Gait & posture最新文献

{"title":"Design of cushioned footwear for children with obesity based on gait dynamics and motion simulation","authors":"Yihong Zhao, Shiyang Yan, Ruoyi Li, Luming Yang, Bi Shi","doi":"10.1016/j.gaitpost.2023.07.276","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.276","url":null,"abstract":"Obesity will cause changes in foot structure and plantar pressure distribution, increasing the risk of foot pain and injury [1]. Functional footwear (outsole) is an essential way to distribute the local plantar pressure for children with obesity. However, the traditional design and research of outsoles need to go through the whole process of design, molding, production, fitting experiments, and so on, which is a long time and high-cost consumption. How to obtain the optimal design scheme of cushioned footwear for children with obesity through finite element analysis? Based on the database of foot morphology of children with obesity, a 3D outsole model was established, and the arch height of the outsole was set as 30%, 60%, and 100% of the arch height of children with obesity. Based on the anthropometric data, biomechanical data, and CT imaging data of children with obesity, a biomechanical simulation model of the lower limb musculoskeletal system and a finite element model of the foot were established. To verify the validity of the finite element model, the simulation results of the maximum principal stress of children with obesity during walking were compared with the actual measured data.The structure of the outsole is preliminarily constructed in Solidworks. The arch height (30%, 60%, and 100%) of the outsole was set to simulate the support at the arch. The foot-outsole-ground structure was assembled, and the pressure on the foot-shoe interface was simulated in ANSYS Workbench, to explore the dispersion effect of different arch heights. After obtaining the best design scheme, the actual relief effect of the outsoles was tested through the try-on trials. The simulation results showed that the 60% arch height support could effectively achieve the dispersion of plantar pressure in the plantar toe area and heel area. The try-on results showed that, when wearing the cushioned footwear, the peak pressure in the central forefoot and heel were relieved by 36.8% and 43.8%, respectively, from176.5 kPa and 310.9 kPa to 111.6 kPa and 174.7 kPa. Fig. 1 (a) 3D model of coushioned outsole. (b) Finite element analysis and verfication results. (c) Construction and assembly of the outsole structure. (d) The finite element analysis results between foot and outsole with the 60% arch height. (e) The cushioned footwear. (f) The cushioned effects of the outsole in the try-on experiments.Download : Download high-res image (244KB)Download : Download full-size image Through finite element analysis and fitting verification test, we found that when the arch height of the outsole is 60% of the arch height of the children with obesity, the decompression function is the best, which can transfer the pressure of the front palm and heel to the arch and toe. Finite element analysis makes functional shoe development process more efficient.","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135299048","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"There is life outside the gait lab: Effectiveness of a self-organising neural map for recognising 24/7 activities of daily living","authors":"Gabor Barton, Jacob Beesley, Jasmine Milnes, Gabriela Czanner, Lynne Boddy","doi":"10.1016/j.gaitpost.2023.08.018","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.08.018","url":null,"abstract":"","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"30 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135299061","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Reliability of 3D kinematic recording of jaw and head movements","authors":"Evelina Nilsson, Helena Grip, Catharina Österlund","doi":"10.1016/j.gaitpost.2023.07.171","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.171","url":null,"abstract":"The jaw and neck sensorimotor systems are functionally integrated during jaw functions1,2. The jaw border movements include maximum opening, laterotrusion to left and right, protrusion and retrusion3. Three-dimensional (3D) kinematic movement analysis provide data to distinguish natural movement patterns from those adapted to pain and dysfunction. Therefore, the reliability of kinematics is crucial to assess movement variability of integrated jaw-neck motor capacity. Can we ensure a high accuracy of the novel method intended to use for estimation of maximum jaw movements and geometric characterization (area and volume)? Is there a high test-retest reliability and intraindividual consistency for a group of healthy participants performing maximum jaw movements? 3D kinematic analysis was used for movement recognition. The first part included three glass beakers of different sizes, with known volumes and the cross-sectional area was estimated with a geometrical algorithm. The percentage deviation between target values and estimated values was calculated and to test the agreement a linear regression was made. The second part included 17 healthy participants (25.37 years ± 2.36). Maximum jaw movements were performed in a pre-determined movement pattern to track reflective marker positions of jaw and head segments. Movement amplitudes, magnitudes, areas, and volumes were analyzed. Intraclass correlation coefficient (ICC)4 estimates and Bland-Altman plots5 were used to assess test – retest reliability. Coefficient of variation (CV)6 tested the within session reliability. Preliminary results for the beakers showed a total percentage deviation from the target area and volume of 0.03 (SD 0.59) and 0.72 (SD 0.81), respectively. The linear regression showed a linear agreement between estimated and target value with R2=0.99. Preliminary results of test – retest reliability per movement outcome variable showed moderate to excellent reliability according to ICC-classification4. The limits of agreement between test and retest session presented with Bland-Altman plots showed good agreement between first and second measurement. The intra individual movement variability expressed as CV showed good repeatability. Jaw movements including the horizontal directions displayed widest ICC 95% confidence interval and highest CV values. (Fig. 1. Coefficient of variation - box plots).Download : Download high-res image (67KB)Download : Download full-size image This study addressed reliability of kinematic parameters of maximum jaw movements and its geometrics. The preliminary main findings indicate high accuracy of the novel method for estimations of volume and area. The agreement between sessions was considered good as well as consistency in repeated movements. Moreover, the more complex movement, the lower reliability and higher variability was seen. In future research of jaw-neck motor function the presented method is suggested to enables valid analysis of jaw movement perf","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"16 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135299063","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The effect of perturbation on hip kinematics of transtibial amputees","authors":"Nimet Sermenli Aydın, İlke Kurt, Halit Selçuk, Sinem Salar, Sezer Ulukaya, Hilal Keklicek","doi":"10.1016/j.gaitpost.2023.07.229","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.229","url":null,"abstract":"Gait adaptation to perturbations is essential for safe interaction with the physical environment and therefore it is important to understand how people with lower-limb amputation adapt their gait to changing conditions (1). Previous studies tried to find some deviation patterns and understand the kinematic strategies of amputee's gait (2). However, there is limited information available on the hip kinematics of amputees during gait and there is no study has yet investigated the effect of the perturbation on the hip kinematics of amputees. How does unpredictable continuous perturbation during gait affect the hip kinematics of unilateral transtibial amputees? Individuals with unilateral trans-tibial amputations and using prostheses with an active vacuum plus carbon foot combination were included in to study. Kinematic data of the hip were collected from 11 amputees and 10 healthy controls during walking on two different ground conditions. Participants walked at least 512 steps at their preferred speed on a motorized treadmill’s (ReaxRun Pro) flat ground condition and then the gait analysis was repeated on a perturbed (5% unpredictable perturbation) ground condition. RehaGait- Pro system was used for evaluation of the kinematics of the hip(min-max hip angles and variability of the hip min-max angles) during gait. Negative values indicated hip hyperextension, positive values indicated hip flexion. The statistical analysis was performed by pairing the residual limbs of amputees with the non-dominant side of the healthy group (RL side), and the sound limbs with the dominant side of the healthy group (HL side). It was observed that the hip hyperextension angle on the sound limb side was bigger in the amputees than in the control group on flat (d=0.462; p=0.034) and perturbated ground (d=0.584; p=0.007). The effect size was larger on the perturbed ground. There was no difference in the maximum hip angles and variability of max-min hip angles between the groups in both ground conditions (p>0.05). The results showed in Table.Download : Download high-res image (142KB)Download : Download full-size image Amputation-related changes were observed in hip kinematics during walking under both ground conditions. However, this change was more prominent on the perturbated ground. The reason for the higher hip hyperextension values in amputees is thought to be due to their efforts to compensate for the ankle (exp. strong plantar flexion) movements. On the unpredictable perturbation ground, the limitation of ankle movements, which is one of the first adaptive mechanisms in adaptation to the ground (exp. subtalar rotations plus plantarflexion), may have made the situation more evident. Future studies may focus on the effect of gait training on perturbed surfaces on gait kinematics, which is an indicator of adaptation to variable conditions.","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"20 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135297885","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A comparative study on wearables and single-camera video for upper-limb out-of-the-lab activity recognition with different deep learning architectures","authors":"Mario Martínez Zarzuela, David González-Ortega, Míriam Antón-Rodríguez, Francisco Javier Díaz-Pernas, Henning Müller, Cristina Simón-Martínez","doi":"10.1016/j.gaitpost.2023.07.149","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.149","url":null,"abstract":"The use of a wide range of computer vision solutions, and more recently high-end Inertial Measurement Units (IMU) have become increasingly popular for assessing human physical activity in clinical and research settings [1]. Nevertheless, to increase the feasibility of patient tracking in out-of-the-lab settings, it is necessary to use a reduced number of devices for movement acquisition. Promising solutions in this context are IMU-based wearables and single camera systems [2]. Additionally, the development of machine learning systems able to recognize and digest clinically relevant data in-the-wild is needed, and therefore determining the ideal input to those is crucial [3]. For upper-limb activity recognition out-of-the-lab, do wearables or single camera offer better performance? Recordings from 16 healthy subjects performing 8 upper-limb activities from the VIDIMU dataset [4] were used. For wearable recordings, the subjects wore 5 IMU-based wearables and adopted a neutral pose (N-pose) for calibration. Joint angles were estimated with inverse kinematics algorithms in OpenSense [5]. Single-camera video recordings occurred simultaneously. Joint angles were estimated with inverse kinematics algorithms in OpenSense. Single-camera video recordings occurred simultaneously, and the subject’s pose was estimated with DeepStream [6]. We compared various Deep Learning architectures (DNN, CNN, CNN-LSTM, LSTM-CNN, LSTM, LSTM-AE) for recognizing daily living activities. The input to the different neural architectures consisted in a 2-second time series containing the estimated joint angles and their 2D FFT. Every network was trained using 2 subjects for validation, a batch size of 20, Adam as the optimizer, and combining early stopping and other regularization techniques. Performance metrics were extracted from 4-fold cross-validation experiments. In all neural networks, performance was higher with IMU-based wearables data compared to video. The best network was an LSTM AutoEncoder (6 layers, 700 K parameters; wearable data accuracy:0.985, F1-score:0.936 (Fig. 1); video data accuracy:0.962, F1-score:0.842). Remarkably, when using video as input there were no significant differences in the performance metrics obtained among different architectures. On the contrary, the F1 scores using IMU data varied significantly (DNN: 0.849, CNN: 0.889, CNN-LSTM: 0.879, LSTM-CNN: 0.904, LSTM: 0.920, LSTM-AE: 0.936).Download : Download high-res image (108KB)Download : Download full-size image Wearables and video present advantages and disadvantages. While IMUs can provide accurate information about the orientation and acceleration of body parts, body-to-segment calibration and drift can affect data reliability. Similarly, a single camera can easily track the position of different body joints, but the recorded data does not yet reliably represent the movement with all degrees of freedom. Our experiments confirm that despite the current limitations of wearables, with a very si","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"22 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135297896","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comparison of neuromuscular and abductor strengthening exercises in the hip abductor muscle strength: A randomized controlled trial","authors":"Sofia Pastrouma, Filippos Kasiotis, Aikaterini - Evanthia Gkanatsiou, Natalia Kitsouli, Konstantinos Vassis, Zacharias Dimitriadis, Savvas Spanos, Ioannis Poulis","doi":"10.1016/j.gaitpost.2023.07.193","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.193","url":null,"abstract":"Decreased hip abductor strength has been associated with a range of knee pathologies. Hip abduction muscles play a critical role in providing pelvic stability and leg alignment during weight-bearing movements by eccentrically controlling hip adduction. Poor hip control can result in abnormal lower extremity motions, and studies have reported that lower limb problems such as patellofemoral pain (PFP),1 knee osteoarthritis (OA),2 and ACL injuries,3,4 are linked with altered lower limb kinematics, with a higher prevalence in females.5 A body of literature suggests that increased dynamic knee valgus is associated with hip abductor weakness.6 Moreover, several studies have shown that interventions such as neuromuscular training (NMT) programs can lower the incidence of lower extremity problems. An NMT program emphasizing optimal alignment of the trunk and lower limb joints relative to each other, along with quality movement performance while dynamically and functionally strengthening the lower limb muscles, may be better at enhancing hip muscle strength. Therefore, we performed a randomized controlled trial evaluating the effects of NMT in comparison to a progressive resistance training program (PRT) on muscle hip abductor strength. To investigate whether a 6-week NMT can improve the hip abductor muscle strength better than a PRT. The present study was a single-blind randomized controlled trial aiming to investigate the effects of two interventions on asymptomatic females, aged 18-35 years old. Sample size calculation revealed that 26 participants per group were required. Following a baseline assessment, 52 participants were randomly assigned to either a 6-week PRT or NMT intervention involving 3 sessions per week. The PRT intervention consisted of hip abductor exercises performed in an open kinetic chain, with three to four sets of ten repetitions at a target intensity of 6-8 RPE.7,8,9 The NMT intervention focused on improving functional stability, balance, proprioception, strength, agility, postural function, and orientation,10,11 consisting of weight-bearing positions. The participants were assessed after the 6-week intervention. Mean peak hip abduction, concentric and eccentric torque, were measured by a blinded assessor on a Biodex System 3 Pro isokinetic dynamometer at 60°/s. Dependent t-tests showed significant improvements in CON60, and ECC60 after both interventions (<0.05) (Table 1). Two-way mixed ANOVAs did not reveal statistically significant Group*Time interactions for the CON60 and ECC60. The results from the comparison of the effectiveness of each intervention are visually presented in Figs. 1 and 2. Download : Download high-res image (114KB)Download : Download full-size image Both PRT and NMT improved abductor strength. However, both groups had similar overall differences in strength before and after the intervention. Since no intervention is superior to the other, neuromuscular training might be clinically preferred as it combines dyn","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"43 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135298030","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"The volume of the lateral gastrocnemius appears reduced in some Idiopathic toe walkers","authors":"Anne Mcnee, Jonathan Noble, Stuart Evans, Karen Ziegler, Stephen Ng Man Sun, Alison Hulme, Nicola Fry, Adam Shortland","doi":"10.1016/j.gaitpost.2023.07.172","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.172","url":null,"abstract":"Plantarflexion contractures are often the focus for intervention in children who toe walk (TW). Caserta et.al1 found reduced plantarflexor strength in TW and greater proportions of type 1 fibres were identified in the plantarflexors2. Variable but mild differences in kinematics have been found between children with mild bilateral cerebral palsy (CP) and TW3,4. Children with CP have reduced muscle volumes compared to typically developing children5. Plantarflexor morphology in TW has not yet been described. Is ankle plantarflexor volume reduced in children who toe walk? Eight children (5male) aged 7-15 yr (mean=11.86 yrs) referred to our orthopaedic department for toe walking and plantarflexion contractures, with no underlying diagnosis, had a routine examination in the gait laboratory. They were matched for age and sex to children with CP (GMFCS I-II) who had also been examined. Assessment included gait analysis and 2D ultrasound imaging of the lateral gastrocnemius(LG). Muscle volumes were estimated by the Vanmechelen et.al6 method, normalised to mass. Selective motor control (SCALE) was assessed according to Fowler et.al7. Mobility was assessed using the Gillette Functional Assessment Questionnaire (GFAQ) 8. Data was compared to a large database of controls (unpaired t-test) and between groups (paired t-test). One limb per subject was randomly selected for analysis. All children had plantarflexor contractures: mean passive dorsiflexion range (knee extended) of -9.4° (SD10.9°) for TW and -6.5° (SD7.2°) for CP. TW had close to normal motor control (SCALE:Median=10, Range=8-10) whereas CP had a greater variability (SCALE:Median=9.5, Range=5-10). Walking function was within normal limits for TW (GFAQ Median=10 Range=8-10) but more variable for CP (GFAQ Median=8 Range=5-10). No difference in speed/cadence was found between groups (p=0.5/p=0.86) and these were within normal limits. All children were in ankle plantarflexion at initial contact (no difference between groups, p=0.48). Mean ankle dorsiflexion in stance and swing were not different between groups (p=0.94, p=0.84). For four TW children, normalised mean LG volume was significantly smaller than controls (1.07vs1.53 ml/kg) (p<0.01) but no different to CP (1.01 ml/kg) (p=0.64). The other TW had LG CSA which was too great for the US field of view. In the presence of an ankle plantarflexion contracture, TW children show less variability in selective motor control and functional mobility to a matched CP group. TW and CP show similar kinematics at the ankle, cadence and speed. A subgroup of TW children had reduced normalised LG compared to control data, comparable in size to the CP group. Other subjects’ muscles were larger and could not be measured. This suggests subgroups of TW with different muscle sizes, which has implications for aetiology and management. Further work is required to further elucidate the triceps surae muscle morphology in TW and relationship between morphology and toe walking.","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"11 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135298032","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Gait deviations in rare genetic syndromes: is there a common denomitator for patients with Dravet, HVDAS and TSC?","authors":"Patricia Van De Walle, An Jansen, An-Sofie Schoonjans, Anke Van Dijck, Colette Puts, Iris van Hal, Marijn Weren, Kinaci Esra, Ann Hallemans","doi":"10.1016/j.gaitpost.2023.08.027","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.08.027","url":null,"abstract":"","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"57 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135298034","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Age related changes in lower-limb joint coordination during gait in children with bilateral cerebral palsy","authors":"Damien Kiernan, Ailish Malone","doi":"10.1016/j.gaitpost.2023.08.024","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.08.024","url":null,"abstract":"","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"24 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135298035","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Explainable machine learning approach on biomechanical features to identify weakness in a population-based setting on aging","authors":"Mariapia Musci, Simona Aresta, Francesco Bottiglione, Michele Ruta, Tommaso Di Noia, Rodolfo Sardone, Ilaria Bortone","doi":"10.1016/j.gaitpost.2023.07.167","DOIUrl":"https://doi.org/10.1016/j.gaitpost.2023.07.167","url":null,"abstract":"Weakness, as measured by maximal Hand Grip Strength (HGS), represents one of the five criteria used in Fried's definition of frailty [1] and is associated with a wide range of health conditions, which makes it challenging to delineate what body system processes are responsible for weakness. [2]. Still, poor studies have investigated the associations between HGS and dynamic functional assessments [3]. To identify a pattern of functional characteristics, extracted from the 5-repetitions-sit-to-stand (5STS) test biomechanical signals best predict weakness. An Explanation approach to a Machine Learning model was also used. In a subcohort of the longitudinal study of aging [4], 86 subjects over 65 performed the 5STS test [5,6]. They were equipped with an IMU on the L5 vertebra and four sEMG probes (BTS Bioengineering) on the Gastrocnemius Medialis and Tibialis Anterior both side muscles. Several kinematic and muscular features were extracted from the cycle, standing and sitting phases. A handgrip dynamometer was used to measure HGS. Men and women who were considered weak had HGS<26 kg and <16 kg, respectively. Socio-demographic information (age, sex and BMI) was also included. The final dataset consisted of 119 features for all subjects. We first performed the undersampling of the majority class (without weakness); then the dataset was divided into 70% training and 30% testing and normalised using the z-score method. Because of the curse of dimensionality, a pipeline for feature selection and hyperparameter tuning, using the GridSearchCV method, was defined to obtain the best Kernel-SVM model. The best model was chosen according to the accuracy score. To evaluate our model accuracy, precision and recall were calculated. All the analyses were performed using the Scikit-Learn library [7] with Python 3.6. To explain our model Python's SHAP library was used [8]. From the hyperparameter tuning, we obtained six features: hip power (Whip), power along the vertical axis (Wvert), and cycle jerk along the vertical axis and its coefficient of variation, age, and sex. Fig. 1 shows the boxplots for the biomechanical selected variables.The model showed 90.0% and 85.7% accuracy on the training and testing sets, respectively. The precision of 100%, recall of 71%, and f1-score 83%, while the precision of 78%, recall of 100%, and f1-score of 88% was obtained on the class without weakness and its counterpart, respectively.The explainability analysis showed that age, Wvert and Whip were the three most important variables in predicting weakness in absolute terms. Sex resulted being the least important variable. Picture 1 - \"Boxplot of the biomechanical selected features according to the weakness condition\"Download : Download high-res image (71KB)Download : Download full-size image Measures of HGS are associated with deficits in several physical functions. In a population-based setting, we identified biomechanical features from 5STS related to stability that could help pre","PeriodicalId":94018,"journal":{"name":"Gait & posture","volume":"1 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2023-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"135298039","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}