Intersegmental dynamics imbalance in gait impairments of children with spastic hemiplegia: insights from different torque contribution Authors: Zijun Lu, Qiong Wang, Xiaocong Chen, Yu Liu, Wenfei Zhu, Yuliang Sun Published: 2025 ABSTRACT AbstractBackgroundChallenges in the rehabilitation of gait impairments among children with spastic hemiplegia (CSH) suggest significant difficulties in controlling lower limb intersegmental dynamics (ISD), which may be closely associated with underlying proprioceptive and neuromuscular impairments. To quantify the ISD imbalance of lower limb in CSH, we developed a novel protocol utilizing motion capture and kinetic analysis to assess multi-joint torque contributions during the gait cycle.MethodsIn this prospective, hospital and school-based, case-control study, 21 typically developing children (age = 7.7 ± 1.2 years, 15 boys and 6 girls) and 21 children with spastic hemiplegia (age = 7.9 ± 1.7 years, 14 boys and 7 girls) were tested. The kinematic and kinetic data were collected synchronously and gait parameters were analyzed. The intersegmental dynamics analysis was conducted by a customized program based on ISD formulation.ResultsThe knee flexion angles (p = 0.045) and ankle dorsiflexion angles (p<0.001) of CSH were significantly lower than TDC. The Co-direction Ratio (CDR) of external contact torque of CSH was significantly lower than TDC (p = 0.015) during the first double stance phase. During the single stance phase, the CDR of hip external contact torque (p = 0.042), knee muscle torque p = 0.014), knee interactive torque (p < 0.001) of CSH were significantly greater than TDC. The net impulse of knee external contact torque (p < 0.001) and knee muscle torque (p < 0.001) of CSH were significantly greater than TDC. The net impulse of ankle gravitational torque of CSH was significantly lower than TDC (p = 0.03) during the swing phase.ConclusionA mismatch between muscle and passive torques may be associated with gait disturbance of CSH, characterized by a pathological shift in leading joint (hip to knee), inefficient passive torque utilization, and excessive yet ineffective muscle contractions. The findings suggest the potential relevance of rehabilitation strategies that focus on improving coordination during the stance phase and enhancing gravitational torque utilization during the swing phase.Supplementary InformationThe online version contains supplementary material available at 10.1186/s12984-025-01791-w. Introduction Pediatric cerebral palsy (PCP) is one of the most prevalent neurodevelopmental disorders associated with childhood disability worldwide [1, 2], with spastic hemiplegia emerging as a prevalent subtype characterized by unilateral impairments in gait and balance [3]. These gait impairments are the primary cause of recurrent falls of children with PCP, with their generally smaller, weaker, and less developed bones than typically developing children [4], leading to a greater risk of injuries and further limiting their independence, which imposes a substantial economic burden on families and society while significantly complicating their rehabilitation due to physical fragility and motor control deficits [5, 6]. CSH exhibit gait impairments characterized by excessive hip flexion, limited knee extension, and reduced ankle dorsiflexion during stance phase, accompanied by decreased gait speed and balance [7]. These deficits primarily arise from underlying neuromuscular impairments, including muscle spasticity, weakness, and impaired motor control [4]. Since gait requires the central nervous system to simultaneously coordinate multiple joints rather than isolated joint movements [8], understanding the multi-joint motor control mechanisms underlying these gait abnormalities is crucial for developing more effective interventions [9]. Multi-joint movements, particularly in a linked system, require CNS coordination of active torque (muscle torque) and passive torques (gravitational, external contact, and interactive torque) [10, 11]. Bernstein proposed that individuals with greater coordination effectively utilize passive torques, leveraging external forces such as gravity to enhance movement efficiency [12]. A variety of neurodevelopmental disorders, including CSH, may disrupt this coordination. Current studies on gait of CSH have mainly focused on identifying abnormal gait patterns through kinematic and kinetic analyses. Few studies describe the deficits and active compensation in CSH through the multi-joint motor control of lower limb. Zago et al. [13] identified distinct multi-segmental gait deviations using Principal Component Analysis. Children with spastic diplegia exhibited abnormal multi-joint coordination, particularly in knee flexion, hip adduction/abduction, pelvic obliquity, and trunk flexion. Similarly, Kuntze et al. utilized k-means clustering to identify four distinct multi-joint gait patterns in children with spastic diplegic cerebral palsy, revealing varied coordination deficits across the hip, knee, and ankle [14]. However, these studies focused on spastic diplegia and kinematic patterns during the stance phase [13, 14], without investigating the kinetic and neuromuscular mechanisms, such as joint torques or muscle spasticity, that underlie multi-joint coordination deficits and gait abnormalities [12]. ISD is a well-established theoretical framework that has been widely applied in the analysis of multi-joint movement patterns, including walking [15], vertical jump [16], arm movement [17, 18], et al. Our previous research has also utilized this approach to investigate segmental interactions during running [19]. In this context, ISD offers a valuable tool for uncovering the neuromechanical mechanisms underlying gait impairments in CSH, which may provide a foundation for personalized rehabilitation strategies. Additionally, ISD-derived torque could serve as biomarkers to track rehabilitation progress, offering clinicians a quantitative tool to assess intervention efficacy. The ISD formulation used here has been validated previously [20], confirming its applicability to multi-joint motor analysis. This study aims to quantify the multi-joint motor control of lower limb in CSH. The function and coordination of different torque components were analyzed. Based on previous studies, we hypothesized that CSH’s coordination of lower limb was significantly lower than typically developing children (TDC), unable to effectively utilize passive torques generated by environmental and joint movements. Methods This case-control study was approved by our local ethics committee (the ethics committee of Shaanxi Normal University, 202216002) under the Declaration of Helsinki. Participants The sample size was calculated by G Power (G* Power 3.1.9.7). Given an effect size (Cohen’s d) of 0.6, a desired statistical power of 0.95, and a significance level (α) of 0.05, resulting in a minimum sample size of 32 [21]. To account for an anticipated attrition rate of 20%, 1 typically developing children with no history of brain injury or other neurological conditions and 21 children with spastic hemiplegia were tested (Table 1). Participants and their immediate family members of guardians gave written informed consent to the experiment. Table 1Basic information of subjects (N = 42)N = 42TDCCSHAge (years)7.7 ± 1.27.9 ± 1.7SexMale1514Female67Affected sideRight limb–9Left limb12Height(cm)133.7 ± 7.1130.8 ± 6.9Weight(kg)34.1 ± 6.926.2 ± 4.1 Basic information of subjects (N = 42) Participants were recruited from two sources: Children with spastic hemiplegia were prospectively recruited in collaboration with Xi’an Children’s Hospital specifically for this study. Typically developing children were recruited through volunteer enrollment at the Affiliated Primary School of Shaanxi Normal University, and an age- and sex-matched control group was established to ensure comparability with the CSH participants. Criteria for inclusion of CSH: (1) Participants were diagnosed with spastic hemiplegia according to “Rehabilitation guidelines for cerebral palsy in China (2021)”, the stage of progression assessed as mid to late. (2) Participants were able to walk independently for 8 m or more. (3) Participants had not undergone neurosurgery or lower limb orthopedic surgery within the past year. Criteria for exclusion of CSH: (1) Spastic hemiplegia combined with mental retardation, visual and auditory abnormalities, epilepsy and other neurological disorders. (2) Incomplete clinical information or lack of consent from the attending physician. Apparatus and measurement The walkway (8 × 1.5 m) has two 900 × 600 mm force platforms (Model 9260AA6, Kistler Instrument, Switzerland, 1000 Hz), which was situated in a volume covered by an optoelectronic movement analysis system comprising 10 infrared cameras (Oqus700+, Qualisys AB, Sweden, 200 Hz) (Fig. 1). After the warm up, the subjects were required to walk on the walkway at a self-selected spontaneous speed. A valid acquisition was recorded by ensuring that the TDC’s right foot or CSH’s affect foot was completely stepped on one force platform. Each subject was required to successfully complete 3 valid acquisitions with a 30s break between each set of tests. All investigators were thoroughly trained, and all investigations were carried out under constant conditions (temperature controlled at 27 °C and kept quiet during the tests). The schematic diagram of experiment environment Data analysis The gait cycle of this study starts with the heel strike of right/affect foot and ends with second heel strike of the same foot. Gait events of the foot strike and foot off were identified with the Visual 3D using a threshold of 10 N on the vertical ground reaction force. A Cardan sequence with X-Y-Z defined the order of rotations according to the right-hand rule about the segment coordinate axes. Kinematic and kinetic signals were filtered using a fourth-order Butterworth digital filter with cutoff frequencies of 14 Hz and 50 Hz, respectively. The intersegmental dynamics analysis was conducted by a customized program based on ISD formulation, which has been validated in several studies [20]. The human body’s inertia parameters are determined according to the Chinese national standard (GB/T 17245 − 2004). Based on previous studies, we partitioned the torque at lower limb into five main components: net torque, muscle torque, interactive torque, gravitational torque, and external contact torque [22]. Net torque is the sum of all the other torque components. Muscle torque is primarily generated by muscle contraction. Interactive torque represents the rotational influence of forces resulting from segmental motion around the joints. Gravitational torque was calculated from gravitational forces acting at the center of mass of each segment. External contact torque in this study mainly results from ground reaction force. The calculation results were standardized based on the subjects’ height and weight to establish force/torque balance equations for the three segments of the lower limb rotating around their centers of mass. Further calculations of the torques at the subjects’ hip, knee, and ankle joints are performed using the Analysis of torque signs and the Analysis of net torque impulses (See Appendix A). The Co-direction Ratio (CDR): The analysis of torque signs. The ratio of the time during the gait cycle in which the other four torque components act in the same direction as the net torque [22]. where take muscle torque as an example, NET means the net torque, MUS means the muscle torque, CDR means the ratio of the time when muscle torque act in the same direction as the net torque, tSign means the time of muscle torque acts in the same direction as the net torque, and T means the time of gait cycle. The analysis of net torque impulses: The contribution of the other four torque components to the net torque during the gait cycle [22]. where take muscle torque as an example, NET means the net torque, MUS means the muscle torque. Statistical analysis Statistical analyses were performed using SPSS (SPSS Statistics v 26, IBM Corp., US) and MATLAB (MATLAB R2017a, The MathWorks Inc., US). Shapiro-Wilk test was used to assess the normality of the data. Cases (CSH, n = 21) and controls (TDC, n = 21) were frequency-matched for age (CSH: 7.9 ± 1.7 years; TDC: 7.7 ± 1.2 years) and sex (CSH: 14 boys, 7 girls; TDC: 15 boys, 6 girls) to minimize confounding. In the analysis of torque signs and the analysis of net torque impulses, for data that follow a normal distribution, use the independent samples t-test. For data that do not follow a normal distribution, use the Mann-Whitney U test to analyze statistical differences. Independent sample t-test implemented in the spm1d tool box was conducted to examine the difference between TDC and CSH in the joint angle and torque, which allows the comparison of trends in successive phases. Cohen’s effects size (ES) measures were computed as the significant differences were observed. For spm1d condition, the effect sizes were computed for each time point and averaged for the duration of the significant clusters [23]. Significance level was set at p < 0.05. Results The study only analyzed the joint angle and torque of the lower limb in the sagittal plane. Knee angles differed significantly at 45–66% of the gait cycle (p = 0.045, ES = 0.21). Ankle angles differed significantly at 0–5%, 27–51%, and 91–100% (p = 0.035, ES = 0.45; p < 0.001, ES = 0.32; p < 0.001, ES = 0.54) (Fig. 2). Hip torque differed significantly at 12–35% (p = 0.036, ES = 0.28), knee torque at 12–23% and 51–55% (p < 0.001, ES = 0.33; p = 0.03, ES = 0.25), and ankle torque at 40–55% (p < 0.001, ES = 0.39) (Fig. 3). The joint angle of the lower limb in the sagittal plane The joint torque of the lower limb in the sagittal plane During the first double stance phase. The CDR of external contact torque of CSH was significantly lower than TDC (F = 7.985, p = 0.015, ES = 0.45) (Fig. 4A). During the single stance phase, the CDR of hip external contact torque (F = 5.177, p = 0.042, ES = 0.33), knee muscle torque (F = 8.710, p = 0.014, ES = 0.17), knee interactive torque (F = 26.712, p < 0.001, ES = 0.36) of CSH were significantly greater than TDC. The knee external contact torque (F = 8.948, p = 0.011, ES = 0.47) and ankle muscle torque (F = 8.387, p = 0.013, ES = 0.28) of CSH were significantly lower than TDC (Fig. 4B). During the swing phase, the CDR of ankle gravitational torque of CSH was significantly lower than TDC (Z=-2.747, p < 0.001, ES = 0.49) (Fig. 4D). The CDR of gravitational torque, external contact torque, interactive torque, and muscle torque in lower limb: A represents the first double stance phase, B represents the single stance phase, C represents the second double stance phase, D represents the swing phase During the first double stance phase, the net impulse of hip external contact torque (F = 13.497, p < 0.01, ES = 0.55) and knee interactive torque (Z=-2.364, p = 0.018, ES = 0.17) of CSH was significantly greater than TDC. The net impulse of hip external contact torque of CSH was positive (I = 0.10), which was contrary to TDC (I=-0.07) (Fig. 5A). During the single stance phase, the knee external contact torque (F = 27.34, p < 0.001, ES = 0.74) and knee muscle torque (F = 27.34, p < 0.001, ES = 0.49) of CSH were significantly greater than TDC. The net impulse of ankle gravitational torque of CSH was significantly lower than TDC (F = 5.793, p = 0.03, ES = 0.18). The net impulse of hip external contact torque of CSH was positive (I = 0.10), which was contrary to TDC (I=-0.1) (Fig. 5B). During the second double stance phase, the net impulse of knee external contact torque of CSH was significantly lower than TDC (Z=-2.236, p = 0.025, ES = 0.59) (Fig. 5C). There was no significant difference during the swing phase (p > 0.05) (Fig. 5D). The net impulse of gravitational torque, external contact torque, interactive torque, and muscle torque in lower limb: A represents the first double stance phase, B represents the single stance phase, C represents the second double stance phase, D represents the swing phase Discussion This study investigates the multi-joint motor control of lower limb in the children with spastic hemiplegia, exploring the mechanisms of gait abnormalities. The gait of CSH is characterized by increased hip flexion, reduced ankle dorsiflexion, and a knee joint pattern showing reduced flexion angles suggestive of an extension tendency. The observed gait abnormalities in CSH may be associated with multi-joint motor control deficits, including mismatches between interactive and muscle torques during stance, and reduced utilization of gravitational torque during swing. These results largely support our hypotheses, the CSH may be unable to effectively utilize passive torques generated by environmental and joint movement. The knee flexion angles of CSH were significantly lower than TDC during the single stance phase and second double stance phase, which was consistent with previous studies [24]. The CSH’s muscles around the knee are in a state of over-contraction and unable to support the knee effectively. Furthermore, a comparative analysis of lower limb joint angles was conducted between the non-affected side of CSH and right lower limb of TDC, CSH demonstrated significantly greater hip flexion angles and reduced knee extension (Appendix C, Fig. S1), which was consistent with previous study [25]. This pattern suggests a compensatory mechanism on the non-affected side, likely intended to offset deficiencies in propulsion and stability from the affected limb. In contrast to the joint angle, CSH exhibited lower knee extension torque, during the single stance phase, even displayed flexion torque. This may indicate an attempt to achieve greater knee extension. Based on the gait pattern observed in TDC, the knee is expected to demonstrate an extension tendency during this phase to facilitate effective push-off. However, the excessive flexion tendency in CSH may reflect an inability to adequately extend the knee when extension is functionally required, potentially compromising gait efficiency. It may also be attributed to spastic muscle activity around the knee joint. In addition, the ankle dorsiflexion angles of CSH were significantly lower than TDC during the terminal single stance phase. The ankle of TDC has changed from plantarflexion to dorsiflexion during the pre-swing phase, while the CSH’ remained in plantarflexion for most of the swing phase. This obvious foot drop pattern was the most common characteristic of gait in CSH, which typically caused by cramps of triceps surae [24]. The hip flexion torque of CSH was significantly greater than TDC, which was contrary to the knee flexion torque and ankle dorsiflexion torque during the single stance phase. The excessive hip flexion and limited knee flexion occurred simultaneously during the pre-single stance phase. However, we didn’t find the excessive ankle dorsiflexion until the terminal single stance phase. The temporal ordering of difference may indicate that the passive adaptive strategy of CSH could be predictable. To accomplish this prediction, it’s necessary to analyze the leading joints within different phases [26]. According to the Leading Joint Hypothesis (LJH) [27] movement coordination in multi-joint systems is typically organized around a leading joint that initiates motion, while subordinate joints adjust accordingly through interactive torques. In TDC, our results indicate that the hip serves as the leading joint during the single stance phase, with knee interactive torques playing a compensatory role. This suggests that healthy gait relies on a proximal-to-distal control strategy, where the hip generates the primary force, and passive torques efficiently propagate through the lower limb. However, in CSH, we found a significant increase in the CDR of knee interactive torques and muscle torques, while hip torque contributions were reduced. This suggests a pathological shift in leading joint dominance from the hip to the knee, which could disrupt intersegmental coordination. Previous studies have shown that CP patients often exhibit compensatory strategies that shift mechanical demands to distal joints [8, 28]. Therefore, this shift may not represent an adaptive control strategy but rather a compensatory response to neuromuscular deficits. This altered control strategy may contribute to increased energy expenditure, inefficient force transmission, and instability during weight-bearing phases of gait. The impaired motor output from the primary motor cortex (M1) and corticospinal tract (CST) dysfunction may reduce hip joint control, may be associated with greater reliance on the knee joint for stability. Deficits in sensory feedback have been hypothesized to affect feedforward control. This may partially explain why CSH tend to rely more on the knee joint for reactive stability, potentially at the expense of efficient motion initiation. Efficient gait requires effective integration of passive torques, including interactive torques, gravitational torques, and external contact torques, to minimize unnecessary muscle activation and optimize movement efficiency [12, 22]. In TDC, passive torques were effectively utilized to support motion, whereas CSH demonstrated a reduced ability to integrate passive torques, particularly during stance and swing phases. During the single stance phase, CSH exhibited greater CDR/net impulse of knee muscle torques and less CDR/net impulse of external contact torque than TDC. We further illustrate the trend of different torque components across lower limb in Appendix (Appendix C, Fig. S2). Notably, during the first double stance phase to single stance phase, CSH exhibited a markedly lower knee extensor muscle torque compared to TDC, while simultaneously showing a greater net knee flexor torque. Although the knee musculature in CSH appears to engage in active contraction aimed at producing extension, the resulting joint dynamics still reflect a dominant flexion trend. This pattern may reflect a form of active yet ineffective muscle contraction, in which motor output does not appear to sufficiently counteract passive influences. Possible contributing factors include altered neuromuscular coordination, increased co-contraction, or suboptimal torque scaling. Previous studies indicate that children with spastic hemiplegia rely more on active muscle contractions rather than leveraging passive mechanical forces [19, 26]. This over-reliance on muscle-generated forces increases metabolic cost and reduces gait efficiency. Future studies combining ISD with surface EMG could clarify these mechanisms and inform more targeted interventions aimed at improving neuromuscular efficiency. During the swing phase, we observed that CSH had significantly lower CDR of ankle gravitational torque compared to TDC. Although the ankle torque during swing phase is close to zero in both groups, the gravitational torque component reflects the dynamic interaction between limb posture and segmental mass distribution. The significantly lower ankle gravitational torque in CSH may indicate suboptimal limb positioning and reduced capacity to exploit passive gravitational effects, which are important for efficient limb advancement even in the absence of active muscle force. In typical gait, passive gravitational torques assist in limb advancement, reducing the need for active muscle contraction [29]. This suggests a lack of neuromuscular integration of passive forces, likely due to deficits in sensory feedback and motor planning. These findings align with prior work suggesting that effective gait control relies not only on muscle strength but also on the ability to exploit passive mechanical properties [12, 28]. The inability of CSH to utilize passive torques may be attributed to spasticity-induced joint stiffness, impaired proprioception, and deficits in feedforward motor control. In addition to passive torque deficits, CSH exhibited excessive muscle contractions that failed to effectively compensate for intersegmental dynamics. Our results showed that knee muscle torques were significantly greater in CSH than in TDC, particularly in response to external contact torques. This suggests that muscle contractions in CSH were primarily reactive rather than anticipatory, meaning they were engaged to counteract instability rather than efficiently generating movement. Previous studies have suggested that excessive muscle co-contraction observed in CSH may be related to increased reflex excitability, which is often accompanied by heightened involuntary muscle stiffness [24]. Poor selective motor control, reducing the ability to selectively activate appropriate muscle groups. Reduced descending motor modulation from supraspinal centers has been implicated in compromised voluntary torque adjustment, which may contribute to impaired inter-joint coordination observed in CSH [5]. The misalignment between muscle torques and passive torques in CSH suggests that excessive contraction is not necessarily beneficial, it does not efficiently replace passive torque utilization but instead increases energy cost and movement rigidity. These findings emphasize the importance of training interventions that focus on optimizing torque utilization rather than just strengthening muscles. To address the leading joint shift from hip to knee dominance and inefficient knee extension torque production, exercises such as resisted hip extension (e.g., bridging, prone leg lifts) and functional stepping tasks can help restore hip dominance and reduce compensatory knee loading. Additionally, task-specific dorsiflexion training or FES targeting the tibialis anterior may improve swing-phase foot clearance by enhancing gravitational torque utilization at the ankle. Several possible limitations of this study should be acknowledged. (1) We did not test EMG indicators. Since the children with hemiplegia tend to have greater muscle tension, the EMG could better explain the differences better; (2) Although all kinetic variables were normalized to individual body size, and anthropometric scaling was applied in the ISD, the significant disparity in body weight between the two groups may still influence inertial properties and torque estimations to some extent; and (3) Due to experimental constraints of force plates, we were unable to collect complete joint moment data across a full gait cycle for the non-affected side of CSH. As a result, ISD analysis could not be conducted for the non-affected limb. Conclusion The gait of CSH is characterized by increased hip flexion, reduced ankle dorsiflexion, and a knee joint pattern with smaller flexion angles suggesting an extension tendency, accompanied by the characteristic drop foot gait. The mismatch between muscle torque and passive torque may be a critical contributor to gait impairments in CSH. CSH appeared to show a shift in leading joint dominance (from hip to knee), reduced utilization of passive torques, and excessive but inefficient muscle contractions. These observations may highlight potential targets for rehabilitation strategies that focus on improving coordination during the stance phase and enhancing gravitational torque utilization during the swing phase through better control of limb inertia. Supplementary Information Supplementary Material 1. Acknowledgments We thank the staff of the clinical motion analysis laboratory of Xi'an Children's Hospital. The study was supported by the 2023 General Project for Pedagogy of the National Social Science Fund of China "Research on an Intelligent Integrated Monitoring System for Body Posture in Children and Adolescents and Its Safeguard Mechanisms" (No. BLA230108). Abbreviations Pediatric cerebral palsy Inter Segment Dynamic Co-direction Ratio, the ratio of the time when other torque components act in the same direction as the net torque The contribution of the other four torque components to the net torque during the gait cycle Children with spastic hemiplegia Author contributions Conceptualization, Z.L.; Methodology, Z.L.; Data curation, X.C. and Q.W.; Data analysis, Z.L.; Writing—original draft preparation, Z.L.; Writing—review and editing, W.Z. and Y.S.; Supervision and ISD data processing method development, Y.L. Funding This research was partially supported by The National Social Science Fund of China (BLA230108). Institutional Review Board Statement: This case-control study was approved by our local ethics committee (the ethics committee of Shaanxi Normal University, 202216002) under the Declaration of Helsinki. Data availability The data presented in this study are available on request from the corresponding author. Declarations Ethics approval and consent to participate The informed consent was obtained from the students and their parents before each item was tested. The children signed the informed consent together with their parent/guardian or legally authorized representative (LAR). Consent for publication All authors have read and agreed to the published version of the manuscript. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Wenfei Zhu, Email: wzhu@snnu.edu.cn. Yuliang Sun, Email: ysun@snnu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The data presented in this study are available on request from the corresponding author.