The system follows a four-level closed-loop pipeline running from video capture to score output. RGB cameras record motion at 1920 by 1080 resolution at frame rates between 25 and 60 frames per second, with raw clips cut to 6 to 10 seconds and 8 to 12-second action samples extracted. Preprocessing uniformly resamples sequences to 32 or 48 frames, converts them to 224 by 224 frames, and applies a sliding-window strategy with a temporal overlap of 0.25 to 0.50 to organize keyframe tokens. A composite saliency function combines inter-frame difference, joint displacement magnitude, and local feature variance to rank frames, and the top 8 or 12 frames are selected under saliency peaks and temporal spacing constraints. Before any sliding window is generated, each video is assigned an immutable anonymized athlete identifier, a source-video identifier, and a SHA-256 hash, and a group-stratified split with a fixed random seed of 2025 partitions the data by athlete identity, stratified by sport category and grade label. Every derived subclip inherits its parent partition label, and automated integrity checks confirmed zero repeated athlete identifiers, source-video identifiers, or file hashes across the training, validation, and test sets.

Feature extraction uses a Vision Transformer encoder of the ViT-B/16 type. Each frame is divided into 16 by 16 patches, producing 196 patch tokens per frame plus one learnable classification token, and linear projection maps each patch into a 768-dimensional embedding. Spatial and temporal position embeddings are added so the model can represent both where a posture deviation occurs on the body and when it occurs in the sequence. The encoder applies multi-head self-attention with 12 heads across 12 layers, and the final classification token of each frame is averaged to yield a global action feature vector. The backbone was initialized from ImageNet-21K pretrained weights, while the temporal modules and task-specific heads used Xavier initialization.

A Temporal Dependency Enhancement Module captures phase transitions and movement rhythm by decomposing each frame-level feature into three components: a global pose component, a motion-difference component computed between adjacent frames, and a second-order temporal curvature component reflecting rhythmic fluctuation. A sigmoid gate fuses these components into an intermediate representation, after which dependency attention with an exponential temporal decay term weights the contribution of nearby and distant frames. This design encodes action amplitude variation, smooth phase transitions, and the location of rhythmic peaks, capabilities that matter most in multi-phase movements such as the run-up, takeoff, flight, and landing of track-and-field jumps.

A Local Detail Attention Module focuses on fine-grained posture deviations. A 17-keypoint human-pose estimator is applied to every frame, and three body regions are defined: an upper-limb mask from shoulder, elbow, and wrist keypoints, a trunk mask from the shoulder–hip quadrilateral, and a lower-limb mask from hip, knee, ankle, and supporting-foot keypoints. When keypoint confidence falls below 0.50 because of rapid movement or self-occlusion, masks from the previous frame are warped forward by an affine transformation estimated with RANSAC, with propagation limited to three consecutive frames. Each mask is projected onto a 14 by 14 patch grid, and region-level attention compares pooled features against a class-wise prototype built from high-score training samples, with a normative offset derived from the local pose distance. This mechanism directs the model’s attention to the body regions where technical errors actually occur, such as foot placement on the balance beam or racquet-arm angles during a serve.

The scoring head jointly optimizes continuous regression and grade classification. Each of the 3160 videos was independently scored by three certified coaches or judges with more than five years of professional experience, and inter-rater reliability was verified using a two-way random-effects, absolute-agreement, average-measures intraclass correlation coefficient for the 0–100 scores and quadratic-weighted Cohen’s kappa for the ordinal grades before the ratings were averaged. The label system weights four dimensions—motion normativity, completion, stability, and rhythm consistency—at 0.40, 0.25, 0.20, and 0.15, values tuned on the validation set and explicitly not presented as universal judging standards. The composite score maps to five grades from A (90–100) to E (0–59), and the joint loss combines mean squared error for the continuous score, cross-entropy for grade classification, and a ranking-consistency term applied to pairs within the same action subcategory whose true scores differ by more than 5 points.

The dataset comprised 1200 gymnastics videos, 960 track-and-field jumping videos, and 1000 racquet-sport videos covering badminton forehand clears and tennis serves, split 8:1:1 into 2528 training, 316 validation, and 316 held-out test videos. On the held-out test set, the full model achieved a root mean square error of 3.49, a mean absolute error of 2.72, a Grade Accuracy of 84.0 percent, and a Spearman rank correlation of 0.908. Under the same split and local reimplementation protocol, these figures represent modest improvements over included action-recognition and video-Transformer baselines, with overall RMSE decreasing from 3.57 for ViViT-B/16 and Grade Accuracy rising from 83.2 to 84.0 percent. Racquet sports proved easiest to score, with an RMSE of 3.21 and Grade Accuracy of 86.3 percent, while track-and-field jumping remained hardest at 3.85, which the authors attribute to the limits of RGB video in representing takeoff force, flight posture, and landing-buffer information.

Ablation experiments quantify each module’s contribution. Removing the Temporal Dependency Enhancement Module raised the overall RMSE from 3.49 to 4.28, an increase of 0.79 points, although a category-level paired test across the three sport groups yielded a 95 percent confidence interval of −0.03 to 1.55 points and did not reach the conventional 0.05 significance level. Removing the Local Detail Attention Module caused the largest deterioration, with RMSE rising to 4.41 and Grade Accuracy falling to 76.9 percent, a 0.92-point RMSE increase. Removing the joint optimization loss raised RMSE to 3.97 and cut Grade Accuracy to 80.5 percent, an increase of 0.48 points supported by a statistically significant category-level test with a confidence interval of 0.23 to 0.69 points. Stratified error analysis showed that severe occlusion, oblique viewpoints, high-speed actions, and borderline scores between 60 and 79 points all inflated errors, with the borderline band showing the lowest Grade Accuracy at 78.2 percent.

An interpretability analysis compared the model’s regional attention with expert-annotated fault regions across all 316 test videos. In the high-score range, the model weighted the lower-limb region most heavily at 37.2 percent, while in the low-score range upper-limb attention rose to 42.7 percent. Mean spatial overlap between the top 20 percent of model attention and expert-consensus masks reached 81.3 percent for the lower-limb region in gymnastics and 76.9 percent for the upper-limb region in racquet sports, but only 58.7 percent for upper limbs in track-and-field jumping, a value the authors identify as a limitation of the current visual explanation rather than reliable biomechanical localization. The authors emphasize several constraints on their claims: the results support scoring potential only under the current dataset, annotation protocol, and RGB-video setting and do not establish cross-dataset or cross-sport generalizability. On a single NVIDIA A100 GPU, inference took 36.8 milliseconds per 32-frame clip, a hardware-specific benchmark positioned for offline scoring and post-training analysis rather than unrestricted real-time deployment, and future work should add multi-view video, skeleton sequences, kinetic priors, and external validation against official judging records.

Subject of Research: Visual Transformer-based automatic scoring of sports motion quality from RGB video

Article Title: Sports motion scoring algorithms using visual transformers

Article References: Gong, C., Zhang, H., & Gong, L. (2026). Sports motion scoring algorithms using visual transformers. Discover Artificial Intelligence, 6(1), Article 1617. https://doi.org/10.1007/s44163-026-02345-2

Image Credits: AI Generated

DOI: 10.1007/s44163-026-02345-2

Keywords: Vision Transformer, sports motion scoring, RGB video, temporal dependency enhancement, local detail attention, joint optimization, action recognition, movement quality assessment, gymnastics, track-and-field jumping, racquet sports, machine learning

The study focuses on functional mobility, a broad clinical term describing the ability to perform movement-related tasks required in daily life. Walking, rising from a chair, turning, climbing stairs and maintaining balance all depend on more than muscle strength alone. They require the nervous system to integrate information from the eyes, inner ear, muscles, joints and skin, then coordinate an appropriate motor response. Cancer treatment can disrupt this system indirectly through fatigue, neuropathy, pain, reduced physical activity, fear of movement and changes in body awareness. A sensorimotor rehabilitation program targets these interactions rather than treating mobility as a simple measure of power or endurance.

That distinction is central to the trial’s scientific importance. Conventional exercise programs for cancer survivors often emphasize aerobic capacity, resistance training or flexibility. These approaches can be highly effective, but they may not fully address how a person controls movement. A movement quality-focused intervention asks different questions. Is weight transferred smoothly from one leg to the other? Does the trunk remain stable during turning? Can the participant adjust posture when balance is challenged? Are movements symmetrical, economical and appropriately timed? These details can reveal hidden limitations that may not be captured by a basic strength test or a short walking assessment.

In a sensorimotor intervention, participants may be guided through activities that challenge postural control, body awareness, coordination and the ability to adapt to changing conditions. Exercises can involve shifting the center of mass, controlling the pelvis and trunk, stepping in multiple directions, responding to external cues or performing tasks with altered visual input. The technical goal is not simply to make muscles contract more strongly, but to refine communication between the central nervous system and the musculoskeletal system. Repeated practice can help the brain update its internal model of the body, improve anticipatory postural adjustments and make corrective reactions faster and more precise.

The randomized controlled design gives the findings particular weight. By assigning participants to an intervention or comparison condition, researchers can estimate whether changes in functional mobility are linked to the rehabilitation program rather than to spontaneous recovery, increased motivation or the passage of time. A multicenter format adds another layer of relevance because participants are recruited and treated across different clinical locations. If a movement-based protocol produces benefits under these conditions, it may be more likely to function beyond a single specialist clinic. At the same time, multicenter trials must manage differences in therapist training, equipment, patient populations and local treatment routines, making protocol fidelity an important part of interpretation.

For breast cancer survivors, mobility problems can arise through several overlapping pathways. Peripheral neuropathy may reduce the accuracy of information arriving from the feet, making uneven surfaces or low-light environments more difficult to navigate. Shoulder pain or restricted arm movement can alter trunk mechanics and balance strategies. Lymphedema, whether present in the arm or surrounding tissues, may influence comfort and movement confidence. Fatigue can reduce the willingness to engage in demanding activity, while fear of pain or injury can lead to protective movement patterns. Over time, these adaptations may become inefficient, increasing physical effort during ordinary tasks and encouraging further inactivity.

A movement quality approach is therefore potentially valuable because it addresses the organization of movement, not just its visible outcome. Two people may complete the same walking test in the same amount of time while using very different strategies. One may move with stable alignment and adaptable balance; another may rely on excessive trunk motion, stiffened joints or cautious, energy-consuming steps. Detailed assessments of functional mobility can capture some of these differences through timed transfers, walking tasks, turning tests, balance challenges or composite performance batteries. However, the clinical meaning of an improvement depends on whether it is large enough to exceed measurement error and whether it translates into greater independence in everyday life.

The trial also arrives at a moment when cancer rehabilitation is becoming more personalized. Survivors are not a uniform group: age, cancer stage, treatment history, reconstruction status, neuropathy, musculoskeletal health and pre-existing activity levels can all shape rehabilitation needs. A sensorimotor program may be especially relevant for individuals who remain physically active but feel unstable, hesitant or inefficient during movement. It could also complement resistance and aerobic exercise rather than replace them. Strength provides the capacity to generate force, cardiovascular training supports sustained activity, and sensorimotor practice helps the nervous system apply those resources in a coordinated way. The most effective long-term programs may combine all three.

Interpreting the study will require attention to more than whether the intervention group improved statistically. Researchers and clinicians will need to examine the size of the changes, the durability of any benefits, adherence to the sessions and the extent to which improvements were visible in daily function. It will also matter whether the program was safe, acceptable and feasible for survivors coping with treatment-related fatigue or limited access to rehabilitation services. If the findings show meaningful gains in mobility, they could support a broader shift in survivorship care, encouraging clinicians to evaluate balance, coordination and movement strategy alongside pain, strength and aerobic fitness. If the effects are modest or limited to specific subgroups, the results could still help identify who benefits most and which components of sensorimotor training deserve further study.

The broader message is that recovery after breast cancer is not only a biological process; it is also a process of relearning how to inhabit and trust the body. A person may be free of active disease yet continue to navigate uncertainty every time they step off a curb, rise from a chair or turn quickly in a crowded room. By focusing on the quality and adaptability of movement, the multicenter trial brings attention to a dimension of survivorship that is often overlooked. Its significance ultimately rests on whether laboratory and clinic-based gains become safer, more confident and more independent movement in real life. That is the standard by which sensorimotor rehabilitation may prove its most lasting impact.

Subject of Research: Sensorimotor-based, movement quality-focused rehabilitation and functional mobility in breast cancer survivors.

Article Title: Effects of a sensorimotor-based, movement quality-focused intervention on functional mobility in breast cancer survivors: A multicenter randomized controlled trial

Image Credits: AI Generated

Keywords: Breast cancer survivors, cancer rehabilitation, sensorimotor training, movement quality, functional mobility, balance, postural control, randomized controlled trial, survivorship care, physical therapy