Motion Signals Research Articles

Multi-level processing of emotions in life motion signals revealed through pupil responses

Perceiving emotions from the movements of other biological entities is critical for human survival and interpersonal interactions. Here, we report that emotional information conveyed by point-light biological motion (BM) triggered automatic physiological responses as reflected in pupil size. Specifically, happy BM evoked larger pupil size than neutral and sad BM, while sad BM induced a smaller pupil response than neutral BM. Moreover, this happy over sad pupil dilation effect is negatively correlated with individual autistic traits. Notably, emotional BM with only local motion features retained could also exert modulations on pupils. Compared with intact BM, both happy and sad local BM evoked stronger pupil responses than neutral local BM starting from an earlier time point, with no difference between the happy and sad conditions. These results revealed a fine-grained pupil-related emotional modulation induced by intact BM and a coarse but rapid modulation by local BM, demonstrating multi-level processing of emotions in life motion signals. Taken together, our findings shed new light on BM emotion processing, and highlight the potential of utilizing the emotion-modulated pupil response to facilitate the diagnosis of social cognitive disorders.

Multi-level processing of emotions in life motion signals revealed through pupil responses.

Perceiving emotions from the movements of other biological entities is critical for human survival and interpersonal interactions. Here, we report that emotional information conveyed by point-light biological motion (BM) triggered automatic physiological responses as reflected in pupil size. Specifically, happy BM evoked larger pupil size than neutral and sad BM, while sad BM induced a smaller pupil response than neutral BM. Moreover, this happy over sad pupil dilation effect is negatively correlated with individual autistic traits. Notably, emotional BM with only local motion features retained could also exert modulations on pupils. Compared with intact BM, both happy and sad local BM evoked stronger pupil responses than neutral local BM starting from an earlier time point, with no difference between the happy and sad conditions. These results revealed a fine-grained pupil-related emotional modulation induced by intact BM and a coarse but rapid modulation by local BM, demonstrating multi-level processing of emotions in life motion signals. Taken together, our findings shed new light on BM emotion processing, and highlight the potential of utilizing the emotion-modulated pupil response to facilitate the diagnosis of social cognitive disorders.

Controllable structure, super adhesion, multifunctional, conductive pressure-sensitive adhesive for electronic skin: Enhancing cohesion by cellulose-derived covalent crosslinking

Ensuring firm adhesion to the skin is essential for improving the detection of temperature and movement signals in electronic skin (E-skin). Self-adhesive eutectogel, an ideal material for E-skin, could act as an innovative pressure-sensitive adhesives (PSAs) that can be easily applied through simple pressing without requiring heating or irradiation. Herein, we synthesized a cellulose-derived polymer crosslinking agent (MCCAC) to create a eutectogel-based conductive PSA (CPSA). The MCCAC formed strong covalent bonds with polymer chains, significantly enhancing the cohesion of the CPSAs. As a result, the CPSAs exhibited high tensile lap-shear strength (99.5 kPa) among self-adhesive eutectogels. Unlike the construction of double-network structures, this method included a straightforward chemical reaction and a cohesion-enhancing strategy, and could adjust crosslinking density by changing the MCCAC content. The preparation process was simple, and the products were controllable. More importantly, this cohesion-enhancing strategy did not compromise the fluidity of polymer chains, which ensured that CPSAs maintain effective interfacial adhesion to adhere on skin, offering exceptional capabilities for capturing body temperature signals (−3.66 %/℃) as well as applications in motion detection and information transmission. This work resolved the cohesion and interfacial adhesion imbalances observed in previous self-adhesive eutectogels and provided a promising approach for designing the novel cellulose-reinforced CPSA for E-skins.

Classification of Co-Manipulation Modus with Human-Human Teams for Future Application to Human-Robot Systems

Despite the existence of robots that can physically lift heavy loads, robots that can collaborate with people to move heavy objects are not readily available. This article makes progress toward effective human-robot co-manipulation by studying 30 human-human dyads that collaboratively manipulated an object weighing \(27 \mathrm{kg}\) without being co-located (i.e., participants were at either end of the extended object). Participants maneuvered around different obstacles with the object while exhibiting one of four modi–the manner or objective with which a team moves an object together–at any given time. Using force and motion signals to classify modus or behavior was the primary objective of this work. Our results showed that two of the originally proposed modi were very similar, such that one could effectively be removed while still spanning the space of common behaviors during our co-manipulation tasks. The three modi used in classification were quickly , smoothly and avoiding obstacles . Using a deep convolutional neural network (CNN), we classified three modi with up to 89% accuracy from a validation set. The capability to detect or classify modus during co-manipulation has the potential to greatly improve human-robot performance by helping to define appropriate robot behavior or controller parameters depending on the objective or modus of the team.

A novel Short-time Fourier transform-based algorithm for detection of wandering in the patients with Alzheimer’s disease

ABSTRACT As the elderly population rises, the growth rate of age-related diseases such as Alzheimer’s disease and cognitive impairments increases. Wandering is one of the first and the most progressive and challenging behaviors; that manifest with certain patterns in the mobility behavior of the patients in the early stages of the disease. Timely diagnosis of wandering can prevent irreparable damages, including the risk of losing the patient, severe physical injuries caused by accidents, and even death. So far, numerous studies have focused on the development of wandering detection algorithms. Most of these mainly rely on two approaches of extracting features from patients’ trajectory history and estimating the path complexity. Motion signal processing methods have rarely been used in this area. An important consideration for these patients is the instability in their movement behaviors, which has received limited attention in previous research, leading to less compatibility of models with this feature. Therefore, this paper proposes an algorithm based on motion signal processing using the Short-time Fourier transform (STFT). This algorithm detects wandering patterns only by using the patient’s trajectory data and changes in their zero and non-zero frequency components. The efficiency of the proposed algorithm has been evaluated using the Geolife open-source dataset, considering the macro-average of metrics such as accuracy, precision, specificity, recall, and F-score, achieving respective values of 96.38%, 94.89%, 96.36%, 96.36%, and 95.58%. The results have validated the proposed algorithm’s strong performance in diagnosing wandering behavior, which, on one hand, helps prevent adverse consequences and, on the other hand, aids in the diagnosis and predicting the progression of the disease to severe Alzheimer’s stage.

Super-Tough, Non-Swelling Zwitterionic Hydrogel Sensor Based on the Hofmeister Effect for Potential Motion Monitoring of Marine Animals.

Hydrogel-based electronic devices in aquatic environments have sparked widespread research interest. Nevertheless, the challenge of developing hydrogel electronics underwater has not been profoundly surmounted because of the fragility and swelling of hydrogels in aquatic environments. In this work, a zwitterionic double network hydrogel comprisedof polyvinyl alcohol (PVA), poly(sulfobetaine methacrylate) (PSBMA), and sulfuric acid (H2SO4) demonstrates super-tough and non-swelling performance. The Hofmeister effect of H2SO4 and PSBMA induces the PVA chains to form numerous nanocrystalline domains, which serve as the primary physical crosslinking points and provide effective energy dissipation. H2SO4 induces a strong salting-out effect to facilitate PVA crystallization and the formation of a dense and stable network structure that inhibits swelling. The resulting hydrogel exhibits an ultra-high toughness of 4.61MJm-3, non-swelling, and long-term stability for up to a month in pure water and seawater. Based on this, a hydrogel-based seawater strain sensor has been developed to monitor the underwater movements of marine animal models. Reliable and stable sensing performance ensures real-time collection of underwater motion signals, despite the impacts of water flow and the interference of ions. This study provides a facile approach to designing super-tough and non-swelling hydrogels and further expands the application of underwater electronic devices.

Visual mental imagery of nonpredictive central social cues triggers automatic attentional orienting

Previous research has demonstrated that social cues (e.g., eye gaze, walking direction of biological motion) can automatically guide people's focus of attention, a well-known phenomenon called social attention. The current research shows that voluntarily generated social cues via visual mental imagery, without being physically presented, can produce robust attentional orienting similar to the classic social attentional orienting effect. Combining a visual imagery task with a dot-probe task, we found that imagining a non-predictive gaze cue could orient attention towards the gazed-at hemifield. Such attentional effect persisted even when the imagery gaze cue was counter-predictive of the target hemifield, and could be generalized to biological motion cue. Besides, this effect could not be simply attributed to low-level motion signal embedded in gaze cues. More importantly, an eye-tracking experiment carefully monitoring potential eye movements demonstrated the imagery-induced attentional orienting effect induced by social cues, but not by non-social cues (i.e., arrows), suggesting that such effect is specialized to visual imagery of social cues. These findings accentuate the demarcation between social and non-social attentional orienting, and may take a preliminary step in conceptualizing voluntary visual imagery as a form of internally directed attention.

Tracking 3D motion of instruments in microsurgery: A comparative study of stereoscopic marker-based vs. deep learning method for objective analysis of surgical skills

Background:In microsurgery, integrating motor and visual information is crucial for achieving precise bimanual maneuvers. Computer vision serves as a useful tool for analyzing surgical maneuvers and enhancing situational awareness in medical assistive systems. Furthermore, the combination of deep learning with computer vision shows promise for objectively assessing surgical skills. Methods:This study compared two methods for tracking 3D movement during microsurgery: Mitracks3D, a stereoscopic tracker using color-marked instruments, and a deep learning method for instruments identification based on YOLOv8. Additionally, drawing an analogy between single-camera and new stereoscopic recording systems, 2D and 3D motion signals from microsurgery training videos on the peg transfer task were obtained to analyze surgeons’ psychomotor skills using six motion analysis parameters (MAPs). Results:When comparing the two 3D tracking methods, the motion signals revealed a relative error (ϵ) of less than 10%, with a correlation coefficient (ρ) exceeding 0.9892 ±0.0020, tracking loss below 2.0398%, and instruments detection time under 1 ms for Mitracks3D and 7.09 ±0.62 ms for YOLOv8. When comparing the performance between participants using 2D and 3D signals, statistically significant differences were found in 4 of the 6 explored MAPs. Additionally, a high cosine similarity (cs>0.99) was found between the performance scores of the 6 MAPs, indicating compatibility between the 2D and 3D recording modalities. Conclusions:The deep learning-based approach is as effective as stereo-tracking with markers; both methods produced remarkably similar 3D motion signals. The compatibility in evaluation between 2D and 3D was confirmed. In essence, adopting computer vision methods such as Mitracks3D and YOLOv8 provides a compelling alternative for constructing objective assessment tools for surgical skills.

Wood derived conductive aerogel with ultrahigh specific surface area and exceptional mechanical flexibility for pressure sensing

Conductive nanocellulose aerogels possess durability and sensitivity by using as piezoresistive sensors. Those aerogels are mostly prepared via a “bottom-up” strategy by using nanocellulose or dissolved and regenerated cellulose as raw materials. However, there is a pressing need to further enhance their sustainability and economic viability. To alleviate these issues, cellulose aerogel can alternatively be made via a “top-down” technique that entails the direct removal of contaminants from naturally porous biomass. Herein, conductive aerogels with exceptional mechanical flexibility and ultrahigh specific surface area were created. Delignified wood was partially dissolved and regenerated to created nanocellulose fibrils with spider web structure inside the pores of wood. After modification with silane coupling agent and Carbon Nanotubes (CNTs), flexible conductive wood aerogel (C-WA) were obtained. The specific surface area of C-WA was as high as 333.82 m2/g, which was 45 times of that of wood. C-WA could maintain its good resilience even after 1000 cycles, showing high compression resilience and excellent fatigue resistance (stress retention rate was 87.22 %, height retention rate was 99.5 %). The utilization of C-WA as a pressure sensor demonstrated remarkable attributes, including ultra-fast response and stability within the designated range. Furthermore, it exhibited exceptional sensing performance even after enduring 2000 stable cycles. The system is capable of real-time, accurate, and efficient detection of human motion signals, showcasing immense potential for future applications in wearable electronic products.

High-strength self-healing multi-functional hydrogels with worm-like surface through hydrothermal-freeze-thaw method

Soft self-healing materials are promising candidates for flexible electronic devices due to their exceptional compatibility, extensibility, and self-restorability. Generally, these materials suffer from low tensile strength and susceptibility to fracture because of the restricted microstructure design. Herein, we propose a hydrothermal-freeze-thaw method to construct high-strength self-healing hydrogels with even interconnected networks and distinctive wrinkled surfaces. The integration of the wrinkled outer surface with the three-dimensional internal network confers the self-healing hydrogel with enhanced mechanical strength. This hydrogel achieves a tensile strength of 223 kPa, a breaking elongation of 442%, an adhesion strength of 57.6 kPa, and an adhesion energy of 237.2 J m-2. Meanwhile, the hydrogel demonstrates impressive self-repair capability (repair efficiency: 93%). Moreover, the density functional theory (DFT) calculations are used to substantiate the stable existence of hydrogen bonding between the PPPBG hydrogel and water molecules which ensures the durability of the PPPBG hydrogel for long-term application. The measurements demonstrate that this multifunctional hydrogel possesses the requisite sensitivity and durability to serve as a strain sensor, which monitors a spectrum of motion signals including subtle vocalizations, pronounced facial expressions, and limb articulations. This work presents a viable strategy for healthcare monitoring, soft robotics, and interactive electronic skins.

Self-powered highly stretchable ferroelectret nanogenerator towards intelligent sports

Ferroelectret nanogenerators (FENGs), recognized for their porous structures that facilitate charge retention, thereby creating giant electric dipoles and exhibiting remarkable piezoelectric properties, are utilized in the development of various flexible transducers. However, despite their flexibility, most developed ferroelectret nanogenerators lack adequate stretchability and satisfactory transverse piezoelectric properties, significantly inhibiting their widespread deployment in wearable or skin-mounted electronics. Here, we introduce a highly stretchable ferroelectret nanogenerator (HS-FENG) built from laser-induced graphene (LIG), Ecoflex and anhydrous ethanol, demonstrating exceptional flexibility and stretchability, along with longitudinal and transverse piezoelectric effects. The stretchability of HS-FENG can reach a record of 468 %, while the quasi-static piezoelectric coefficients d33 and d31 are approximately 120 pC/N and 70 pC/N, respectively. To our knowledge, this is the first demonstration of the developed FENG with remarkably high stretchability. Furthermore, leveraging the performance of the created HS-FENG, we construct a skin-mounted intelligent kinesiology tape capable of effectively monitoring motion signals from human muscles and joints, thereby offering a deeper understanding of movement for users across different levels of physical activity, from professional athletes to individuals undergoing rehabilitation. The development of intelligent kinesiology tape exemplifies the potential of HS-FENG technology in enhancing professional athletic training and personalized healthcare. It contributes to the advancement of inconspicuous skin-mounted biomechanical feedback systems and human-machine interfaces, marking progress in the field.

Genome-Wide Identification of the Remorin Gene Family in Poplar and Their Responses to Abiotic Stresses.

The Remorin (REM) gene family is a plant-specific, oligomeric, filamentous family protein located on the cell membrane, which is important for plant growth and stress responses. In this study, a total of 22 PtREMs were identified in the genome of Populus trichocarpa. Subcellular localization analysis showed that they were predictively distributed in the cell membrane and nucleus. Only five PtREMs members contain both Remorin_C- and Remorin_N-conserved domains, and most of them only contain the Remorin_C domain. A total of 20 gene duplication pairs were found, all of which belonged to fragment duplication. Molecular evolutionary analysis showed the PtREMs have undergone purified selection. Lots of cis-acting elements assigned into categories of plant growth and development, stress response, hormone response and light response were detected in the promoters of PtREMs. PtREMs showed distinct gene expression patterns in response to diverse stress conditions where the mRNA levels of PtREM4.1, PtREM4.2 and PtREM6.11 were induced in most cases. A co-expression network centered by PtREMs was constructed to uncover the possible functions of PtREMs in protein modification, microtube-based movement and hormone signaling. The obtained results shed new light on understanding the roles of PtREMs in coping with environmental stresses in poplar species.

Late feature fusion using neural network with voting classifier for Parkinson’s disease detection

Parkinson’s disease (PD) is classified as a neurological, progressive illness brought on by cell death in the posterior midbrain. Early PD detection will assist doctors in reducing the disease’s consequences. A collection of skilled models that may be applied to regression as well as classification is known as artificial intelligence (AI). PD can be detected using a variety of dataset formats, including text, speech, and picture datasets. For the purpose of classifying Parkinson’s disease, this study suggests merging deep with machine learning recognition approaches. The three primary components of the suggested approach are designed to enhance the accuracy of Parkinson’s disease early diagnosis. These sections cover the topics of categorising, combining, and separating. Convolutional Neural Networks (CNN) as well as attention procedures are used to create feature extractors. The related motion signals are fed to a combination of convolutional neural network and long-short-memory model for feature extraction. Besides, for the classification of patients from non-suffers of Parkinson’s disease, Random Forest, Logistic Regression, Support Vector Machine, Extreme Boot Classifier, and voting classifier were used. Our result shows that for the PD handwriting and related motion datasets, using the proposed CNN with an attention and voting classifier yields 99.95% accuracy, 99.99% precision, 99.98% sensitivity, and 99.95% F1-score. Based on these results, it is warranted to conclude that the proposed methodology of feature extraction from photos of handwriting and relating motor symptoms, fusing of those features, and following it with a voting classifier yields excellent results for PD classification.

An efficient approach for EMG controlled pattern recognition system based on MUAP identification and segregation

An Electromyography (EMG) based pattern recognition system constitutes various steps of signal processing and control engineering from signal acquisition to real-time control. Efficient control of external devices largely depends on the signal processing steps executed before the final output. This work presents a new approach to signal processing using Motor Unit Action Potential (MUAP) based signal decomposition and segmentation. An MUAP is a neurological response during muscle contraction. Due to the higher contact area of surface electrodes, MUAPs from multiple muscles are captured. An MUAP generated from a single muscle usually has identical waveshapes and similar discharging rates and usually lasts for 8–15 ms. These are known as primary MUAPs. The proposed algorithm identifies and uses the primary observed MUAPs for feature extraction and classification. Firstly, noise signals are eliminated by a determined noise margin, which also separates the active muscle movement signals. Next, a novel MUAP identification algorithm is implemented to detect the MUAP trains. Then, identified primary MUAPs are used to make segments with variable widths to extract feature vectors. Based on the correlation score of all the primary MUAPs, the segmentation is performed, which results in segmentation width varying from 110–200 ms. The achieved segmentation width is lesser than the conventional overlapping and non-overlapping methods — the proposed approach results in a 20 to 50% reduction in the segmentation width. Four different classifiers are tested during the machine learning stage to investigate the performance of the proposed approach. The obtained feature sets are then used to train the Linear Discriminant Analysis (LDA), K-Nearest Neighbor (kNN), Decision Tree (DT), and Random Forest (RF) classifiers. The classifiers are tested with precision, recall, F1 score, and accuracy. The kNN and DT classifiers performed better than the LDA and RF classifiers. The maximum precision and recall are 100% while the maximum achieved accuracy is 98.56%. The comparative results show higher accuracy even at lower segmentation widths than the conventional constant window scheme. The kNN and DT classifiers provide a 5% to 15% increment in accuracy compared to the constant window segmentation-based approach.

MXene-based self-adhesive, ultrasensitive, highly tough flexible hydrogel pressure sensors for motion monitoring and robotic tactile sensing

With the rapid development of artificial intelligence and human–computer interaction fields, wearable electronic devices represented by flexible pressure sensors have a wider range of application scenarios. In this paper, we used a composite hydrogel made by one-pot polymerization using a combination of acrylamide (AM), sodium carboxymethyl cellulose (CMC), dopamine (DA), and MXene (Ti3C2Tx). The composite hydrogel has the advantages of self-adhesion, high sensitivity, large detection range (0–500 kPa), low detection limit, high toughness (compression > 90 %), stability (more than 6700 s) and other advantages of hydrogel flexible pressure strain sensor. Sodium carboxymethyl cellulose (CMC), as a multifunctional thickener and gelling agent, significantly improves rheological properties and viscosity in hydrogels, forms crosslinked structures to enhance mechanical properties and provides reliable toughness for pressure sensors. Dopamine (DA), which is rich in catechol groups, can adhere to the material surface through oxidative polymerization to enhance bond strength. Hydrogel pressure sensors for human motion detection can accurately track joint movement, respond sensitively to changes in knocks and touches, and display specific characters such as Morse code. The technology extends to robotic systems, where pressure-activated electrical signals are used to control robot movements. Therefore, it has a very promising application in the field of human–computer interaction, human-body movement signal detection, etc. PAM/CMC/DA/MXene hydrogel pressure sensors have excellent mechanical, electrical, and adhesion properties, which provide a more reliable solution for the fields of flexible wearable e-skin, human–computer interaction, and tactile sensing.

High‐Performance Multipedal Shape Strain Sensors for Human Motion and Electrophysiological Signal Monitoring

AbstractStrain sensors from conducting polymer hydrogel have been widely employed in various wearable devices, electronic skins, and biomedical applications. These sensors provide outstanding flexibility and high sensitivity by integrating conducting polymer with hydrogels, making them particularly suitable for monitoring human motion and physiological signals like heart rate or muscle activity. Despite their extensive application potential, conducting polymer hydrogel face several technical challenges in practical use, including poor mechanical properties, lack of long‐term stability, and difficulty in customizable design. This work introduces a method for fabricating a multipedal strain sensor using poly(3,4‐ethylenedioxythiophene):polystyrene sulfonate (PEDOT:PSS)/polyvinyl alcohol (PVA) dimethyl sulfoxide (DMSO)hydrogels through screen printing and demonstrates its application in human motion monitoring. The multipedal strain sensor demonstrates a low Young's modulus (200 kPa), high stretchability (400%), and excellent mechanical cyclic stability (3000 cycles). Furthermore, this strain sensor is further applied to detect human movements such as chewing, smiling, fist clenching, arm bending, and carotid pulse monitoring. Comparative analysis between the multipedal‐designed sensor and the non‐designed sensor highlights the enhanced sensing capabilities of the multipedal sensor. The design of this multipedal sensor holds the potential to broaden the design concepts for strain sensors and offers new insights for wearable devices and electronic skins.

AI-enhanced backpack with double frequency-up conversion vibration energy converter for motion recognition and extended battery life

With growing interest in physical activities, wearable device sales have been increasing each year due to their help in tracking movement. However, keeping these devices running for a long time is still a key challenge. This paper presents an AI-enhanced backpack equipped with a hybrid vibration energy converter that integrates piezoelectric (PE) and electromagnetic (EM) technologies, featuring a double frequency-up conversion (FUC) mechanism. The PE unit enables efficient motion recognition, while the EM unit extends battery life via kinetic energy harvesting. The double FUC mechanism boosts human motion signal strength and resolution by converting sub-Hertz frequencies to hundreds of Hertz. This makes the device particularly effective in environments with performance constraints and interference. By adopting deep learning techniques, the system maintains a high motion recognition accuracy of 97.33 % even under low data sampling rates and significant noise interference. The EM unit effectively harvests kinetic energy, generating an average output power of 4.5 mW during activities such as running at 4.5 Hz, despite of the device’s compact size. Additionally, a power management circuit with human motion-stimulated switch optimizes battery life by managing the system’s sleep and active states. This innovative energy management approach ensures that during human activity, the system conserves energy by remaining in sleep state for at least 30 % of the time, significantly extending battery life and enhancing the backpack’s suitability for outdoor applications.

Human in the collaborative loop: a strategy for integrating human activity recognition and non-invasive brain-machine interfaces to control collaborative robots.

Human activity recognition (HAR) and brain-machine interface (BMI) are two emerging technologies that can enhance human-robot collaboration (HRC) in domains such as industry or healthcare. HAR uses sensors or cameras to capture and analyze the movements and actions of humans, while BMI uses human brain signals to decode action intentions. Both technologies face challenges impacting accuracy, reliability, and usability. In this article, we review the state-of-the-art techniques and methods for HAR and BMI and highlight their strengths and limitations. We then propose a hybrid framework that fuses HAR and BMI data, which can integrate the complementary information from the brain and body motion signals and improve the performance of human state decoding. We also discuss our hybrid method's potential benefits and implications for HRC.

Multi-step prediction of ship heave motion using transformer-enhanced multi-scale CNN

To achieve precise multi-step heave motion prediction for active compensation control in marine equipment, an innovative approach that integrates an attention-fused multi-scale CNN with a Transformer encoder is introduced in this study. The dual-path CNN and point-wise convolution-enhanced Transformer encoder are designed to capture local features across various scales and the global characteristics of heave motion signals, respectively. Furthermore, a novel optimization objective which combines a slope factor-optimized Huber loss function with a maximum square distance loss function is proposed to better approximate the original signals at extreme points. The proposed model is trained and tested on simulated data under multiple sea conditions. In typical active compensation scenarios (Hs=3.5m), the proposed model demonstrates superior performance compared to baseline methods, achieving an RMSE of 0.0083 m, an MAE of 0.0066 m, and the upper whisker of the box plot for the prediction error is less than 0.02 m. These outcomes effectively satisfy the accuracy requirements for active compensation systems. The generalization test on real-ship collected data demonstrates excellent performance after fine-tuning. Additionally, the model is considered suitable for deployment in real-world applications due to its memory efficiency and fast inference speed.

Characteristics of motion signal profiles of tonic-clonic, tonic, hyperkinetic, and motor seizures extracted from nocturnal video recordings.

In this study, characteristics of signal profiles formed by motion, oscillation, and sound signals were analyzed to evaluate generalizability and variability in a single patient setting (intra-patient variability) and between patients (inter-patient variability). As a secondary objective, the effect of brivaracetam intervention on signal profiles was explored. Patient data included 13 hyperkinetic seizures, 65 tonic seizures, 13 tonic-clonic seizures, and 138 motor seizures from 11 patients. All patients underwent an 8-week monitoring, and after a 3-week baseline, brivaracetam was initiated. Motion, oscillation, and sound features extracted from the video were used to form signal profiles. Variance of signals was calculated, and combined median and quartile visualizations were used to visualize the results. Similarly, the effect of intervention was visualized. Hyperkinetic motion signals showed a rapid increase in motion and sound signals without oscillations and achieved low intra-patient variance. Tonic component created a recognizable peak in motion signal typical for tonic and tonic-clonic seizures. For tonic seizures, inter-patient variance was low. Motor signal profiles were varying, and they did not form a generalizable signal profile. Visually recognizable changes were observed in the signal profiles of two patients. Video-based motion signal analysis enabled the extraction of motion features characteristic for different motor seizure types which might be useful in further development of this system. Tonic component formed a recognizable seizure signature in the motion signal. Hyperkinetic and motor seizures may have not only significantly different motion signal amplitude but also overlapping signal profile characteristics which might hamper their automatic differentiation. Motion signals might be useful in the assessment of movement intensity changes to evaluate the treatment effect. Further research is needed to test generalizability and to increase reliability of the results.