Predicting Free Achilles Tendon Strain From Motion Capture Data Using Artificial Intelligence

IEEE Transactions on Neural Systems and Rehabilitation Engineering, Год журнала: 2023, Номер 31, С. 3086 - 3094

Опубликована: Янв. 1, 2023

The Achilles tendon (AT) is sensitive to mechanical loading, with appropriate strain improving tissue and material properties. Estimating free AT currently possible through personalized neuromusculoskeletal (NMSK) modeling; however, this approach time-consuming requires extensive laboratory data. To enable in-field assessments, we developed an artificial intelligence (AI) workflow predict during running from motion capture Ten keypoints commonly used in pose estimation algorithms (e.g., OpenPose) were synthesized data noise was added represent real-world obtained using video cameras. Two AI workflows compared: (1) a Long Short-Term Memory (LSTM) neural network that predicted directly (called LSTM only workflow); (2) force which subsequently converted force-strain curve LSTM+ workflow). models trained evaluated estimates of validated NMSK model curve. effect different input features (position, velocity, acceleration keypoints, height mass) on predictions also assessed. significantly improved the compared (p<0.001). best positions velocities as well mass participants input, average time-series root mean square error (RMSE) 1.72±0.95% r ² 0.92±0.10, peak RMSE 2.20% 0.54. In conclusion, showed feasibility predicting accurate low fidelity

Язык: Английский

---

Estimation of Kinetics Using IMUs to Monitor and Aid in Clinical Decision-Making during ACL Rehabilitation: A Systematic Review

Sensors, Год журнала: 2024, Номер 24(7), С. 2163 - 2163

Опубликована: Март 28, 2024

After an ACL injury, rehabilitation consists of multiple phases, and progress between these phases is guided by subjective visual assessments activities such as running, hopping, jump landing, etc. Estimation objective kinetic measures like knee joint moments GRF during assessment can help physiotherapists gain insights on loading tailor protocols. Conventional methods deployed to estimate kinetics require complex, expensive systems are limited laboratory settings. Alternatively, algorithms have been proposed in the literature from kinematics measured using only IMUs. However, knowledge about their accuracy generalizability for patient populations still limited. Therefore, this article aims identify available estimation parameters IMUs evaluate applicability through a comprehensive systematic review. The papers identified search were categorized based modelling techniques interest, subsequently compared accuracies achieved patients rehabilitation. exhibited potential estimating with good accuracy, particularly sagittal movements healthy cohorts. several shortcomings future directions improvement proposed, including extension accommodate multiplanar validation diverse particular population.

Язык: Английский

---

Application of Machine Learning Methods to Investigate Joint Load in Agility on the Football Field: Creating the Model, Part I

Sensors, Год журнала: 2024, Номер 24(11), С. 3652 - 3652

Опубликована: Июнь 5, 2024

Laboratory studies have limitations in screening for anterior cruciate ligament (ACL) injury risk due to their lack of ecological validity. Machine learning (ML) methods coupled with wearable sensors are state-of-art approaches joint load estimation outside the laboratory athletic tasks. The aim this study was investigate ML predicting knee loading during sport-specific agility We explored possibility high and low abduction moments (KAMs) from kinematic data collected a setting through actual KAM kinematics. Xsens MVN Analyze Vicon motion analysis, together Bertec force plates, were used. Talented female football (soccer) players (n = 32, age 14.8 ± 1.0 y, height 167.9 5.1 cm, mass 57.5 8.0 kg) performed unanticipated sidestep cutting movements (number trials analyzed 1105). According findings technical note, classification models that identify exhibiting or preferable ones predict peak magnitude. classifying versus KAMs good approximation (AUC 0.81–0.85) represents step towards testing an ecologically valid environment.

Язык: Английский

---

Deep-learning model for the prediction of lower-limb joint moments using single inertial measurement unit during different locomotive activities

Biomedical Signal Processing and Control, Год журнала: 2023, Номер 86, С. 105372 - 105372

Опубликована: Авг. 21, 2023

Язык: Английский

---

Improving Biological Joint Moment Estimation During Real-World Tasks With EMG and Instrumented Insoles

IEEE Transactions on Biomedical Engineering, Год журнала: 2024, Номер 71(9), С. 2718 - 2727

Опубликована: Апрель 15, 2024

Objective: Real-time measurement of biological joint moment could enhance clinical assessments and generalize exoskeleton control. Accessing moments outside laboratory settings requires harnessing non-invasive wearable sensor data for indirect estimation. Previous approaches have been primarily validated during cyclic tasks, such as walking, but these methods are likely limited when translating to non-cyclic tasks where the mapping from kinematics is not unique. Methods: We trained deep learning models estimate hip knee kinematic sensors, electromyography (EMG), simulated pressure insoles a dataset including 10 18 activities. assessed estimation error on combinations modalities both activity types. Results: Compared kinematics-only baseline, adding EMG reduced RMSE by 16.9% at 30.4% (p<0.05) 21.7% 33.9% (p<0.05). Adding 32.5% 41.2% which was significantly higher than either modality individually All additions improved model performance more Conclusion: These results demonstrate that kinetic information through or improves jointly. additional most important reflect variable sporadic nature real-world. Significance: Improved task generalization pivotal developing robotic systems capable enhancing mobility in everyday life.

Язык: Английский

---

Inverse distance weighting to rapidly generate large simulation datasets

Journal of Biomechanics, Год журнала: 2023, Номер 158, С. 111764 - 111764

Опубликована: Авг. 9, 2023

Язык: Английский

---

Wearable sensor and machine learning estimate tendon load and walking speed during immobilizing boot ambulation

Scientific Reports, Год журнала: 2023, Номер 13(1)

Опубликована: Окт. 23, 2023

Abstract The purpose of this study is to develop a wearable paradigm accurately monitor Achilles tendon loading and walking speed using sensors that reduce subject burden. Ten healthy adults walked in an immobilizing boot under various heel wedge conditions (30°, 5°, 0°) speeds. Three-dimensional motion capture, ground reaction force, 6-axis inertial measurement unit (IMU) signals were collected. We used Least Absolute Shrinkage Selection Operator (LASSO) regression predict peak load speed. effects altering sensor parameters also explored. Walking models (mean absolute percentage error (MAPE): 8.81 ± 4.29%) outperformed (MAPE: 34.93 26.3%). Models trained with subject-specific data performed better than without data. Removing the gyroscope, decreasing sampling frequency, combinations did not change usability models, having inconsequential on model performance. developed simple monitoring uses LASSO (MAPE ≤ 12.6%) while ambulating boot. This provides clinically implementable strategy longitudinally patient activity recovering from injuries.

Язык: Английский

---

Estimation of Ground Reaction Forces during Sports Movements by Sensor Fusion from Inertial Measurement Units with 3D Forward Dynamics Model

Опубликована: Фев. 29, 2024

Rotational jumps are crucial techniques in sports competitions. Estimating ground reaction forces (GRFs), one of the components constituting jumps, through a biomechanical model-based approach enables analysis even environments where force plates or machine learning training data cannot be utilized. In this study, rotational jump movements involving twists on land were measured using inertial measurement units (IMUs) and estimated GRFs body loads 3D forward dynamics model. Our estimation method, based optimization calculations, generated optimized cost functions defined by motion measurements internal loads. To reduce influence dynamic acceleration calculation, orientation sensor fusion composed angular velocity obtained from IMUs an extended Kalman filter was estimated. As result, generating function, it possible to calculate biomechanically valid while following if not all joints covered IMUs. This method allows for independent conditions data.

Язык: Английский

---

Estimation of Ground Reaction Forces during Sports Movements by Sensor Fusion from Inertial Measurement Units with 3D Forward Dynamics Model

Sensors, Год журнала: 2024, Номер 24(9), С. 2706 - 2706

Опубликована: Апрель 24, 2024

Rotational jumps are crucial techniques in sports competitions. Estimating ground reaction forces (GRFs), a constituting component of jumps, through biomechanical model-based approach allows for analysis, even environments where force plates or machine learning training data would be impossible. In this study, rotational jump movements involving twists on land were measured using inertial measurement units (IMUs), and GRFs body loads estimated 3D forward dynamics model. Our optimization calculation-based estimation method generated optimized cost functions defined by motion measurements internal loads. To reduce the influence dynamic acceleration calculation, we orientation sensor fusion, comprising angular velocity from IMUs an extended Kalman filter. As result, generating function-based movements, could calculate biomechanically valid while following if not all joints covered IMUs. The developed study condition- data-independent analysis.

Язык: Английский

---

Trends in real-time artificial intelligence methods in sports: a systematic review

Journal Of Big Data, Год журнала: 2024, Номер 11(1)

Опубликована: Окт. 26, 2024

Abstract This review focuses on the usage of machine learning methods in sports. It closely follows PRISMA framework for writing systematic reviews. We introduce broader field using sensor data feedback sport and cite similar reviews, that focus other aspects field. With its models use signals from simple sensors, this covers a very focused area has not yet been covered by any review. As described problem definition, we well-defined inclusion criteria, have reviewed 72 papers. They present existing solutions, to extract useful information collected various sensors To be included, papers had during sports, sports-related applications result some can used real-time. found is rapidly developing as 46 included were last four years. Furthermore, moving classical techniques deep learning. analyze which input learning, find most commonly accelerometer, followed gyroscope. The common platform single wearable sensor, however, studies multiple often. Dataset sizes sports are relatively small compared fields, but datasets average slightly larger than those do not. preprocessing low-pass filtering feature extraction used. compare different results tested same data, where proved better Most show classification accuracy over 90%, showing tool researched problems. end researching how far implemented. Twenty their beyond research paper provided sort back athletes or coaches. After completing field, propose solution – plan future research. proposed combination best practices implemented further elaborate, see current state conclude article with short summary findings.

Язык: Английский

---