Abstract:The loss of bilateral hand function is a debilitating challenge for millions of individuals that suffered a motor-complete spinal cord injury (SCI). We have recently demonstrated in eight tetraplegic individuals the presence of highly functional spared spinal motor neurons in the extrinsic muscles of the hand that are still capable of generating proportional flexion and extension signals. In this work, we hypothesized that an artificial intelligence (AI) system could automatically learn the spared electromyographic (EMG) patterns that encode the attempted movements of the paralyzed digits. We constrained the AI to continuously output the attempted movements in the form of a digital hand so that this signal could be used to control any assistive system (e.g., exoskeletons, electrical stimulation). We trained a convolutional neural network using data from 13 uninjured (control) participants and 8 motor-complete tetraplegic participants to study the latent space learned by the AI. Our model can automatically differentiate between eight different hand movements, including individual finger flexions, grasps, and pinches, achieving a mean accuracy of 98.3% within the SCI group. Moreover, the model could distinguish with 100% accuracy whether a participant had an injury or not, and it could also facilitate proportional control of certain movements after the injury. Analysis of the latent space of the model revealed that proportionally controllable movements exhibited an elliptical path, while movements lacking proportional control followed a chaotic trajectory. We found that proportional control of a movement can only be correctly estimated if the latent space embedding of the movement follows an elliptical path (correlation = 0.73; p < 0.001). These findings emphasize the reliability of the proposed system for closed-loop applications that require an accurate estimate of the spinal cord motor output.

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is: how to use artificial intelligence systems to automatically identify and decode the patterns of electrical activity in the hand muscles retained in patients with complete spinal cord injury (SCI) in order to achieve precise control of paralyzed finger movements. Specifically, the researchers hope to develop an artificial intelligence system that can automatically learn from these retained electromyographic signals and predict specific hand movements. The successful application of this system can significantly improve the daily living abilities of SCI patients, especially for those who have lost hand function. ### Main research contents: 1. **Background and motivation**: - Complete spinal cord injury leads to the loss of hand function, affecting the quality of life of millions of patients. - Research has found that even in patients with complete spinal cord injury, there are still some retained spinal cord motor neurons, which can generate proportional flexion and extension signals. - This paper hypothesizes that by using an artificial intelligence system, the attempted movement patterns can be automatically learned and decoded from these retained electromyographic signals. 2. **Methods**: - Use a convolutional neural network (CNN) to train the model. The data comes from 13 healthy control group participants and 8 complete spinal cord injury patients. - Record the forearm electromyographic signals of the participants through surface electromyography (sEMG), and at the same time ask them to attempt to perform a series of pre - recorded hand movements. - Use UMAP (Uniform Manifold Approximation and Projection) technology to project the high - dimensional latent space onto a two - dimensional space for easy analysis and visualization. 3. **Results**: - The model can distinguish 8 different hand movements, including single - finger flexion, grasping and pinching actions, with an average accuracy rate of 98.3% for the SCI group. - The model can distinguish whether the participants are injured with 100% accuracy, and can achieve proportional control of certain movements after injury. - Latent space analysis shows that proportionally controllable movements present an elliptical path in the two - dimensional projection, while movements lacking proportional control show chaotic trajectories. - The study found that only when the movement path of the latent space embedding is elliptical can the proportional control be correctly estimated (correlation = 0.73; p < 0.001). ### Conclusion: This study shows the possibility of using artificial intelligence systems to automatically learn and decode hand movements from retained electromyographic signals, which is of great significance for the rehabilitation of patients with complete spinal cord injury. By analyzing the structure of the latent space, researchers can better understand how these retained motor neurons encode movement signals, thus providing new ideas for future rehabilitation technologies.

Identification of Spared and Proportionally Controllable Hand Motor Dimensions in Motor Complete Spinal Cord Injuries Using Latent Manifold Analysis

Identification of Spared and Proportionally Controllable Hand Motor Dimensions in Motor Complete Spinal Cord Injuries Using Latent Manifold Analysis

Sensing and decoding the neural drive to paralyzed muscles during attempted movements of a person with tetraplegia using a sleeve array

Assisted Grasping in Individuals with Tetraplegia: Improving Control through Residual Muscle Contraction and Movement

Offline and online myoelectric pattern recognition analysis and real-time control of a robotic hand after spinal cord injury

Decoding of unimanual and bimanual reach-and-grasp actions from EMG and IMU signals in persons with cervical spinal cord injury

Prediction of imagined single-joint movements in a person with high-level tetraplegia

A direct spinal cord–computer interface enables the control of the paralysed hand in spinal cord injury

Unsupervised decoding of spinal motor neuron spike trains for estimating hand kinematics following targeted muscle reinnervation

MyoGestic: EMG Interfacing Framework for Decoding Multiple Spared Degrees of Freedom of the Hand in Individuals with Neural Lesions

Determining grasp selection from arm trajectories via deep learning to enable functional hand movement in tetraplegia

Towards non-invasive brain-computer interface for hand/arm control in users with spinal cord injury

Implantable brain–computer interface for neuroprosthetic-enabled volitional hand grasp restoration in spinal cord injury

Proportional and Simultaneous Real-Time Control of the Full Human Hand From High-Density Electromyography

Towards unlocking motor control in spinal cord injured by applying an online EEG-based framework to decode motor intention, trajectory and error processing

Body machine interface: remapping motor skills after spinal cord injury

Real-Time Control of an Exoskeleton Hand Robot with Myoelectric Pattern Recognition

Simultaneous and Proportional Control of Wrist and Hand Movements Based on a Neural-Driven Musculoskeletal Model.

A deep learning-based approach for unbiased kinematic analysis in CNS injury

Intramuscular EMG-Driven Musculoskeletal Modelling: Towards Implanted Muscle Interfacing in Spinal Cord Injury Patients

Reconnecting the Hand and Arm to the Brain: Efficacy of Neural Interfaces for Sensorimotor Restoration After Tetraplegia