Post-stroke hand gesture recognition via one-shot transfer learning using prototypical networks.

BACKGROUND: In-home rehabilitation systems are a promising, potential alternative to conventional therapy for stroke survivors. Unfortunately, physiological differences between participants and sensor displacement in wearable sensors pose a significant challenge to classifier performance, particularly for people with stroke who may encounter difficulties repeatedly performing trials. This makes it challenging to create reliable in-home rehabilitation systems that can accurately classify gestures. METHODS: Twenty individuals who suffered a stroke performed seven different gestures (mass flexion, mass extension, wrist volar flexion, wrist dorsiflexion, forearm pronation, forearm supination, and rest) related to activities of daily living. They performed these gestures while wearing EMG sensors on the forearm, as well as FMG sensors and an IMU on the wrist. We developed a model based on prototypical networks for one-shot transfer learning, K-Best feature selection, and increased window size to improve model accuracy. Our model was evaluated against conventional transfer learning with neural networks, as well as subject-dependent and subject-independent classifiers: neural networks, LGBM, LDA, and SVM. RESULTS: Our proposed model achieved 82.2% hand-gesture classification accuracy, which was better (P<0.05) than one-shot transfer learning with neural networks (63.17%), neural networks (59.72%), LGBM (65.09%), LDA (63.35%), and SVM (54.5%). In addition, our model performed similarly to subject-dependent classifiers, slightly lower than SVM (83.84%) but higher than neural networks (81.62%), LGBM (80.79%), and LDA (74.89%). Using K-Best features improved the accuracy in 3 of the 6 classifiers used for evaluation, while not affecting the accuracy in the other classifiers. Increasing the window size improved the accuracy of all the classifiers by an average of 4.28%. CONCLUSION: Our proposed model showed significant improvements in hand-gesture recognition accuracy in individuals who have had a stroke as compared with conventional transfer learning, neural networks and traditional machine learning approaches. In addition, K-Best feature selection and increased window size can further improve the accuracy. This approach could help to alleviate the impact of physiological differences and create a subject-independent model for stroke survivors that improves the classification accuracy of wearable sensors. TRIAL REGISTRATION NUMBER: The study was registered in Chinese Clinical Trial Registry with registration number CHiCTR1800017568 in 2018/08/04.

Automated, IMU-based spine angle estimation and IMU location identification for telerehabilitation.

BACKGROUND: Telerehabilitation is a promising avenue for improving patient outcomes and expanding accessibility. However, there is currently no spine-related assessment for telerehabilitation that covers multiple exercises. METHODS: We propose a wearable system with two inertial measurement units (IMUs) to identify IMU locations and estimate spine angles for ten commonly prescribed spinal degeneration rehabilitation exercises (supine chin tuck head lift rotation, dead bug unilateral isometric hold, pilates saw, catcow full spine, wall angel, quadruped neck flexion/extension, adductor open book, side plank hip dip, bird dog hip spinal flexion, and windmill single leg). Twelve healthy subjects performed these spine-related exercises, and wearable IMU data were collected for spine angle estimation and IMU location identification. RESULTS: Results demonstrated average mean absolute spinal angle estimation errors of 2.59 ∘ and average classification accuracy of 92.97%. The proposed system effectively identified IMU locations and assessed spine-related rehabilitation exercises while demonstrating robustness to individual differences and exercise variations. CONCLUSION: This inexpensive, convenient, and user-friendly approach to spine degeneration rehabilitation could potentially be implemented at home or provide remote assessment, offering a promising avenue to enhance patient outcomes and improve accessibility for spine-related rehabilitation. TRIAL REGISTRATION:  No. E2021013P in Shanghai Jiao Tong University.

EchoGest: Soft Ultrasonic Waveguides Based Sensing Skin for Subject-Independent Hand Gesture Recognition.

Gesture recognition is crucial for enhancing human-computer interaction and is particularly pivotal in rehabilitation contexts, aiding individuals recovering from physical impairments and significantly improving their mobility and interactive capabilities. However, current wearable hand gesture recognition approaches are often limited in detection performance, wearability, and generalization. We thus introduce EchoGest, a novel hand gesture recognition system based on soft, stretchable, transparent artificial skin with integrated ultrasonic waveguides. Our presented system is the first to use soft ultrasonic waveguides for hand gesture recognition. EcoflexTM 00-31 and EcoflexTM 00-45 Near ClearTM silicone elastomers were employed to fabricate the artificial skin and ultrasonic waveguides, while 0.1 mm diameter silver-plated copper wires connected the transducers in the waveguides to the electrical system. The wires are enclosed within an additional elastomer layer, achieving a sensing skin with a total thickness of around 500 µ m. Ten participants wore the EchoGest system and performed static hand gestures from two gesture sets: 8 daily life gestures and 10 American Sign Language (ASL) digits 0-9. Leave-One-Subject-Out Cross-Validation analysis demonstrated accuracies of 91.13% for daily life gestures and 88.5% for ASL gestures. The EchoGest system has significant potential in rehabilitation, particularly for tracking and evaluating hand mobility, which could substantially reduce the workload of therapists in both clinical and home-based settings. Integrating this technology could revolutionize hand gesture recognition applications, from real-time sign language translation to innovative rehabilitation techniques.

Graph-Driven Simultaneous and Proportional Estimation of Wrist Angle and Grasp Force via High-Density EMG.

Myoelectric prostheses are generally unable to accurately control the position and force simultaneously, prohibiting natural and intuitive human-machine interaction. This issue is attributed to the limitations of myoelectric interfaces in effectively decoding multi-degree-of-freedom (multi-DoF) kinematic and kinetic information. We thus propose a novel multi-task, spatial-temporal model driven by graphical high-density electromyography (HD-EMG) for simultaneous and proportional control of wrist angle and grasp force. Twelve subjects were recruited to perform three multi-DoF movements, including wrist pronation/supination, wrist flexion/extension, and wrist abduction/adduction while varying grasp force. Experimental results demonstrated that the proposed model outperformed five baseline models, with the normalized root mean square error of 13.2% and 9.7% and the correlation coefficient of 89.6% and 91.9% for wrist angle and grasp force estimation, respectively. In addition, the proposed model still maintained comparable accuracy even with a significant reduction in the number of HD-EMG electrodes. To the best of our knowledge, this is the first study to achieve simultaneous and proportional wrist angle and grasp force control via HD-EMG and has the potential to empower prostheses users to perform a broader range of tasks with greater precision and control, ultimately enhancing their independence and quality of life.

Self-Supervised Learning Improves Accuracy and Data Efficiency for IMU-Based Ground Reaction Force Estimation.

Objective: Recent deep learning techniques hold promise to enable IMU-driven kinetic assessment; however, they require large extents of ground reaction force (GRF) data to serve as labels for supervised model training. We thus propose using existing self-supervised learning (SSL) techniques to leverage large IMU datasets to pre-train deep learning models, which can improve the accuracy and data efficiency of IMU-based GRF estimation. Methods: We performed SSL by masking a random portion of the input IMU data and training a transformer model to reconstruct the masked portion. We systematically compared a series of masking ratios across three pre-training datasets that included real IMU data, synthetic IMU data, or a combination of the two. Finally, we built models that used pre-training and labeled data to estimate GRF during three prediction tasks: overground walking, treadmill walking, and drop landing. Results: When using the same amount of labeled data, SSL pre-training significantly improved the accuracy of 3-axis GRF estimation during walking compared to baseline models trained by conventional supervised learning. Fine-tuning SSL model with 1-10% of walking data yielded comparable accuracy to training baseline model with 100% of walking data. The optimal masking ratio for SSL is 6.25-12.5%. Conclusion: SSL leveraged large real and synthetic IMU datasets to increase the accuracy and data efficiency of deep-learning-based GRF estimation, reducing the need for labeled data. Significance: This work, with its open-source code and models, may unlock broader use cases of IMU-driven kinetic assessment by mitigating the scarcity of GRF measurements in practical applications.

Self-Supervised Learning Improves Accuracy and Data Efficiency for IMU-Based Ground Reaction Force Estimation.

OBJECTIVE: Recent deep learning techniques hold promise to enable IMU-driven kinetic assessment; however, they require large extents of ground reaction force (GRF) data to serve as labels for supervised model training. We thus propose using existing self-supervised learning (SSL) techniques to leverage large IMU datasets to pre-train deep learning models, which can improve the accuracy and data efficiency of IMU-based GRF estimation. METHODS: We performed SSL by masking a random portion of the input IMU data and training a transformer model to reconstruct the masked portion. We systematically compared a series of masking ratios across three pre-training datasets that included real IMU data, synthetic IMU data, or a combination of the two. Finally, we built models that used pre-training and labeled data to estimate GRF during three prediction tasks: overground walking, treadmill walking, and drop landing. RESULTS: When using the same amount of labeled data, SSL pre-training significantly improved the accuracy of 3-axis GRF estimation during walking compared to baseline models trained by conventional supervised learning. Fine-tuning SSL model with 1-10% of walking data yielded comparable accuracy to training baseline model with 100% of walking data. The optimal masking ratio for SSL is 6.25-12.5%. CONCLUSION: SSL leveraged large real and synthetic IMU datasets to increase the accuracy and data efficiency of deep-learning-based GRF estimation, reducing the need for labeled data. SIGNIFICANCE: This work, with its open-source code and models, may unlock broader use cases of IMU-driven kinetic assessment by mitigating the scarcity of GRF measurements in practical applications.

3D Knee and Hip Angle Estimation With Reduced Wearable IMUs via Transfer Learning During Yoga, Golf, Swimming, Badminton, and Dance.

Wearable lower-limb joint angle estimation using a reduced inertial measurement unit (IMU) sensor set could enable quick, economical sports injury risk assessment and motion capture; however the vast majority of existing research requires a full IMU set attached to every related body segment and is implemented in only a single movement, typically walking. We thus implemented 3-dimensional knee and hip angle estimation with a reduced IMU sensor set during yoga, golf, swimming (simulated lower body swimming in a seated posture), badminton, and dance movements. Additionally, current deep-learning models undergo an accuracy drop when tested with new and unseen activities, which necessitates collecting large amounts of data for the new activity. However, collecting large datasets for every new activity is time-consuming and expensive. Thus, a transfer learning (TL) approach with long short-term memory neural networks was proposed to enhance the model's generalization ability towards new activities while minimizing the need for a large new-activity dataset. This approach could transfer the generic knowledge acquired from training the model in the source-activity domain to the target-activity domain. The maximum improvement in estimation accuracy (RMSE) achieved by TL is 23.6 degrees for knee flexion/extension and 22.2 degrees for hip flexion/extension compared to without TL. These results extend the application of motion capture with reduced sensor configurations to a broader range of activities relevant to injury prevention and sports training. Moreover, they enhance the capacity of data-driven models in scenarios where acquiring a substantial amount of training data is challenging.

Design of a Wearable Real-Time Hand Motion Tracking System Using an Array of Soft Polymer Acoustic Waveguides.

Robust hand motion tracking holds promise for improved human-machine interaction in diverse fields, including virtual reality, and automated sign language translation. However, current wearable hand motion tracking approaches are typically limited in detection performance, wearability, and durability. This article presents a hand motion tracking system using multiple soft polymer acoustic waveguides (SPAWs). The innovative use of SPAWs as strain sensors offers several advantages that address the limitations. SPAWs are easily manufactured by casting a soft polymer shaped as a soft acoustic waveguide and containing a commercially available small ceramic piezoelectric transducer. When used as strain sensors, SPAWs demonstrate high stretchability (up to 100%), high linearity (R2 > 0.996 in all quasi-static, dynamic, and durability tensile tests), negligible hysteresis (<0.7410% under strain of up to 100%), excellent repeatability, and outstanding durability (up to 100,000 cycles). SPAWs also show high accuracy for continuous finger angle estimation (average root-mean-square errors [RMSE] <2.00°) at various flexion-extension speeds. Finally, a hand-tracking system is designed based on a SPAW array. An example application is developed to demonstrate the performance of SPAWs in real-time hand motion tracking in a three-dimensional (3D) virtual environment. To our knowledge, the system detailed in this article is the first to use soft acoustic waveguides to capture human motion. This work is part of an ongoing effort to develop soft sensors using both time and frequency domains, with the goal of extracting decoupled signals from simple sensing structures. As such, it represents a novel and promising path toward soft, simple, and wearable multimodal sensors.

IMU Shoulder Angle Estimation: Effects of Sensor-to-Segment Misalignment and Sensor Orientation Error.

Accurate shoulder joint angle estimation is crucial for analyzing joint kinematics and kinetics across a spectrum of movement applications including in athletic performance evaluation, injury prevention, and rehabilitation. However, accurate IMU-based shoulder angle estimation is challenging and the specific influence of key error factors on shoulder angle estimation is unclear. We thus propose an analytical model based on quaternions and rotation vectors that decouples and quantifies the effects of two key error factors, namely sensor-to-segment misalignment and sensor orientation estimation error, on shoulder joint rotation error. To validate this model, we conducted experiments involving twenty-five subjects who performed five activities: yoga, golf, swimming, dance, and badminton. Results showed that improving sensor-to-segment misalignment along the segment's extension/flexion dimension had the most significant impact in reducing the magnitude of shoulder joint rotation error. Specifically, a 1° improvement in thorax and upper arm calibration resulted in a reduction of 0.40° and 0.57° in error magnitude. In comparison, improving IMU heading estimation was only roughly half as effective (0.23° per 1°). This study clarifies the relationship between shoulder angle estimation error and its contributing factors, and identifies effective strategies for improving these error factors. These findings have significant implications for enhancing the accuracy of IMU-based shoulder angle estimation, thereby facilitating advancements in IMU-based upper limb rehabilitation, human-machine interaction, and athletic performance evaluation.

Synthetic EMG Based on Adversarial Style Transfer Can Effectively Attack Biometric-Based Personal Identification Models.

Biometric-based personal identification models are generally considered to be accurate and secure because biological signals are too complex and person-specific to be fabricated, and EMG signals, in particular, have been used as biological identification tokens due to their high dimension and non-linearity. We investigate the possibility of effectively attacking EMG-based identification models with adversarial biological input via a novel EMG signal individual-style transformer based on a generative adversarial network and tiny leaked data segments. Since two same EMG segments do not exist in nature; the leaked data can't be used to attack the model directly or it will be easily detected. Therefore, it is necessary to extract the style with the leaked personal signals and generate the attack signals with different contents. With our proposed method and tiny leaked personal EMG fragments, numerous EMG signals with different content can be generated in that person's style. EMG hand gesture data from eighteen subjects and three well-recognized deep EMG classifiers were used to demonstrate the effectiveness of the proposed attack methods. The proposed methods achieved an average of 99.41% success rate on confusing identification models and an average of 91.51% success rate on manipulating identification models. These results demonstrate that EMG classifiers based on deep neural networks can be vulnerable to synthetic data attacks. The proof-of-concept results reveal that synthetic EMG biological signals must be considered in biological identification system design across a vast array of relevant biometric systems to ensure personal identification security for individuals and institutions.

Challenges and advances in the use of wearable sensors for lower extremity biomechanics.

The use of wearable sensors for the collection of lower extremity biomechanical data is increasing in popularity, in part due to the ease of collecting data and the ability to capture movement outside of traditional biomechanics laboratories. Consequently, an increasing number of researchers are facing the challenges that come with utilizing the data captured by wearable sensors. These challenges include identifying/calculating meaningful measures from unfamiliar data types (measures of acceleration and angular velocity instead of positions and joint angles), defining sensor-to-segment alignments for calculating traditional biomechanics metrics, using reduced sensor sets and machine learning to predict unmeasured signals, making decisions about when and how to make algorithms freely available, and developing or replicating methods to perform basic processing tasks such as recognizing activities of interest or identifying gait events. In this perspective article, we present our own approaches to common challenges in lower extremity biomechanics research using wearable sensors and share our perspectives on approaching several of these challenges. We present these perspectives with examples that come mostly from gait research, but many of the concepts also apply to other contexts where researchers may use wearable sensors. Our goal is to introduce common challenges to new users of wearable sensors, and to promote dialogue amongst experienced users towards best practices.

Real-Time Ground Reaction Force and Knee Extension Moment Estimation During Drop Landings Via Modular LSTM Modeling and Wearable IMUs.

This work investigates real-time estimation of vertical ground reaction force (vGRF) and external knee extension moment (KEM) during single- and double-leg drop landings via wearable inertial measurement units (IMUs) and machine learning. A real-time, modular LSTM model with four sub-deep neural networks was developed to estimate vGRF and KEM. Sixteen subjects wore eight IMUs on the chest, waist, right and left thighs, shanks, and feet and performed drop landing trials. Ground embedded force plates and an optical motion capture system were used for model training and evaluation. During single-leg drop landings, accuracy for the vGRF and KEM estimation was R2 = 0.88 ± 0.12 and R2 = 0.84 ± 0.14, respectively, and during double-leg drop landings, accuracy for the vGRF and KEM estimation was R2 = 0.85 ± 0.11 and R2 = 0.84 ± 0.12, respectively. The best vGRF and KEM estimations of the model with the optimal LSTM unit number (130) require eight IMUs placed on the eight selected locations during single-leg drop landings. During double-leg drop landings, the best estimation on a leg only needs five IMUs placed on the chest, waist, and the leg's shank, thigh, and foot. The proposed modular LSTM-based model with optimally-configurable wearable IMUs can accurately estimate vGRF and KEM in real-time with relatively low computational cost during single- and double-leg drop landing tasks. This investigation could potentially enable in-field, non-contact anterior cruciate ligament injury risk screening and intervention training programs.

Multi-day monitoring of foot progression angles during unsupervised, real-world walking in people with and without knee osteoarthritis.

BACKGROUND: Foot progression angle is a biomechanical target in gait modification interventions for knee osteoarthritis. To date, it has only been evaluated within laboratory settings. METHODS: Adults with symptomatic knee osteoarthritis (n = 30) and healthy adults (n = 15) completed two conditions: 1) treadmill walking in the laboratory (5-min), and 2) real-world walking outside of the laboratory (1-week). Foot progression angle was estimated via shoe-embedded inertial sensing. We calculated the foot progression angle magnitude (median) and variability (interquartile range, coefficient of variation), and used mixed models to compare outcomes between the conditions, participant groups, and disease severities. Reliability was quantified by the intraclass correlation coefficient, standardized error of the measurement, and the minimum detectable change. FINDINGS: Foot progression angle magnitude did not differ between groups or conditions but variability significantly higher in real-world walking (P < 0.001). Structural and symptomatic severity were unrelated to FPA in either walking condition, except for real-world coefficient of variation which was higher for moderate-severe structural osteoarthritis compared to the treadmill for those with mild structural severity (P < 0.034). All real-world outcomes showed excellent reliability including intraclass correlation coefficients above 0.95. The participants recorded a mean (standard deviation) of 298 (33) and 10,447 (5232) steps in the laboratory and real-world walking conditions, respectively. INTERPRETATION: This study provides the first characterization of foot progression angles during real-world walking in people with and without symptomatic knee osteoarthritis. These results indicate that foot progression angles can be feasibly and reliably measured in unsupervised real-world walking conditions.

A scoping review of portable sensing for out-of-lab anterior cruciate ligament injury prevention and rehabilitation.

Anterior cruciate ligament (ACL) injury and ACL reconstruction (ACLR) surgery are common. Laboratory-based biomechanical assessment can evaluate ACL injury risk and rehabilitation progress after ACLR; however, lab-based measurements are expensive and inaccessible to most people. Portable sensors such as wearables and cameras can be deployed during sporting activities, in clinics, and in patient homes. Although many portable sensing approaches have demonstrated promising results during various assessments related to ACL injury, they have not yet been widely adopted as tools for out-of-lab assessment. The purpose of this review is to summarize research on out-of-lab portable sensing applied to ACL and ACLR and offer our perspectives on new opportunities for future research and development. We identified 49 original research articles on out-of-lab ACL-related assessment; the most common sensing modalities were inertial measurement units, depth cameras, and RGB cameras. The studies combined portable sensors with direct feature extraction, physics-based modeling, or machine learning to estimate a range of biomechanical parameters (e.g., knee kinematics and kinetics) during jump-landing tasks, cutting, squats, and gait. Many of the reviewed studies depict proof-of-concept methods for potential future clinical applications including ACL injury risk screening, injury prevention training, and rehabilitation assessment. By synthesizing these results, we describe important opportunities that exist for clinical validation of existing approaches, using sophisticated modeling techniques, standardization of data collection, and creation of large benchmark datasets. If successful, these advances will enable widespread use of portable-sensing approaches to identify ACL injury risk factors, mitigate high-risk movements prior to injury, and optimize rehabilitation paradigms.

A Novel PPG-FMG-ACC Wristband for Hand Gesture Recognition.

Wrist-based hand gesture recognition has the potential to unlock naturalistic human-computer interaction for a vast array of virtual and augmented reality applications. Photoplethysmography (PPG), force myography (FMG), and accelerometry (ACC) have generally been proposed as isolated single sensing modalities for gesture recognition, but any of these alone is inherently limited in the amount of biological information it can collect during finger and hand movements. We thus propose a novel, wrist-based, PPG-FMG-ACC combined sensing approach based on a multi-head attention mechanism fusion convolutional neural network (CNN-AF) for gesture recognition. Nine subjects performed twelve hand gestures involving various wrist and finger postures. Experimental results showed that multi-modal fusion improved classification performance significantly ( p 0.01) compared to any single sensing modality, and the F1-score of the combined PPG-FMG-ACC approach was 40.1% higher than PPG alone, 27.4% higher than ACC alone, and 11.9% higher than FMG alone. To the best of our knowledge, this paper is the first to combine wrist-based PPG, FMG, and ACC signals for hand gesture recognition. These results could serve to inform wrist-based gesture recognition design (e.g., via a smartwatch) and thus expand the capabilities of intuitive and ubiquitous human-machine interaction.

Proposal of a Wearable Multimodal Sensing-Based Serious Games Approach for Hand Movement Training After Stroke.

Stroke often leads to hand motor dysfunction, and effective rehabilitation requires keeping patients engaged and motivated. Among the existing automated rehabilitation approaches, data glove-based systems are not easy to wear for patients due to spasticity, and single sensor-based approaches generally provided prohibitively limited information. We thus propose a wearable multimodal serious games approach for hand movement training after stroke. A force myography (FMG), electromyography (EMG), and inertial measurement unit (IMU)-based multi-sensor fusion model was proposed for hand movement classification, which was worn on the user's affected arm. Two movement recognition-based serious games were developed for hand movement and cognition training. Ten stroke patients with mild to moderate motor impairments (Brunnstrom Stage for Hand II-VI) performed experiments while playing interactive serious games requiring 12 activities-of-daily-living (ADLs) hand movements taken from the Fugl Meyer Assessment. Feasibility was evaluated by movement classification accuracy and qualitative patient questionnaires. The offline classification accuracy using combined FMG-EMG-IMU was 81.0% for the 12 movements, which was significantly higher than any single sensing modality; only EMG, only FMG, and only IMU were 69.6, 63.2, and 47.8%, respectively. Patients reported that they were more enthusiastic about hand movement training while playing the serious games as compared to conventional methods and strongly agreed that they subjectively felt that the proposed training could be beneficial for improving upper limb motor function. These results showed that multimodal-sensor fusion improved hand gesture classification accuracy for stroke patients and demonstrated the potential of this proposed approach to be used as upper limb movement training after stroke.

Predicting knee adduction moment response to gait retraining with minimal clinical data.

Knee osteoarthritis is a progressive disease mediated by high joint loads. Foot progression angle modifications that reduce the knee adduction moment (KAM), a surrogate of knee loading, have demonstrated efficacy in alleviating pain and improving function. Although changes to the foot progression angle are overall beneficial, KAM reductions are not consistent across patients. Moreover, customized interventions are time-consuming and require instrumentation not commonly available in the clinic. We present a regression model that uses minimal clinical data-a set of six features easily obtained in the clinic-to predict the extent of first peak KAM reduction after toe-in gait retraining. For such a model to generalize, the training data must be large and variable. Given the lack of large public datasets that contain different gaits for the same patient, we generated this dataset synthetically. Insights learned from a ground-truth dataset with both baseline and toe-in gait trials (N = 12) enabled the creation of a large (N = 138) synthetic dataset for training the predictive model. On a test set of data collected by a separate research group (N = 15), the first peak KAM reduction was predicted with a mean absolute error of 0.134% body weight * height (%BW*HT). This error is smaller than the standard deviation of the first peak KAM during baseline walking averaged across test subjects (0.306%BW*HT). This work demonstrates the feasibility of training predictive models with synthetic data and provides clinicians with a new tool to predict the outcome of patient-specific gait retraining without requiring gait lab instrumentation.

Reducing Slip Risk: A Feasibility Study of Gait Training with Semi-Real-Time Feedback of Foot-Floor Contact Angle.

Slip-induced falls, responsible for approximately 40% of falls, can lead to severe injuries and in extreme cases, death. A large foot-floor contact angle (FFCA) during the heel-strike event has been associated with an increased risk of slip-induced falls. The goals of this feasibility study were to design and assess a method for detecting FFCA and providing cues to the user to generate a compensatory FFCA response during a future heel-strike event. The long-term goal of this research is to train gait in order to minimize the likelihood of a slip event due to a large FFCA. An inertial measurement unit (IMU) was used to estimate FFCA, and a speaker provided auditory semi-real-time feedback when the FFCA was outside of a 10-20 degree target range following a heel-strike event. In addition to training with the FFCA feedback during a 10-min treadmill training period, the healthy young participants completed pre- and post-training overground walking trials. Results showed that training with FFCA feedback increased FFCA events within the target range by 16% for "high-risk" walkers (i.e., participants that walked with more than 75% of their FFCAs outside the target range) both during feedback treadmill trials and post-training overground trials without feedback, supporting the feasibility of training FFCA using a semi-real-time FFCA feedback system.

Strike index estimation using a convolutional neural network with a single, shoe-mounted inertial sensor.

Strike index is a measurement of the center of pressure position relative to the foot length, and it is regarded as a gold standard in classifying strike pattern in runners. However, strike index requires sophisticated laboratory equipment, e.g., force plates and optical motion capture. We present a method of estimating strike index using data from a shoe-mounted inertial measurement unit (IMU) analyzed by a participant-independent convolutional neural network (CNN), which consists of convolutional, max-pooling, and fully-connected layers. To promote data variability, 16 participants were required to land with three strike patterns (rearfoot, midfoot, and forefoot strike) while running on an instrumented treadmill in four conditions i.e., two footwear types and two running speeds. Using the proposed approach, strike index was estimated with a root mean square error of 6.9% and a R2 of 0.89. Training and testing the model with different variations of the data collected showed that the model was robust to changes in speed. The proposed approach enables accurate estimation of strike index outside of traditional gait laboratories. This solution potentially improves running performance and reduces injury risk in distance runners.