • Nine combinations of three dorsiflexion stiffnesses (low: 120 Nm/rad, medium: 235 Nm/rad, high: 350 Nm/rad) and three neutral angles (early: -5° plantarflexion, neutral: 0°, and late: +5° dorsiflexion) in a full factorial design.
  • The baseline 0° angle represents a neutral foot position, as in 90° relative to the shank.
  • Hinged angle joint
• Bilateral, Solid AFOs
  • Rigid ankle joint, with the foot set at 0°(shank 90° relative to foot)
• Shoes Alone
  • Nike Flex Train

Acclimation

During an initial acclimation period, participants walked for a total of 360 steps with the robotic AFO followed by 15 minutes of rest. Participants walked on level ground for four sets of 30 steps. Midway through each set, the spring stiffness was changed from low to high or high to low. Participants then ascended 120 steps and descended 120, about ten flights of stairs, for acclimation. Midway through each flight of stairs, the stiffness was changed between low and high. The robotic AFO was provided with an extended acclimation to accustom participants to the variable configurations; since the solid AFO was not adjustable, it was not assigned an extended acclimation.

Timed Up and Down Stairs Tests

Participants performed a modified Timed Up and Down Stairs (TUDS) test. The original TUDS test has been shown to have high interrater, intrareader, and test-rest reliability (interclass correlation coefficients >0.94) in children with CP [@zaino2004, @ayed2022] and shows concurrent validity with other types of mobility tests [@zaino2004]. The test measured the time required to ascend and descend a 12-step flight of stairs (tread depth: 30.5 cm, riser height: 17.8 cm). Participants were allowed to use the railing on either side of the stairs as needed, but only one participant (P5) used it. We modified the protocol in two ways. First, ascent and descent were analyzed separately to isolate the effects of stiffness and neutral angle for each ambulation mode. Second, participants performed a familiarization flight before the measured flight. The first flight for either ascent or descent was used for familiarization with each experimental condition, before participants continued in the same direction, ascent or descent, during the second flight where the timed trial occurred. During the timed trials, participants stood 30 cm away from the first step and were instructed to ascend and descend the stairs as quickly as possible until both feet contacted the top or bottom landing. A wireless laser timer system (YZ Laser Timer), placed 30 cm from the first and final step, measured the time to the nearest thousandths of a second. Participants were allowed to choose any strategy they wished, though most preferred a step-over-step method. P5 used a step-to-step method, and as such, their biomechanical data was limited to include every second step to keep sample counts similar between participants.

A total of nine trials with the robotic AFO were performed, each one corresponding to a different configuration of stiffness and neutral angle. Shoes alone and the rigid AFOs were tested once each.

The shoes alone condition was always performed first, while the AFO conditions were applied next. Four participants started with the bilateral solid AFOs while three started with the bilateral robotic AFOs. Within the robotic AFO condition, neutral angle and stiffness formed a nested block pattern. Neutral angle was presented as the supergroup, with early, neutral, and late conditions each forming a separate block of trials. Stiffness, the subgroup, was presented in ascending or descending order (high-medium-low) within each block of neutral angle conditions. For example, P1 had a trial order as follows: (1) early neutral angle with high-medium-low stiffness levels; (2) neutral 0° neutral angle with high-medium-low stiffness levels; and (3) late neutral angle with high-medium-low stiffness levels. The order was reversed for every other participant, such that P2 had a trial order as follows: (1) late neutral angle with low-medium-high stiffness levels; (2) neutral 0° neutral angle with low-medium-high stiffness levels; (3) early neutral angle with low-medium-high stiffness levels. Fatigue and carry-over effects may have impacted the data; however, the use of block randomization, reverse ordering, and frequent rest mitigates these impacts.

Sensor Placement for Biomechanical Measurements

We measured muscle activity, ankle and knee kinematics, and gait instability via trunk accelerations. Four Trigno Avanti sensors (Delsys) were used as both electromyography (EMG) sensors and inertial measurement units (IMUs), each equipped with 3-axis gyroscope, 3-axis accelerometer, and 3-axis magnetometer sensors. Participants identified their more impaired leg prior to Avanti sensor placement. Two sensors were placed over the soleus and the vastus lateralis to measure plantar flexor and quadriceps activity, which play a dominant role in stair ambulation [@mcfadyen1988, @zelik2016]. The soleus Avanti sensor measured muscle activity, while peak vertical acceleration of this segment was used to identify heel strikes. The vastus lateralis sensor measured muscle activity and, using proprietary sensor-fusion algorithms [@trigno2023], estimated thigh orientation.

Acceleration and orientation modes on these devices are mutually exclusive. Because the soleus IMU was configured in acceleration mode, a third Avanti sensor was placed just above the medial malleolus to measure shank orientation. Knowledge of both the shank and thigh orientations enabled measurement of the knee angle defined by the relative angle of the two sensors in the sagittal plane. This method was previously validated against traditional optical motion tracking [@bergmann2009]. An initial calibration was performed to account for misalignment between the IMU sensors. Participants were instructed to sit still, with both feet flat on the ground, while a handheld goniometer, placed over the lateral epicondyle of the femur, was aligned with the shank and thigh to measure the knee angle. Participants wore the same standard shoes during calibration. Simultaneously, the Delsys acquisition software recorded 30 seconds of data from which the offset between the IMU-derived and manual measurement was calculated. Knee kinematics can determine foot clearance when ascending and descending stairs [@lewerenz2019], and thus are a relevant factor for assessing stair ambulation.

The fourth Avanti sensor was placed over the L3 vertebra to measure 3-axis trunk accelerations, which have been previously used to measure gait instability in children with CP [@valenciano2022, @iosa2012]. Greater focus placed on maintaining control during stair ambulation may affect stair performance and thus may be a cause for lower speed [@lewerenz2019]. Thus, changes in instability may affect ambulatory speed. Next, an AS5600 magnetic encoder at the ankle within the robotic AFO measured ankle angle. Finally, we used a custom, validated torque sensor to measure the peak torque supplied by the device [@hopkins2024]. The mechanical power developed by the device was calculated as the product of ankle angular velocity (the time derivative of ankle angle) and measured ankle torque.

Clinical Outcomes

• Timed Up and Down Stairs Time
  • Stair ascent and descent measured separately
  • YZ Laser Time, Amazon. 0.001 sec resolution

Biomechanical Outcomes

• Ankle Kinematics
  • Sagittal Plane only
  • Recorded at 50 Hz
  • AS5600, Adafruit
  • Measured for the robotic AFO only
• Knee Kinematics
  • Recorded at 74 Hz, resampled to 100 Hz
  • Knee angle, measured by the difference in sagittal plane angle between an Inertial measurement Unit (Trigno Avanti Sensors, Delsys) on the shank and on the thigh
• Muscle activity
  • We generated a linear envelope by band-pass filtering the data using a fourth order Butterworth filter with cutoff frequencies 15--380 Hz, full wave rectification, followed by low-pass filtering the data with a fourth order, 7Hz cutoff frequency Butterworth filter [@mcfadyen1988].
  • Soleus Activity
    • Recorded at 1926 Hz, resampled to 100 Hz
    • Trigno Avanti Sensors, Delsys
  • Vastus Lateralis Activity
    • Recorded at 1778 Hz, resampled to 100 Hz
    • Trigno Avanti Sensors, Delsys
    • All Avanti sensors used to measure muscle activity were placed on the more impaired limb using SENIAM Recommendations [@zelik2016].
• Trunk Instability
  • Root mean square of acceleration in anteroposterior, mediolateral, and vertical directions [@trigno2023]
  • Recorded at 370 Hz, filtered at 20 Hz
  • Trigno Avanti Sensors, Delsys, placed on the L3 vertebrae
• Robotic AFO Output
  • Torque
    • Recorded at 50 Hz
    • Custom torque sensor [@bergmann2009]
  • Power
    • Calculated as the time differential of the ankle angle times the device torque
    • Measured for robotic AFO only
• User Experience Outcomes
  • OMNI Scale of Perceived Exertion Administered after each trial [@lewerenz2019, @valenciano2022]
  • Activities-specific Balance Confidence scale [@iosa2012, @hopkins2024] question on perceived risk of failing during stair ascent or descent administered after every trial

Data Description

• RegressionMasterDataset.csv
  • Participant
    • Participant ID
  • Age
    • Participant Age (yrs)
  • Weight
    • Participant Weight (kg)
  • Height
    • Participant Height (cm)
  • GMFCS
    • Gross Motor Function Classification Scale (I,II)
  • Stiffness
    • (Nm/rad)
    • 120 Nm/rad, 235 Nm/rad, 350 Nm/rad
    • -10000000 was used as a null value.
    • Rows 2,3, 24,25, 46, 47, 68, 69, 90, 91, 112,113, 134,135, correspond to Shoes Alone
    • Rows 4,5, 26,27, 48, 49, 70, 71, 92, 93, 114,115, 136,137, correspond to Solid AFOs
  • NeutralAngle
    • (°)
    • -5°, 0°, 5°
    • -10000000 was used as a null value.
    • Rows 2,3, 24,25, 46, 47, 68, 69, 90, 91, 112,113, 134,135, correspond to Shoes Alone
    • Rows 4,5, 26,27, 48, 49, 70, 71, 92, 93, 114,115, 136,137, correspond to Solid AFOs
  • Direction
    • “Ascend” for stair ascent trials
    • “Descend” for stair descent trials
  • AnkleROM
    • (°)
    • Ankle range of motion = Peak dorsiflexion-peak plantarflexion
    • Averaged across all gait cycles
  • AnkleExcursion
    • (°)
    • Difference between peak dorsiflexion and the ankle angle measured at the first instance of a 1Nm load on the torque sensor during the gait cycle
    • Averaged across all gait cycles
  • PeakDorsi
    • Peak Dorsiflexion (°)
    • Maximum angle during a step
    • Averaged across all gait cycles
  • PeakPlantar
    • Peak plantarflexion (°)
    • Minimum angle during a step
    • Averaged across all gait cycles
  • PeakTorque
    • (Nm)
    • Maximum torque measured during a step
    • Averaged across all gait cycles
  • PeakPower
    • (W)
    • Power = torque * time derivative of ankle angle
    • Averaged across all gait cycles
  • KneeROM
    • Knee Range of Motion (°)
    • Knee angle = maximum knee angle – minimum knee angle
    • Averaged across all gait cycles
  • Anterioposterior
    • (g)
    • Root mean square of trunk acceleration in the anteroposterior direction during a trial
  • Mediolateral
    • (g)
    • Root mean square of trunk acceleration in the mediolateral direction during a trial
  • Vertical
    • (g)
    • Root mean square of trunk acceleration in the vertical direction during a trial
  • Time
    • (seconds)
    • Timed Up And Down Stairs test time
  • Exertion
    • Perceived Exertion
    • 1 represents resting
    • 10 represents maximal effort activities
  • Confidence
    • Balance Confidence rating 0-100
    • 0 indicates “no confidence”, the participant feels likely to fall
    • 100 indicates “complete confidence”, the individual feels unlikely to fall
  • SoleusiEMG
    • (no unit)
    • Time normalized, integrated electromyography, then normalized to shoes alone
    • Averaged across all gait cycles
    • Data recorded from the soleus
  • VastusiEMG
    • (no unit)
    • Time normalized, integrated electromyography (iEMG), then normalized to shoes alone (iEMG)
    • Averaged across all gait cycles
    • Data recorded from the vastus lateralis
  • SoleusiEMG_raw
    • (mV *s)
    • Time normalized, integrated electromyography, no normalization to shoes
    • Averaged across all gait cycles
    • Data recorded from the soleus
  • VastusiEMG_raw
    • (mV *s)
    • Time normalized, integrated electromyography (iEMG), no normalization to shoes
    • Averaged across all gait cycles
    • Data recorded from the vastus lateralis
  • Device Condition
    • Describes the device condition for that trial
    • Shoes Alone, Rigid AFO, Clutched AFO
      • Configuration for clutched AFO is described in columns for Stiffness and NeutralAngle
• MixedEffectsModeling_Simplified.m
  • Runs mixed effects modeling on the outcomes
  • Uses the following model for each outcoe described above
  • Outcome = S+NA+ S ×NA+(1 | P) (1)
  • where the model included fixed effect terms for stiffness (S), neutral angle (NA), and their interaction. The model included a random intercept for participant (P) to account for repeated measures within-participants.
• PlotRelationships.m
  • Plots a simple line plot for each participant with the independent variable being the outcome of choice, while the dependent variable is either stiffness or neutral angle
• AnovaAnalysisHolm.m
  • Runs repeated-measures ANOVA with Holm-Bonferroni corrections (function in bonf_holm.m), and then plots the significant comparisons
• bonf_holm1.m
  • Bonferroni-Holm (1979) correction for multiple comparisons. This is a sequentially rejective version of the simple Bonferroni correction for multiple comparisons and strongly controls the family-wise error rate at level alpha [@groppe2026].
  • Created by David Groppe, accessed online [@groppe2026]

Usage Notes

Wearable robotics is a growing field of interest, including the emergence of novel robotic foot ankle orthoses. The device tested here is one of several newly developed ankle foot orthoses capable of adjusting stiffness in real time, comparable to the inGAIT-VSO [@fang2022], ADJUST [@miedema2023], and others [@kumar2020, @diller2016, @seniam2024, @orekhov2021]. Past research into AFO development has focused on level walking [@robertson2000], while these data support the broader research community by assessing how AFO stiffness and neutral angle impact stair ambulation. This benefit may enable development of improved designs and control strategies for robotic AFOs that adapt across different activities, like level walking or stair ambulation, to better fit individual needs, which fixed stiffness AFOs, the current standard-of-care, cannot do. Having an AFO with adjustable stiffness is an important desire for 72.2% of AFO users [@robertson2005], and this work contributes to addressing that need by furthering our understanding of interactions between the human body and adjustable robotic AFOs. A novel part of the dataset is the inclusion of neutral angles as a variable, since AFO design has predominantly been evaluated with stiffness as the main focus of research [@robertson2000]. Usage of both neutral angle and stiffness provides the ability to assess interaction effects, which have so far been relegated to simulation studies [@powell1995]. As such, this dataset can provide valuable information regarding how to modulate both AFO stiffness and neutral angle to improve stair ascent and descent in CP. Additionally, this dataset provides biomechanical data in CP for split Timed Up and Down Stairs (TUDS) mobility tests with shoes alone and solid AFOs. As such, it may be valuable to the clinical community and future studies for the purpose of developing better mobility interventions.

This dataset is limited in its small, all male sample size. Most participants had high functional ability as five people were in the Gross Motor Function Classification Scale Level I. Most participants did not typically use AFOs, though P5 used crutches routinely. The balance confidence used here as an isolated, absolute metric has not been validated when separated from the Activities-specific Balance Confidence Scale. The ankle angle and torque data from the robotic AFO for P1 were unavailable due to a device disconnection following the testing. The same issue occurred with the 5° neutral angle conditions for P6. All other data, such as iEMG, for these participants were stored by external systems. P3’s low stiffness, 0° neutral angle descent trial was excluded due to a loss of balance during descent; no injury occurred.

The manuscript for this study has been published in the Annals of Biomedical Engineering [@hopkins2026].

Release Notes

Version 1.0.0

Initial release

Ethics

This work offers the potential to improve AFO design for individuals with cerebral palsy. Robotic AFOs may provide the adaptability individuals desire from their AFOs. The data indicated here suggests that high stiffness, more dorsiflexed settings may help individuals ambulate faster while descending stairs after mixed effects modeling. The major risks for this study are slips, trips, falls, and discomfort due to rubbing of the AFO padding. Handrail use was allowed as needed while a researcher walked next to the participant to mitigate risk of falling. Rubbing was addressed by adding additional foam to the device or athletic tape around the point-of-contact on the leg as needed.

The Institutional Review Board of Northern Arizona University approved this study (protocol #2137266). This study was performed in line with the principles of the Declaration of Helsinki. Clinical Trial Number: not applicable. The inclusion criteria for this study were age 10 years old or older, spastic cerebral palsy, the ability to follow written and verbal instructions, walk for at least 6 minutes and up and down at least ten flights of stairs. Exclusion criteria included any lower limb surgery or botulinum toxin injections within the last 6 months. Participants under the age of 18 provided verbal assent, while their parents and legal guardian provided written consent. Participants older than 18 provided written consent.

Acknowledgements

This work was supported in part by the Flinn Foundation. Research reported in this publication was partially supported by the Eunice Kennedy Shriver National Institute Of Child Health & Human Development of the National Institutes of Health under Award Number F31HD119948. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Conflicts of Interest

Z. F. Lerner is a co-founder with shareholder interest in Biomotum, Inc., a university start-up company, seeking to commercialize the device used in this study. He also has intellectual property inventorship rights covering aspects of the design and control of the exoskeleton device used in this study.

References

1. @wright2020 Wright E, DiBello SA. Principles of Ankle-Foot Orthosis Prescription in Ambulatory Bilateral Cerebral Palsy. Phys Med Rehabil Clin N Am. 2020;31:69–89. https://doi.org/10.1016/j.pmr.2019.09.007
2. @bayon2023 Bayón C, Hoorn M van, Barrientos A, Rocon E, Trost JP, Asseldonk EHF van. Perspectives on ankle-foot technology for improving gait performance of children with Cerebral Palsy in daily-life: requirements, needs and wishes. J NeuroEngineering Rehabil. 2023;20:44. https://doi.org/10.1186/s12984-023-01162-3
3. @sienkothomas2002 Sienko Thomas S, Buckon CE, Jakobson-Huston S, Sussman MD, Aiona MD. Stair locomotion in children with spastic hemiplegia: the impact of three different ankle foot orthosis (AFOs) configurations. Gait Posture. 2002;16:180–7. https://doi.org/10.1016/S0966-6362(02)00002-4
4. @ozturk2015 Öztürk B, Uygur F, Yakut Y. Effect of ankle-foot orthoses on functional performance and physiological cost index in children with cerebral palsy. J Exerc Ther Rehabil [Internet]. 2015;2. https://izlik.org/JA22JL64BZ
5. @hopkins2026 Hopkins S, Gaudette C, Silvetti MJ, Lerner ZF. Exploratory Assessment of Robotic Ankle Foot Orthosis Stiffness and Neutral Angle on Stair Ambulation in Individuals with Cerebral Palsy. Ann Biomed Eng [Internet]. 2026 [cited 2026 Sept 16]; https://doi.org/10.1007/s10439-026-04305-4
6. @barhaim2004 Bar-Haim S, Belokopytov M, Harries N, Frank A. A stair-climbing test for ambulatory assessment of children with cerebral palsy. Gait Posture. 2004;20:183–8. https://doi.org/10.1016/j.gaitpost.2003.09.002
7. @noortlvan2024 Noort L van, Crey NV, Rouse EJ, Martínez-Caballero I, Asseldonk EHF van, Bayón C. A usability study on the inGAIT-VSO: effects of a variable-stiffness ankle-foot orthosis on the walking performance of children with cerebral palsy [Internet]. 2024 [cited 2024 May 23]. https://doi.org/10.21203/rs.3.rs-4350951/v1
8. @miedema2023 Miedema R, Waterval N, Meijneke C, Harlaar J, Nollet F, Brehm M. ADJUST: A stiffness adjustable ankle-foot-orthosis for rapid human-in-the-loop orthosis selection. Gait Posture. 2023;106:S126–7. https://doi.org/10.1016/j.gaitpost.2023.07.156
9. @kumar2020 Kumar S, Zwall MR, Bolívar-Nieto EA, Gregg RD, Gans N. Extremum Seeking Control for Stiffness Auto-Tuning of a Quasi-Passive Ankle Exoskeleton. IEEE Robot Autom Lett. 2020;5:4604–11. https://doi.org/10.1109/LRA.2020.3001541
10. @diller2016 Diller S, Majidi C, Collins SH. A lightweight, low-power electroadhesive clutch and spring for exoskeleton actuation. 2016 IEEE Int Conf Robot Autom ICRA [Internet]. 2016 [cited 2025 Nov 24]. p. 682–9. https://doi.org/10.1109/ICRA.2016.7487194
11. @zaino2004 Zaino CA, Marchese VG, Westcott SL. Timed up and down stairs test: preliminary reliability and validity of a new measure of functional mobility. Pediatr Phys Ther Off Publ Sect Pediatr Am Phys Ther Assoc. 2004;16:90–8. https://doi.org/10.1097/01.PEP.0000127564.08922.6A
12. @ayed2022 Ayed AE, Hanna S, Abdelazeim FH. Reliability of the TUDS test for children with cerebral palsy. Bull Fac Phys Ther. 2022;27:46. https://doi.org/10.1186/s43161-022-00104-9
13. @mcfadyen1988 McFadyen BJ, Winter DA. An integrated biomechanical analysis of normal stair ascent and descent. J Biomech. 1988;21:733–44. https://doi.org/10.1016/0021-9290(88)90282-5
14. @zelik2016 Zelik KE, Adamczyk PG. A unified perspective on ankle push-off in human walking. J Exp Biol. 2016;219:3676–83. https://doi.org/10.1242/jeb.140376
15. @trigno2023 Trigno Wireless Biofeedback System [Internet]. 2023. https://delsys.com/downloads/USERSGUIDE/MAN-031-1-7%20Trigno%20Wireless%20Biofeedback%20System.pdf
16. @bergmann2009 Bergmann JH, Mayagoitia RE, Smith IC. A portable system for collecting anatomical joint angles during stair ascent: a comparison with an optical tracking device. Dyn Med. 2009;8:3. https://doi.org/10.1186/1476-5918-8-3
17. @lewerenz2019 Lewerenz A, Wolf SI, Dreher T, Krautwurst BK. Performance of stair negotiation in patients with cerebral palsy and stiff knee gait. Gait Posture. 2019;71:14–9. https://doi.org/10.1016/j.gaitpost.2019.04.005
18. @valenciano2022 Valenciano PJ, Conceição NR, Marcori AJ, Teixeira LA. Use of accelerometry to investigate standing and dynamic body balance in people with cerebral palsy: A systematic review. Gait Posture. 2022;96:357–64. https://doi.org/10.1016/j.gaitpost.2022.06.017
19. @iosa2012 Iosa M, Marro T, Paolucci S, Morelli D. Stability and harmony of gait in children with cerebral palsy. Res Dev Disabil. 2012;33:129–35. https://doi.org/10.1016/j.ridd.2011.08.031
20. @hopkins2024 Hopkins S, Bowersock C, Rouse EJ, Lerner ZF. A Quasi-Passive Robotic Ankle Foot Orthosis With Speed-Adaptive Stiffness. IEEE Robot Autom Lett. 2024;9:1740–7. https://doi.org/10.1109/LRA.2024.3349829
21. @groppe2026 Groppe D. Bonferroni-Holm Correction for Multiple Comparison [Internet]. 2026 [cited 2026 Oct 29]. https://www.mathworks.com/matlabcentral/fileexchange/28303-bonferroni-holm-correction-for-multiple-comparisons. Accessed 29 Oct 2026
22. @fang2022 Fang Y, Lerner ZF. How Ankle Exoskeleton Assistance Affects the Mechanics of Incline Walking and Stair Ascent in Cerebral Palsy. 2022 Int Conf Rehabil Robot ICORR [Internet]. 2022 [cited 2024 July 31]. p. 1–6. https://doi.org/10.1109/ICORR55369.2022.9896476
23. @seniam2024 Welcome to SENIAM [Internet]. [cited 2024 Nov 1]. http://www.seniam.org/. Accessed 1 Nov 2024
24. @orekhov2021 Orekhov G, Fang Y, Cuddeback CF, Lerner ZF. Usability and performance validation of an ultra-lightweight and versatile untethered robotic ankle exoskeleton. J NeuroEngineering Rehabil. 2021;18:163. https://doi.org/10.1186/s12984-021-00954-9
25. @robertson2000 Robertson RJ, Goss FL, Boer NF, Peoples JA, Foreman AJ, Dabayebeh IM, et al. Children’s OMNI Scale of Perceived Exertion: mixed gender and race validation. Med Sci Sports Exerc. 2000;32:452.
26. @robertson2005 Robertson RJ, Goss FL, Andreacci JL, Dube JJ, Rutkowski JJ, Snee BM, et al. Validation of the Children’s OMNI RPE Scale for Stepping Exercise: Med Sci Sports Exerc. 2005;37:290–8. https://doi.org/10.1249/01.MSS.0000149888.39928.9F
27. @powell1995 Powell LE, Myers AM. The Activities-specific Balance Confidence (ABC) Scale. J Gerontol Ser A. 1995;50A:M28–34. https://doi.org/10.1093/gerona/50A.1.M28
28. @opheim2012 Opheim A, Jahnsen R, Olsson E, Stanghelle JK. Balance in relation to walking deterioration in adults with spastic bilateral cerebral palsy. Phys Ther. 2012;92:279–88. https://doi.org/10.2522/ptj.20100432
29. @vancrey2023 Van Crey N, Cavallin M, Shepherd M, Rouse EJ. Design of a Quasi-Passive Ankle-Foot Orthosis with Customizable, Variable Stiffness. 2023 Int Conf Rehabil Robot ICORR [Internet]. 2023 [cited 2024 June 13]. p. 1–6. https://doi.org/10.1109/ICORR58425.2023.10304820
30. @rong2022 Rong H, Ramezani S, Ambro A, Lei CH, Choi H. Design and characterization of a variable-stiffness ankle-foot orthosis. Prosthet Orthot Int. 2022;10.1097/PXR.0000000000000323. https://doi.org/10.1097/PXR.0000000000000323
31. @kobayashi2011 Kobayashi T, Leung AKL, Akazawa Y, Hutchins SW. Design of a stiffness-adjustable ankle-foot orthosis and its effect on ankle joint kinematics in patients with stroke. Gait Posture. 2011;33:721–3. https://doi.org/10.1016/j.gaitpost.2011.02.005
32. @totah2019 Totah D, Menon M, Jones-Hershinow C, Barton K, Gates DH. The impact of ankle-foot orthosis stiffness on gait: A systematic literature review. Gait Posture. 2019;69:101–11. https://doi.org/10.1016/j.gaitpost.2019.01.020
33. @waterval2023 Waterval NFJ, Brehm MA, Veerkamp K, Geijtenbeek T, Harlaar J, Nollet F, et al. Interacting effects of AFO stiffness, neutral angle and footplate stiffness on gait in case of plantarflexor weakness: A predictive simulation study. J Biomech. 2023;157:111730. https://doi.org/10.1016/j.jbiomech.2023.111730