Whole-body Motion Research Articles

Motor imagery and action observation of whole-body movements for experienced motor repertoire: an fNIRS study

Motor imagery and action observation of whole-body movements for experienced motor repertoire: an fNIRS study

Development of a low-cost and user-friendly system to create personalized human digital twin.

Creating a digital twin has enormous potential in biomedical engineering. However, it is also technically challenging. No existing system can allow people who don't have the art-and-design background to create their own digital twin. To fill this gap, this study proposes a low-cost wearable system and a user-friendly framework for creating personalized digital twins with a fast speed and high fidelity. The personalized human digital twin can capture synchronized facial expressions, gaze direction, and whole-body movements for real-time rendering. The system simplifies the complex process of creating digital humans, and allows for the creation of data-driven characters without specialized skills.Clinical Relevance- This system can be used to help doctors keep track record of the patient's health status in a more visual and realistic way, supporting them in making more accurate clinical decisions, and facilitating a more detailed medical intervention.

Sensory prediction error drives subconscious motor learning outside of the laboratory.

Sensorimotor adaptation is supported by at least two parallel learning systems: an intentionally controlled explicit strategy and an involuntary implicit learning system. Past work focused on constrained reaches or finger movements in laboratory environments has shown subconscious learning systems to be driven in part by sensory prediction error (SPE), i.e., the mismatch between the realized and expected outcome of an action. We designed a ball rolling task to explore whether SPEs can drive implicit motor adaptation during complex whole body movements that impart physical motion on external objects. After applying a visual shift, participants rapidly adapted their rolling angles to reduce the error between the ball and the target. We removed all visual feedback and told participants to aim their throw directly toward the primary target, revealing an unintentional 5.06° implicit adjustment to reach angles that decayed over time. To determine whether this implicit adaptation was driven by SPE, we gave participants a second aiming target that would "solve" the visual shift, as in the study by Mazzoni and Krakauer (Mazzoni P, Krakauer JW. J Neurosci 26: 3642-3645, 2006). Remarkably, after rapidly reducing ball-rolling error to zero (due to enhancements in strategic aiming), the additional aiming target caused rolling angles to deviate beyond the primary target by 3.15°. This involuntary overcompensation, which worsened task performance, is a hallmark of SPE-driven implicit learning. These results show that SPE-driven implicit processes, previously observed within simplified finger or planar reaching movements, actively contribute to motor adaptation in more complex naturalistic skill-based tasks.NEW & NOTEWORTHY Implicit and explicit learning systems have been detected using simple, constrained movements inside the laboratory. How these systems impact movements during complex whole body, skill-based tasks has not been established. Here, we demonstrate that sensory prediction errors significantly impact how a person updates their movements, replicating findings from the laboratory in an unconstrained ball-rolling task. This real-world validation is an important step toward explaining how subconscious learning helps humans execute common motor skills in dynamic environments.

A Dynamical Systems Approach to Characterizing Brain-Body Interactions during Movement: Challenges, Interpretations, and Recommendations.

Brain-body interactions (BBIs) have been the focus of intense scrutiny since the inception of the scientific method, playing a foundational role in the earliest debates over the philosophy of science. Contemporary investigations of BBIs to elucidate the neural principles of motor control have benefited from advances in neuroimaging, device engineering, and signal processing. However, these studies generally suffer from two major limitations. First, they rely on interpretations of 'brain' activity that are behavioral in nature, rather than neuroanatomical or biophysical. Second, they employ methodological approaches that are inconsistent with a dynamical systems approach to neuromotor control. These limitations represent a fundamental challenge to the use of BBIs for answering basic and applied research questions in neuroimaging and neurorehabilitation. Thus, this review is written as a tutorial to address both limitations for those interested in studying BBIs through a dynamical systems lens. First, we outline current best practices for acquiring, interpreting, and cleaning scalp-measured electroencephalography (EEG) acquired during whole-body movement. Second, we discuss historical and current theories for modeling EEG and kinematic data as dynamical systems. Third, we provide worked examples from both canonical model systems and from empirical EEG and kinematic data collected from two subjects during an overground walking task.

A -2- month old child with Spinal Muscular Atrophy Type 1 with Pneumonia: A case report

Spinal muscular atrophy (SMA) is a autosomal recessive disease characterized by muscular weakness, hypotonia and atrophy. SMA is caused by mutation or deletion of the survivor motor neuron gene (SMN1), which is located in the telomere region of chromosome 5q13. Incidence of the disease is 1:6000-10000 newborn. Confirmatory diagnosis can be established by molecular genetic analysis. Here a 2- month- old female baby was admitted into the department of pediatrics of Sylhet Women’s Medical College Hospital with because of less movement of whole body for 1 month and repeated cough with respiratory distress for same duration. Baby was dyspneic, chest in drawing and crepitation was present over both lung fields. Signs of lower motor neuron lesions were present. It was confirmed by Polymerase chain reaction (PCR) for SMN gene. Final diagnosis was Spinal muscular atrophy with pneumonia. Symptomatic, Supportive and antibiotic treatment were given. Genetic counseling, regular respiratory follow up was advised.

High Utility Teleoperation Framework for Legged Manipulators Through Leveraging Whole-Body Control

Legged manipulators are a prime candidate for reducing risk to human lives through completing tasks in hazardous environments. However, controlling these systems in real-world applications requires a highly functional teleoperation framework, capable of leveraging all utility of the robot to complete tasks. In this work, such a teleoperation framework is presented, where a wearable whole-body motion capture suit is integrated with a whole-body controller specialised for teleoperation and a set of teleoperation strategies that enable the control of all main frames of the robot along with additional functions. Within the whole-body controller, all tasks and constraints can be configured dynamically due to their modularity, hence enabling seamless transitions between each teleoperation strategy. As a result, this not only enables the realisation of trajectories outside the workspace without the whole-body controller but also the ability to complete tasks that would require an additional manipulator if just the gripper frames of the robot were controllable. To validate the presented framework, a set of real robot experiments have been completed to demonstrate all teleoperation strategies and analyse their proficiency.

Is Social Training Delivered with a Head-Mounted Display Suitable for Patients with Hereditary Ataxia?

Social cognition is fundamental in everyday life to understand "others' behavior", which is a key feature of social abilities. Previous studies demonstrated the efficacy of a rehabilitative intervention in semi-immersive virtual reality (VR) controlled by whole-body motion to improve the ability of patients with cerebellar disorders to predict others' intentions (VR-SPIRIT). Patients with severe ataxia that have difficulties at multiple levels of social processing could benefit from this intervention in terms of improving their social prediction skills, but they may have difficulties in controlling VR with whole-body movements. Therefore, we implemented VR-SPIRIT on a wearable, affordable, and easy-to-use technology, such as the Oculus Quest, a head-mounted display. The aim of this work was to evaluate the usability and tolerability of this VR application. We recruited 10 patients (37.7 ± 14.8 years old, seven males) with different types of hereditary ataxia who performed a single VR-SPIRIT session using the Oculus Quest viewer. After the session, patients answered a series of questionnaires to investigate the overall usability of the system and its potential effects in terms of cyber sickness. The preliminary results demonstrated system usability and tolerability. Indeed, only three patients did not complete the session due to different problems (dizziness, nausea, and boredom). In future studies, more patients will be enrolled to assess the effectiveness of the application, paving the way for the implementation of social training that can also be delivered at home.

Individual muscle contributions to the acceleration of the centre of mass during gait in people with mild-to-moderate hip osteoarthritis

BackgroundPeople with mild-to-moderate hip osteoarthritis (OA) exhibit hip muscle weakness, alterations in hip kinematics and kinetics and hip contact forces during gait compared to healthy controls. However, it is unclear if those with hip OA use different motor control strategies to coordinate the motion of the centre of mass (COM) during gait. Such information could provide further critical assessment of conservative management strategies implemented for people with hip OA. Research questionDo muscle contributions to the acceleration of the COM during walking differ between individuals with mild-to-moderate hip OA and controls? MethodsEleven individuals with mild-to-moderate hip OA and 10 healthy controls walked at a self-selected speed while whole-body motion and ground reaction forces were measured. Muscle forces during gait were obtained using static optimisation and an induced acceleration analysis was performed to determine individual muscle contributions to the acceleration of the COM during single-leg stance (SLS). Between-group comparisons were made using independent t-tests via Statistical Parametric Modelling. ResultsThere were no between-group differences in spatial-temporal gait parameters or three-dimensional whole-body COM acceleration. The rectus femoris, biceps femoris, iliopsoas and gastrocnemius muscles in the hip OA group contributed less to the fore-aft accelerations of the COM (p < 0.05), and more to the vertical COM acceleration with the gluteus maximus (p < 0.05), during SLS, compared to the control group. SignificanceSubtle differences exist in the way people with mild-to-moderate hip OA use their muscles to accelerate the whole-body centre of mass during the SLS phase of walking relative to healthy controls. These findings improve understanding of the complex functional consequences of hip OA and enhance our understanding of how to monitor the effectiveness of an intervention on biomechanical changes in gait in people with hip OA.

Action-Conditioned Generation of Bimanual Object Manipulation Sequences

The generation of bimanual object manipulation sequences given a semantic action label has broad applications in collaborative robots or augmented reality. This relatively new problem differs from existing works that generate whole-body motions without any object interaction as it now requires the model to additionally learn the spatio-temporal relationship that exists between the human joints and object motion given said label. To tackle this task, we leverage the varying degree each muscle or joint is involved during object manipulation. For instance, the wrists act as the prime movers for the objects while the finger joints are angled to provide a firm grip. The remaining body joints are the least involved in that they are positioned as naturally and comfortably as possible. We thus design an architecture that comprises 3 main components: (i) a graph recurrent network that generates the wrist and object motion, (ii) an attention-based recurrent network that estimates the required finger joint angles given the graph configuration, and (iii) a recurrent network that reconstructs the body pose given the locations of the wrist. We evaluate our approach on the KIT Motion Capture and KIT RGBD Bimanual Manipulation datasets and show improvements over a simplified approach that treats the entire body as a single entity, and existing whole-body-only methods.

Whole-body kinematic adaptations to running on an unstable, irregular, and compliant surface

ABSTRACT The goal of this study was to investigate whole-body kinematic adaptations when running on an unstable, irregular, and compliant surface in comparison to running on asphalt. We hypothesised that the gait pattern (H1) and its stride-to-stride variability (H2) would be affected by the unstable surface but that variability related to some movement features would be reduced over multiple testing days indicative of gait optimisation (H3). Fifteen runners ran on a woodchip and asphalt track on five testing days while their whole-body movements were captured using inertial motion capture and examined using joint angle and principal component analysis. Joint angles and stride-to-stride variability in eight principal running movements were subjected to surface by day analyses of variance. The woodchip track compared to asphalt resulted in (H1) a more crouched gait pattern including more leg flexion and forward trunk lean and (H2) higher stride-to-stride variability in most investigated principal running movements. However, (H3) stride-to-stride variability did not systematically change over testing days. Running on an unstable, irregular, and more compliant surface leads to the adoption a gait pattern and control strategy that are more robust against disturbances caused by the surface but may pose certain risks for overuse injury in trail runners.

Gas turbine failure classification using acoustic emissions with wavelet analysis and deep learning

Compared to vibration monitoring, acoustic emission (AE) monitoring in gas turbines is highly sensitive to changes that do not involve whole-body motion, such as wear, rubbing, and fluid-induced faults. AE signals captured by suitably mounted sensors can potentially provide early indications of abnormal turbine operation before such abnormalities manifest in structural vibration or emitted airborne noise. However, developing an online fault detection system requires extensive real-time data treatment to extract appropriate features and indicators from raw AE records. To build such a system for industrial turbines, researchers need to understand the AE-generating mechanisms associated with turbine operation and the sources of background noise. In this study, we aim to develop such an understanding using a small-scale turbine whose operational conditions can be modified safely to reflect both normal and faulty conditions. Our signal processing approach involves first extracting a time-series envelope using an averaging time selected to enhance major features and eliminate irrelevant noise. We then generate time–frequency features using a continuous wavelet transform, which are used to train a deep convolutional neural network to classify gas turbine conditions. The resulting model demonstrates high accuracy in classifying two normal running conditions and two faulty conditions at various turbine speeds. Overall, the proposed methodology offers a powerful tool for gas turbine condition monitoring, and we make all associated data available in open-source format to facilitate further research in this field.4

Cerebellar contributions to a brainwide network for flexible behavior in mice

The cerebellum regulates nonmotor behavior, but the routes of influence are not well characterized. Here we report a necessary role for the posterior cerebellum in guiding a reversal learning task through a network of diencephalic and neocortical structures, and in flexibility of free behavior. After chemogenetic inhibition of lobule VI vermis or hemispheric crus I Purkinje cells, mice could learn a water Y-maze but were impaired in ability to reverse their initial choice. To map targets of perturbation, we imaged c-Fos activation in cleared whole brains using light-sheet microscopy. Reversal learning activated diencephalic and associative neocortical regions. Distinctive subsets of structures were altered by perturbation of lobule VI (including thalamus and habenula) and crus I (including hypothalamus and prelimbic/orbital cortex), and both perturbations influenced anterior cingulate and infralimbic cortex. To identify functional networks, we used correlated variation in c-Fos activation within each group. Lobule VI inactivation weakened within-thalamus correlations, while crus I inactivation divided neocortical activity into sensorimotor and associative subnetworks. In both groups, high-throughput automated analysis of whole-body movement revealed deficiencies in across-day behavioral habituation to an open-field environment. Taken together, these experiments reveal brainwide systems for cerebellar influence that affect multiple flexible responses.

Is the use of the coefficient of variation a valid way to assess the swimming intra-cycle velocity fluctuation?

Swimming intra-cycle velocity fluctuation has often been assessed using the coefficient of variation, which requires a mathematical assumption of a positive linear relationship between the velocity mean and standard deviation. As this assumption has never been tested, the current study aimed to investigate the within-participant relationship between the mean and standard deviation of the intra-cycle velocity. Cross-sectional study. The intra-trial mean and standard deviation of one stroke cycle centre of mass velocity (vCMmean and vCMSD, respectively) were obtained from 80 front crawl trials (10 participants × eight swimming speeds) using whole-body three-dimensional motion analysis. The linear mixed-effect model and intra-class correlation analysis were used to test the linear relationship between vCMmean and vCMSD (n = 80) and the absolute agreement between vCMmean and vCMSD relative to those during the fastest trial (n = 70). Neither the linear regression model (95 % confidence interval range of the fixed effect of vCMmean: -0.003-0.031) nor the intra-class correlation coefficient (ICC = 0.07; p = 0.26) verified linear relationships between vCMmean and vCMSD, which violated the background assumption of coefficient of variation calculation. When investigating the intra-cycle velocity fluctuation, the coefficient of variation should not be used alone. Researchers and practitioners should always interpret/report the obtained results together with the mean and standard deviation to avoid misleading conclusions and feedback because the coefficient of variation obtained from one cycle velocity data is likely biased by mean velocity.

Exergames to Limit Weight Gain and to Fight Sedentarism in Children and Adolescents with Obesity

Exergames are defined as digital games that require bodily movements to play, stimulating an active gaming experience to function as a form of physical activity (PA). The players interact with the game through whole-body movements improving energy expenditure. Exergames may be effective in improving physical and psychological aspects of children and adolescents with obesity. In this narrative review, we synthesized the current evidence regarding the role of exergames in modifying body composition and weight and in promoting changes in sedentary behavior to define the benefits of active video games as useful tools for fighting sedentarism and to outline the future directions of exergaming as a supplementation exercise rather than a replacement in educational programs for pediatric obesity. Data from the literature indicate that exergames may offer an interesting impact on childhood obesity and may be considered a potential strategy for controlling weight gain and body composition, promote PA, and decrease time spent on sedentary behavior in children and adolescents with obesity. However, exergame use also has some limits, such as children’s poor self-regulation and poor structuring of exergame use. Therefore, a prudent approach should be maintained, and additional high-quality research is needed to determine if exergames can be effectively used in the treatment of childhood obesity and if new digital media, as a supplementation of exercise rather than a replacement, could be considered to combat sedentary behavior in educational programs for pediatric obesity prevention.

Humans need only 200 ms to generate posture-specific muscle activation patterns for successful vertical jumps in reaction to an auditory trigger.

It is currently unknown how the central nervous system controls ballistic whole-body movements like vertical jumps. Here we set out to study the time frame of generating muscle activation patterns for maximum-effort jumps from different initial postures. We had ten healthy male participants make a slow countermovement from an upright position and initiate a maximal vertical jump as soon as possible following an auditory trigger. The trigger was produced when hip height dropped below one of three preselected values, unknown in advance to the participant, so that the participant was uncertain about the posture from which to initiate the jump. Furthermore, we determined the ensuing bottom postures reached during jumps, and from these postures had the participants perform maximum-effort squat jumps in two conditions: whenever they felt ready, or as soon as possible following an auditory trigger. Kinematics and ground reaction forces were measured, and electromyograms were collected from gluteus maximus, biceps femoris, rectus femoris, vastus lateralis, gastrocnemius and soleus. For each muscle, we detected activation onsets, as well as reaction times defined as the delay between trigger onset and activation onset. In the jumps preceded by a slow countermovement, the posture from which to initiate the jump was unknown before trigger onset. Nevertheless, in these jumps, posture-specific muscle activation patterns were already released within 200 ms after trigger onset and reaction times were not longer and jump heights not less than in squat jumps from corresponding bottom postures. Our findings suggest that the generation of muscle activation patterns for jumping does not start before trigger onset and requires only about 200 ms.

Motion planning in dynamic environments using context-aware human trajectory prediction

Over the years, the separate fields of motion planning, mapping, and human trajectory prediction have advanced considerably. However, the literature is still sparse in providing practical frameworks that enable mobile manipulators to perform whole-body movements and account for the predicted motion of moving obstacles. Previous optimisation-based motion planning approaches that use distance fields have suffered from the high computational cost required to update the environment representation. We demonstrate that GPU-accelerated predicted composite distance fields significantly reduce the computation time compared to calculating distance fields from scratch. We integrate this technique with a complete motion planning and perception framework that accounts for the predicted motion of humans in dynamic environments, enabling reactive and pre-emptive motion planning that incorporates predicted motions. To achieve this, we propose and implement a novel human trajectory prediction method that combines intention recognition with trajectory optimisation-based motion planning. We validate our resultant framework on a real-world Toyota Human Support Robot (HSR) using live RGB-D sensor data from the onboard camera. In addition to providing analysis on a publicly available dataset, we release the Oxford Indoor Human Motion (Oxford-IHM) dataset and demonstrate state-of-the-art performance in human trajectory prediction. The Oxford-IHM dataset is a human trajectory prediction dataset in which people walk between regions of interest in an indoor environment. Both static and robot-mounted RGB-D cameras observe the people while tracked with a motion-capture system.

Estimating Whole-Body Walking Motion from Inertial Measurement Units at Wrist and Heels Using Deep Learning

A recurrent-neural-network-based deep-learning model was developed to estimate the three-axis joint angles of an entire body with 17 bones during walking from three inertial measurement units (IMUs) — one each on the left wrist and heels. In this model, the acceleration and angular velocity of the previous 49 frames and current frame were considered as inputs. The architecture comprises two hidden layers (two long short-term memory layers) and a dense layer. The performance of the model was evaluated using the National Institute of Advanced Industrial Science and Technology (AIST) Gait Database 2019 public dataset. Consequently, the root mean squared error of each joint angle was less than 12.28°. A comparison of the estimation results of the same model with IMUs at the pelvis and shanks revealed that the proposed model is advantageous in terms of balanced measurement accuracy and ease of use in realizing whole-body motion capture. Although the accuracy of the model was better than those of previous models in estimating the general whole-body motion from six IMUs, it was worse than that of a previous model in estimating only the lower-limb motion from three IMUs attached to the pelvis and shanks during walking. In the proposed model, IMUs are attached to the left wrist and heels, and whole-body motion can be easily captured using a smartwatch and smart shoes.

Dual-Arm Mobile Manipulation Planning of a Long Deformable Object in Industrial Installation

In this letter, we present a planning method considering regrasp and mobility for the manipulation problem of a long deformable object in an industrial scenario. The object is carried through designated rollers by a dual-arm robot in the presence of environmental constraints. The problem is formulated by modeling the deformable object as a set of line segments and categorizing obstacles into different types. A hierarchical 2.5D manipulation planner is proposed, which first plans the ideal path of a deformable object in a 2D plane and obtains its topological representation. Next, dual-arm regrasps are determined by utilizing the deformation models, and collision-free paths are planned to meet topological constraints for the grippers. The feasibility of the proposed method is experimentally validated using a humanoid robot, by integrating the planner with a loco-manipulation planning framework and QP-based whole-body motion controller.

Data-Driven Discovery of Novel Motor Biomarkers for Freezing-of-Gait in Parkinsonism from Full-body Kinematics using Artificial Intelligence (P6-11.001)

<h3>Objective:</h3> The purpose of this study is to identify novel phenotypes of freezing of gait using a deep neural network model and whole-body kinematics data. <h3>Background:</h3> Freezing of gait (FOG) in parkinsonism contributes to significant morbidity and is a challenge to treat due to its complex representation. Prior studies explored objectively phenotyping FOG primarily by spectral analyses on kinematics during gait assessments, which lack details to capture complex whole-body movements. In this work, we demonstrate the development of a deep learning model that analyzes complex body movements from whole-body 3D kinematics during time-up-and-go tests (TUG) for individuals with parkinsonism to detect FOG phenotypes. <h3>Design/Methods:</h3> We collected whole-body kinematic marker time series from 2015 to 2017 using a motion capture system from 57 patients, including 5 patients with atypical parkinsonism, assessed with TUG tests in the off and on medication state. FOG score was measured with Movement Disorder Society-Unified Parkinson’s Disease Rating Scale Part III motor score (MDS-UPDRS-III). In our experiment, we designed a deep neural network model to simultaneously predict levodopa medication state (ON/OFF), FOG score (0–4), and MDS-UPDRS-III total score from the collected kinematics data. <h3>Results:</h3> As compared to formal clinical assessments by a movement disorders specialist, our model classified levodopa medication state and FOG score with 96.4% and 96.2% F1 scores respectively and regressed MDS-UPDRS-III total score with root mean square error (RMSE) of 2.7 points. <h3>Conclusions:</h3> This model detects time segments having characteristic movements of FOG during walking, e.g. small shuffling steps, akinesia, and tremulousness. Additional findings demonstrate that FOG is not limited to the lower extremity, and also significantly involves movements in the upper body, further supporting that FOG requires phenotyping using whole-body kinematics. Findings from our analysis may lead to novel hypotheses to define more granular FOG phenotypes. <b>Disclosure:</b> Dr. Kwon has nothing to disclose. Prof. Clifford has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Alivecor. Prof. Clifford has stock in Mindchild Medical. Prof. Clifford has received intellectual property interests from a discovery or technology relating to health care. Prof. Clifford has received publishing royalties from a publication relating to health care. Dr. Esper has received research support from Centers for Disease Control and Prevention. Dr. Esper has received publishing royalties from a publication relating to health care. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Acorda. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Sunovion. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Alterity. Dr. Factor has received personal compensation in the range of $10,000-$49,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Lundbeck. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Biogen. The institution of Dr. Factor has received research support from Impax. The institution of Dr. Factor has received research support from Biohaven. The institution of Dr. Factor has received research support from Sunovion. The institution of Dr. Factor has received research support from Vaccinex. The institution of Dr. Factor has received research support from Voyager. The institution of Dr. Factor has received research support from Neurocrine. The institution of Dr. Factor has received research support from US World Meds. The institution of Dr. Factor has received research support from Addex. The institution of Dr. Factor has received research support from Prelinia. The institution of Dr. Factor has received research support from Medtronics. The institution of Dr. Factor has received research support from Boston Scientific. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. McKay has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Biocircuit Technologies. The institution of Dr. McKay has received research support from NIH. The institution of Dr. McKay has received research support from the McCamish Foundation.

Phenotyping Motor-subtypes of Parkinsonism from Full-body Kinematics using Machine Learning (P6-11.007)

<h3>Objective:</h3> The purpose of this study is to use whole-body kinematics data during walking to accurately identify motor subtypes of Parkinson’s disease (PD) and parkinsonism. <h3>Background:</h3> Two major motor subtypes of PD have been described, Postural Instability and Gait Disorder (PIGD) and Tremor Dominant (TD), which differ in the disease prognosis. Prior studies have explored quantifying motor subtypes via gait analysis, which lacks data for whole-body movements that are required when analyzing an entire gait cycle. In this study, we measure whole-body kinematics while walking, which is then analyzed to classify PIGD vs. TD subtypes using machine learning (ML) models. <h3>Design/Methods:</h3> The whole-body kinematic marker time series were collected from 57 patients (including 5 patients with atypical parkinsonism) while walking in off medication state using a motion capture system from 2015 to 2017. From all marker kinematics data, we extracted various features beyond gait-related features, including time series and spectral features. Then, ML models were used to classify the extracted features into the corresponding motor subtypes assessed with Movement Disorder Society-Unified Parkinson’s Disease Rating Scale (MDS-UPDRS). <h3>Results:</h3> Compared to the clinically assessed motor subtypes, our model classified whole-body kinematics while walking with a 97.0% F1 score, which significantly outperformed the gait-based analysis which showed a 55.8% F1 score. Both upper and lower body movements were found to be useful indicators to distinguish between PIGD and TD. <h3>Conclusions:</h3> Our findings strongly supported our hypothesis that accurately quantifying motor subtypes requires more than gait-related or lower-body movement features that are typically used in previous work. This feature analysis may lead to novel hypotheses for fine-grained upper-body motion phenotypes related to motor subtypes. More importantly, our techniques can help monitor changes in motor subtypes for guiding appropriate intervention through objective quantification of walking movements. <b>Disclosure:</b> The institution of an immediate family member of Mr. Gong has received research support from National Science Foundation. Dr. Kwon has nothing to disclose. Prof. Clifford has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Alivecor. Prof. Clifford has stock in Mindchild Medical. Prof. Clifford has received intellectual property interests from a discovery or technology relating to health care. Prof. Clifford has received publishing royalties from a publication relating to health care. Dr. Esper has received research support from Centers for Disease Control and Prevention. Dr. Esper has received publishing royalties from a publication relating to health care. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Acorda. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Sunovion. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving as a Consultant for Alterity. Dr. Factor has received personal compensation in the range of $10,000-$49,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Lundbeck. Dr. Factor has received personal compensation in the range of $5,000-$9,999 for serving on a Scientific Advisory or Data Safety Monitoring board for Biogen. The institution of Dr. Factor has received research support from Impax. The institution of Dr. Factor has received research support from Biohaven. The institution of Dr. Factor has received research support from Sunovion. The institution of Dr. Factor has received research support from Vaccinex. The institution of Dr. Factor has received research support from Voyager. The institution of Dr. Factor has received research support from Neurocrine. The institution of Dr. Factor has received research support from US World Meds. The institution of Dr. Factor has received research support from Addex. The institution of Dr. Factor has received research support from Prelinia. The institution of Dr. Factor has received research support from Medtronics. The institution of Dr. Factor has received research support from Boston Scientific. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. Factor has received publishing royalties from a publication relating to health care. Dr. McKay has received personal compensation in the range of $500-$4,999 for serving as a Consultant for Biocircuit Technologies. The institution of Dr. McKay has received research support from NIH. The institution of Dr. McKay has received research support from the McCamish Foundation.