Mehrdad Davoudi

Three-dimensional instrumented gait analysis is widely used to guide surgical decision-making in children with cerebral palsy (CP), but its interpretation is time-consuming and prone to inter-rater variability. In this single-centre pilot study, we investigated whether a generative large language model (LLM) could consistently generate gait deviation findings and surgical procedure suggestions that align with expert judgement. Kinematic features for lower-limb joints across the gait cycle, stance, and swing were extracted from eight children with unilateral CP using the open-source GaitSharing Toolkit and a structured prompt, then submitted three times per patient to OpenAI’s GPT-5.5 model. The model assessed 28 kinematic deviations and 12 surgical procedure groups using majority voting. One gait analyst and two paediatric orthopaedic surgeons independently rated outputs on a 0–2 ordinal scale, blinded to all clinical information beyond the kinematic curves and diagnosis. Agreement was summarised descriptively as the percentage of the maximum attainable score with 95% confidence intervals (CIs), and quadratic-weighted Cohen’s kappa was used to quantify inter-surgeon agreement. Agreement with the gait expert was highest at the hip (90.6%) and lowest at the knee, particularly in the transverse plane (65.2%). For surgical procedures, agreement with the LLM reached 83.9% and 73.4% for the two surgeons, with the tibialis anterior procedure showing the lowest concordance. Inter-surgeon agreement was 79.2% (95% CI 71.9–85.4) with a kappa of 0.59 (0.47–0.70), indicating moderate agreement. The LLM showed high self-consistency (>90% across runs). These preliminary findings suggest that generative LLMs may be feasible as assistive tools in clinical gait analysis for deviation detection and future treatment planning and should be interpreted as hypothesis-generating, warranting confirmation in larger, more diverse cohorts.

Three-dimensional instrumented gait analysis (3D-IGA) is widely used to guide surgical decision-making in children with cerebral palsy (CP), but its interpretation is time-consuming and prone to inter-rater variability. We investigated whether a generative large language model (LLM) could consistently generate gait deviation findings and surgical procedure suggestions that align with expert judgement. Kinematic features for lower-limb joints across the gait cycle, stance, and swing were extracted from eight children with unilateral CP using the open-source GaitSharing Toolkit and a structured prompt, then submitted three times per patient to OpenAI's GPT-5.5 model. The model assessed 28 kinematic deviations and 12 surgical procedure groups using majority voting. One gait analyst rated the deviations, and two paediatric orthopaedic surgeons independently rated the procedures on a 0–2 ordinal scale, blinded to all clinical information beyond the kinematic curves and diagnosis. Agreement with the gait expert averaged 84.7% in the sagittal plane and was lowest at the knee (62.6%). Surgeons’ agreement with the LLM reached 83.9% and 73.5%, with the tibialis anterior procedure showing the lowest concordance. Inter-surgeon agreement was 79.2%. The LLM showed high self-consistency (>90% across runs). This work demonstrates the potential of generative LLMs as assistive tools in clinical gait analysis for deviation detection and future treatment planning.

Instrumented gait analysis provides objective data for clinical assessment, with surface electromyography (EMG) serving as a key tool in identifying abnormal muscle activation. However, reliable reference data considering both age and gender remain limited. Age- and gender-related differences in lower-limb EMG during gait in typically developing individuals were examined in this study using statistical parametric mapping (SPM). We also determined the minimum sample size required for robust clinical reference data. Our findings revealed significant differences in muscle activation patterns across age and gender. Children exhibited increased rectus femoris activation in initial swing and greater hamstring activation in the midstance, whereas adults demonstrated greater semimembranosus activity at initial contact, increased soleus activation at push-off, and greater rectus femoris activity in late swing. Gender-based differences included greater tibialis anterior activation in females during the terminal stance and increased vastus lateralis activity during swing, whereas males showed greater vastus lateralis and biceps femoris activation during terminal swing. Additionally, significant age–gender interaction effects were observed in the biceps femoris and semimembranosus, with gender-related differences becoming more pronounced in adulthood. Power analysis indicates that at least 47 participants, with a minimum of 12 per subgroup (male children, female children, male adults, and female adults), are required to detect age–gender interactions reliably. We strongly recommend incorporating both age and gender in clinical norm bands to enhance the accuracy of gait assessments and improve clinical and research comparisons.

Instrumented gait analysis provides objective data for clinical assessment, with surface electromyography (EMG) serving as a key tool in identifying abnormal muscle activation. However, reliable reference data considering both age and gender remain limited. This study examines age- and gender-related differences in lower-limb EMG during gait in typically developing individuals using statistical parametric mapping (SPM). We also determine the minimum sample size required for robust clinical reference data. Our findings reveal significant differences in muscle activation patterns across age and gender. Children exhibited increased rectus femoris activation in initial swing and greater hamstring activation in mid-stance, whereas adults demonstrated higher semimembranosus activity at initial contact, increased soleus activation at push-off, and greater rectus femoris activity in late swing. Gender-based differences included higher tibialis anterior activation in females during terminal stance and increased vastus lateralis activity during swing, whereas males showed greater vastus lateralis and biceps femoris activation during terminal swing. Additionally, significant age-gender interaction effects were observed in the biceps femoris and semimembranosus, with gender-related differences becoming more pronounced in adulthood. Power analysis indicates that at least 47 participants, with a minimum of 12 per subgroup (male children, female children, male adults, female adults), are required to detect age-gender interactions reliably. Since our study included 97 subjects, our results are robust. We strongly recommend incorporating both age and gender in clinical norm bands to enhance the accuracy of gait assessments and improve clinical and research comparisons.

This study aimed to first investigate changes in electromyography (EMG) patterns after multilevel surgical treatment in patients with cerebral palsy (CP) and then to assess the connection between the measure of EMG and motor control indices and surgery outcomes. We analyzed retrospective EMG and gait data from 167 patients with CP before and after surgery and from 117 typically developed individuals as a reference group. The patients underwent at least one soft tissue surgery on their shank and foot muscles. Using Repeated Measures ANOVA, we examined the norm-distance (ND) of the kinematics, kinetics, and EMG patterns, in addition to the Kerpape-Rennes EMG-based Gait Index (EDI), EMG Profile Score (EPS), and Walking Dynamic Motor Control Index (DMC) before and after surgery. Participants were divided into different response groups (poor, mild, and good gait quality) according to their pre- and post-treatment Gait Deviation Index (GDI), using the K-means-PSO clustering algorithm. The gait and EMG indices were compared between the responders using the nonparametric Mann–Whitney test. The ND for all kinematics and kinetics parameters significantly improved (p-value < 0.05) after the surgery. Regarding EMG, a significant reduction was only observed in the ND of the rectus femoris (p-value < 0.001) and soleus (p-value = 0.006). Among the indices, DMC was not altered post-operatively (p-value = 0.88). Although EDI and EPS were consistent across responders with a similar pre-treatment gait, a higher DMC was significantly associated with a greater improvement, particularly in patients with poor gait (p-value < 0.05). These findings indicate systematic changes in the EMG of patients with CP following surgery, which can also be demonstrated through indices. DMC is a measure that can potentially serve as a partial predictor of outcomes, particularly in patients with poor pre-operative gait. Future research should investigate the effects of different surgical strategies on the improvement of these patients.

This study aimed to investigate how electromyography (EMG) cluster analysis of the rectus femoris (RF) could help to better interpret gait analysis in patients with cerebral palsy (CP). The retrospective gait data of CP patients were categorized into two groups: initial examination (E1, 881 patients) and subsequent examination (E2, 377 patients). Envelope-formatted EMG data of RF were collected. Using PCA and a combined PSO-K-means algorithm, main clusters were identified. Patients were further classified into crouch, jump, recurvatum, stiff and mild gait for detailed analysis. The clusters (labels) were characterized by a significant peak EMG activity during mid-swing (L1), prolonged EMG activity during stance (L2), and a peak EMG activity during loading response (L3). Notably, L2 contained 76% and 92% of all crouch patients at E1 and E2, respectively. Comparing patients with a crouch gait pattern in L2-E1 and L2-E2, two subgroups emerged: patients with persistent crouch (G1) and patients showing improvement at E2 (G2). The minimum activity of RF during 20–45% of the gait was significantly higher (p = 0.025) in G1 than in G2. A greater chance of improvement from crouch gait might be associated with lower RF activity during the stance phase. Using our findings, we could potentially establish an approach to improve clinical decision-making regarding treatment of patients with CP.

This study aimed to investigate how electromyography (EMG) cluster analysis of the rectus femoris (RF) could help to better interpret gait analysis in patients with cerebral palsy (CP). The retrospective gait data of CP patients were categorized into two groups: initial examination (E1, 881 patients) and subsequent examination (E2, 377 patients). Envelope-formatted EMG data of RF were collected. Using PCA and a combined PSO-K-means algorithm, main clusters were identified. Patients were further classified into crouch, jump, recurvatum, stiff and mild gait for detailed analysis. The clusters (labels) were characterized by a significant peak EMG activity during mid-swing (L1), prolonged EMG activity during stance (L2), and a peak EMG activity during loading response (L3). Notably, L2 contained 76% and 92% of all crouch patients at E1 and E2, respectively. Comparing patients with a crouch gait pattern in L2-E1 and L2-E2, two subgroups emerged: patients with persistent crouch (G1) and patients showing improvement at E2 (G2). The minimum activity of RF during 20-45% of the gait was significantly higher (p= 0.025) in G1 than in G2. A greater chance of improvement from crouch gait might be associated with lower RF activity during the stance phase. Using our findings, we could potentially establish an approach to improve clinical decision-making regarding treatment of patients with CP.

Chronic obstructive pulmonary disease (COPD) is increasingly being recognized as a systemic disease rather than a mere disorder of the lungs. Central (respiratory) and peripheral (limb) muscle weakness are among the main pronounced systemic effects of COPD. While the disease primarily affects the lower limb muscles and contributes to gait impairment, COPD is also associated with an increasing risk of falls in patients (COPDp). Previous studies have reported higher rates of falls among COPDp (1.17 to 1.20 falls/person-year), amounting to four times higher than an age-matched healthy group. Potential fall risk factors include muscle weakness, impaired daily activities, cognitive dysfunction, and gait and balance impairment. Although COPDp often manifest many of these risk factors, there remains a gap in literature regarding falls during walking in this population. This study aimed to 1. analyze the literature to identify the risk factors of falling in COPDp, and 2. investigate the underlying mechanisms by which these risk factors can lead to increased prevalence of falling. The results suggest that in addition to the known risk factors of falling, low back pain and mental fatigue should also be considered as relevant risk factors in the treatment process of these patients. Moreover, respiratory problems, which are common in this population, have demonstrated pronounced effects on energy expenditure, gait, and other types of activities of daily living (ADLs), leading to reduced intensity, disrupted coordination of the trunk-pelvic structure with the lower limbs during gait, and altered motor control performance due to activation of muscles in an inefficient synergic manner. These problems potentially lead to the increased vulnerability of these patients to external disturbances and higher incidence risk of falls and injuries. Cognitive problems, which are typically due to reduced oxygen received by the brain, as well as general inflammation caused by COPD, also play a significant role in gait disruption and balance. Future research is warranted to determine the prevalence of falls in COPDp by examining the response of these patients to Medio-Lateral (ML) and Anterior-Posterior (AP) disturbances during gait in association with traditional and recommended fall risk factors.

Central nervous system (CNS) uses vision, vestibular, and somatosensory information to maintain body stability. Research has shown that there is more lumbar proprioception error among low back pain (LBP) individuals as compared to healthy people. In this study, two groups of 20 healthy people and 20 non-specific low back pain (NSLBP) participants took part in this investigation. This investigation focused on somatosensory sensors and in order to alter proprioception, a vibrator (frequency of 70 Hz, amplitude of 0.5 mm) was placed on the soleus muscle area of each leg and two vibrators were placed bilaterally across the lower back muscles. Individuals, whose vision was occluded, were placed on two surfaces (foam and rigid) on force plate, and trunk angles were recorded simultaneously. Tests were performed in eight separate trials; the independent variables were vibration (four levels) and surface (two levels) for within subjects and two groups (healthy and LBP) for between subjects (4 × 2 × 2). MANOVA and multi-factor ANOVA tests were done. Linear parameters for center of pressure (COP) [deviation of amplitude, deviation of velocity, phase plane portrait (PPP), and overall mean velocity] and non-linear parameters for COP and trunk angle [recurrence quantification analysis (RQA) and Lyapunov exponents] were chosen as dependent variables. Results indicated that NSLBP individuals relied more on ankle proprioception for postural stability. Similarly, RQA parameters for the COP on both sides and for the trunk sagittal angle indicated more repeated patterns of movement among the LBP cohort. Analysis of short and long Lyapunov exponents showed that people with LBP caused no use of all joints in their bodies (non-flexible), are less stable than healthy subjects.

Central nervous system (CNS) uses vision, vestibular, and somatosensory information to maintain body stability. Research has shown that there is more lumbar proprioception error among low back pain (LBP) individuals as compared to healthy people. In this study, two groups of 20 healthy people and 20 non-specific low back pain participants (LBP) took part in this investigation. This investigation focused on somatosensory sensors and in order to alter proprioception, a vibrator (frequency of 70Hz, amplitude of 0.5 mm) was placed on the soleus muscle area of each leg and two vibrators were placed bilaterally across the lower back muscles. Individuals, whose vision was occluded, were placed on two surfaces (foam and rigid) on force plate, and trunk angles were recorded simultaneously. Tests were performed in 8 separate trials; the independent variables were vibration (4 levels) and surface (2 levels) for within subjects and 2 groups (healthy and LBP) for between subjects (4×2×2). MANOVA and multi-factor ANOVA tests were done. Linear parameters for center of pressure (COP) (deviation of amplitude, deviation of velocity, phase plane portrait (PPP), and overall mean velocity) and nonlinear parameters for COP and trunk angle ((recurrence quantification analysis) RQA and Lyapunov exponents) were chosen as dependent variables. Results indicated that NSLBP individuals relied more on ankle proprioception for postural stability. Similarly, RQA parameters for the COP on both sides and for the trunk sagittal angle indicated more repeated patterns of movement among the LBP cohort. Analysis of short and long Lyapunov exponents showed that people with LBP caused no use of all joints in their bodies (non-flexible), are less stable than healthy subjects.