PREFACE: APPLICATION AND PROGRESS OF BIOMECHANICS IN MEDICINE — PART II
Building on previous insights, Part II of this section further illuminates how biomechanical principles are revolutionizing medical science and patient care. This preface sets the stage for the continuation of the topic, highlighting the evolution of biomechanics in medicine and inviting readers to discover new developments and potential directions. Herein are summarized the particulars of the 27 papers that have been selected for inclusion.
Yu Jiao et al. pioneer a single-channel electroencephalogram (EEG) framework for identifying various stress types. Employing tasks like arithmetic operations and memory tests, it extracts multi-domain EEG features from participants under stress. Through rigorous statistical analysis and machine learning, notably using support vector machines, the study achieves high accuracy in stress level classification. Delta, theta, alpha, and full frequency bands prove crucial. This breakthrough paves the way for practical, real-time stress monitoring in clinical and daily life settings.
Innovating medical image segmentation, especially in lung CT scans, Qing Yang et al. integrate U-Net networks with multilevel attention mechanisms, addressing traditional methods’ limitations. Leveraging deep learning, the enhanced model focuses on pertinent features, improving segmentation accuracy and efficiency. Comparative experiments on the 2019nCoVR dataset showcase superior performance and detail enhancement, outperforming conventional techniques in lung nodule analysis, marking a significant advancement in medical diagnostics.
Jieun Park et al. introduce and validate an innovative algorithm utilizing near-infrared light to measure relative blood flow changes. By assessing gradient values with dual-wavelength LED lights, the algorithm demonstrates significant sensitivity to blood concentration fluctuations, crucial for diagnosing Peripheral Vascular Disease. Rigorous testing confirms the system’s efficacy in detecting circulatory alterations, marking a promising step in noninvasive medical diagnostics.
Fujun Han et al. investigate the impact of 10Hz/1.5mT Extremely Low Frequency Electromagnetic Fields (ELF-EMF) on myogenesis in early denervated muscle atrophy stages in SD rats. This study reveals significant benefits. ELF-EMF exposure notably enhances muscle fiber cross-sectional area, upregulates Myf5 mRNA and protein expression, and increases MYOG protein levels. Findings suggest ELF-EMF’s potential in mitigating muscle atrophy effects.
Chenxu Wang et al. focus on motor imagery EEG signal classification, this study innovates with a Takagi–Sugeno–Kang fuzzy system (TSK-FS) model enhanced by low-rank sparse subspace learning (TSK-LSSL). By integrating low-rank and sparse subspace learning, the method optimizes consequent parameter learning, reducing fuzzy rules and redundant parameters. Incorporating a local boundary term, TSK-LSSL excels in mining both global and local data structures, achieving superior classification performance on BCI competition datasets.
Kyu-Beom Kim et al. examine the influence of inaudible theta binaural beats on brain activity, this study reveals that 18kHz baseline frequency paired with 5Hz binaural beats significantly boosts theta wave power in central, parietal, temporal, and occipital brain regions. Utilizing EEG monitoring and power spectrum analysis, findings highlight the capacity of inaudible binaural beats to induce theta waves, independent of auditory perception, offering novel insights into nonsensory neurostimulation techniques.
Yun Gao et al. present M4EEG, a novel matching network-based model utilizing EEG signals for mental health status assessment. Designed for both large-scale supervised learning and few-shot classification, this model leverages a pretrained transformer, decoupling, and cross-connecting features for accurate EEG-based mental health evaluation. Tested on DEAP and AMIGOS datasets, M4EEG showcases superior performance, distinguishing itself in scenarios lacking extensive labeled data, offering a promising tool in mental health monitoring.
Rui Wang et al. aim to enhance feature visibility in vascular medical images, this study introduces a four-step multi-scale Retinex enhancement algorithm. Through block processing, Retinex enhancement, optimized sub-block connection, and gradient filtering, the method effectively sharpens vascular details while minimizing image distortion. Tested on diverse images, the technique outperforms alternatives, significantly boosting information entropy by up to 96.2%. This innovative approach promises enhanced diagnostic accuracy in medical imaging.
Innovating emotion analysis in mental health, Jiayu Han et al. integrate EEG signals and music therapy using a refined transformer model. Addressing the challenge of merging temporal and spatial features, the novel top-k sparse transformer selects key EEG segments for precise emotion detection. Tested on the DEAP dataset, this method outperforms existing models, enhancing sensitivity to emotion-related EEG information and achieving notable classification accuracy improvements across different emotion dimensions. This research bridges a gap in sentiment analysis, offering a promising deep learning application.
In a novel approach, Hyoungjoo Choi et al. integrate stroboscopic glasses (SG) with proprioceptive feedback training (PFT) for athletes with chronic ankle instability (CAI), demonstrating significant improvements in joint position sense, postural stability, and functional performance over eight weeks. The PFT+SG group showed markedly enhanced FAAM-S, IdFAI scores, reduced JPS errors, and fewer giving-way episodes, highlighting the potential of SG to augment rehabilitation and restore athletic performance in CAI patients.
Zongxiang Hu et al. explore the biomechanical impacts of midsole hollow structure in running shoes, this study found that shoes with optimized hollow design (Hollow shoe 2) significantly increased peak impact force timing, reduced maximum and average loading rates, and extended braking, push, and contact times. Notably, Hollow shoe 2 also decreased tibialis anterior muscle activation during the push phase, suggesting complex hollow midsoles offer biomechanical advantages for runners.
Xiaoli Fan et al. examine neck muscle fatigue induced by wearing helmets during motion using electromyography (EMG), aiming to establish a comprehensive evaluation model. Through monitoring EMG signals and subjective feedback from participants running with two helmet types, the research identifies key indicators of neck fatigue, including EMG, root mean square (RMS), mean power frequency (MPF) and Median Frequency (MF) from specific neck muscles. Support Vector Machine (SVM) is employed to create a fatigue classification model with 91.67% accuracy. The findings highlight the trapezius muscle’s high sensitivity to fatigue and validate the effectiveness of EMG in assessing helmet-induced neck fatigue, contributing to future helmet design optimization.
Keke Li et al. introduce a novel depression detection model for college students, employing deep learning networks to analyze facial expressions and physical activity data. Using a transformer model for behavioral features and a multiregional attention network (MRAN) for emotional characteristics, the approach integrates these multimodal insights at a decision-level fusion stage. Experimental validation on a custom dataset reveals the model’s efficacy, thereby facilitating early-stage mental health intervention among students. This innovative tool holds promise for enhancing mental health support services in educational settings.
Zuoliang Liu et al. explore the effect of footwear bending stiffness on lower-limb biomechanics and running efficiency among recreational runners. Using two distinct shoe stiffness levels, researchers found stiffer shoes decreased metatarsophalangeal (MTP) joint motion and work, enhancing running economy without altering other joints’ work. Stiffer soles reduced MTP joint range and velocity, minimizing negative power, suggesting benefits for shoe design targeting improved running performance.
Haiyu Hu et al. study the mechanisms of electro-acupuncture at “Baihuan Shu” (BL 30) and “Huiyang” (BL 35) points in treating chronic prostatitis in rats. Through comparing various groups, including a control, model, electro-acupuncture, inhibitor, and activator group, the research finds that electro-acupuncture improves rat behavior, reduces prostate weights, and modulates the TLR4/NF- pathway proteins. The results suggest electro-acupuncture may alleviate chronic prostatitis symptoms by regulating this inflammatory pathway.
Zixuan Cheng et al. present an intelligent depression detection model that integrates text data and EEG signals using multimodal fusion technology. By combining BERT-TextCNN for text analysis and convolutional neural network (CNN)-long and short-term memory (LSTM) network for EEG processing, the model captures both subjective expressions and objective neural patterns. A weighted fusion strategy merges these modalities, enhancing detection accuracy and robustness, as validated through experiments on a custom dataset. This pioneering approach offers a promising tool for early depression diagnosis, contributing to precision medicine advancements.
Dehong Guo et al. examine the link between body mass index (BMI) and wound healing outcomes, focusing on infection rates, bacterial strains, and healing effects in 160 patients with wounds or fractures. While no significant correlation is found between BMI and infection incidence, a weak connection emerged between BMI and bacterial strain distribution. Wound exposure time, depth, and longer hospital stays are identified as key risk factors for infection. Overall, the study suggests that BMI has a minor impact on wound healing, with other factors playing more prominent roles.
Caimei Chen et al. develop a nomogram to predict acute kidney injury (AKI) risk in severe community-acquired pneumonia (SCAP) patients. Analyzing data from 474 patients, researchers identified age, male gender, chronic renal disease, diabetes, NSAID use, plus high baseline serum creatinine and uric acid levels as independent risk factors. The nomogram, validated with an AUC of 0.811, can aid in assessing AKI likelihood in SCAP patients, thereby facilitating early intervention and management.
Ya-Hui Lin et al. demonstrate an advanced method using a novel multi-W-shaped slit gauge and Taguchi analysis to optimize gamma camera (GC) scan spatial resolution in nuclear medicine. The gauge, adapted from CT scanning, helps quantify SPECT image detail. By varying five acquisition factors and applying Taguchi’s orthogonal array, the research determines the optimal setting for maximum spatial resolution, achieving a minimum detectable difference (MDD) of 6.21mm. This approach, validated by expert evaluations and statistical analysis, promises improved precision in routine diagnostic imaging with GCs.
Jianuo Huang et al. design a novel medical image segmentation model that integrates a transformer with a skipped-features enhancer (SFE) to address the limitations of deep CNNs, particularly in capturing long-range dependencies and mitigating overfitting on small datasets. The model incorporates additional information capturers and a gated multi-head self-attention mechanism to preserve and enhance image details, leading to improved segmentation performance compared to other transformer-based methods. Its effectiveness is demonstrated through evaluations on public medical datasets.
Xiaobing Lu et al. address mental health in college students experiencing learning burnout, utilizing electroencephalography (EEG) signals to develop a detection model. Comparing SVM, k-nearest neighbor (KNN), logistic regression (LR), CNN, long and short-term memory network (LSTM), and CNN-LSTM models, the research finds CNN-LSTM optimal for identifying mental health issues with high accuracy and stability. The work underscores the connection between academic stress, burnout, and mental health, offering a novel, EEG-based early detection method to support timely interventions and student wellbeing.
Zhilin Wang et al. use a novel approach to optimize the end-effector trajectory of a seven-degree-of-freedom biomimetic medical robotic arm using a deep learning method based on the Faster R-CNN network. By replacing conventional search with RPN and utilizing RoI pooling, the method enhances efficiency and reduces impact forces during speed reversals, ensuring smoother trajectories and improved surgical performance. Input data including velocity, angular velocity, acceleration, and angular acceleration contribute to effective optimization outcomes.
Yijie Zhang et al. introduce an enhanced reinforcement learning framework combined with swarm intelligence optimization for accurate heart failure prediction. By integrating the adaptive learning of reinforcement learning with the robust search capabilities of swarm intelligence, the algorithm optimizes decision-making and global search efficiency. Key improvements involve initializing the Q-table and refining reward mechanisms to boost global optimization and prevent local optima. The method effectively predicts heart failure probabilities from patient data, facilitating early intervention and treatment.
Han Li et al. employ multi-rigid body system theory to develop a dynamic analysis model for human bones in figure skating, particularly focusing on waltz jumps. The model characterizes joints with varying degrees-of-freedom and calculates bone centroid displacement. Through experiments with figure skaters, the method captures angle change curves of key joints, offering insights for comparing with standard actions and enhancing scientific training in figure skating.
Yufeng Xie et al. conduct a systematic review and meta-analysis of randomized controlled trials (RCTs) involving 653 patients and find that both moxibustion and fire needle acupuncture effectively treat gouty arthritis (GA), significantly improving total effectiveness and reducing Comprehensive Joint Swelling and Pain. Moxibustion uniquely demonstrated a significant decrease in uric acid levels. However, due to study limitations, further high-quality RCTs are needed to evaluate the effectiveness and safety of moxibustion and fire needle acupuncture in treating GA.
Hao Zang et al. introduce a low-rank sparse representation-based transition subspace learning (LRSRTS) algorithm to enhance epileptic seizure recognition from EEG signals. Addressing inter-domain variability and noise susceptibility, LRSRTS aligns multiple domains via subspace projection and low-rank constraints, while sparse constraints facilitate reconstruction. Imposing nonnegative constraints on the transition subspace, the algorithm establishes a discriminative classifier using source domain samples. Demonstrating promising results on the CHB-MIT dataset, future work will focus on improving efficiency, mitigating negative transfer, and exploring multimodal fusion for enhanced seizure recognition.
Yijie Chen et al. propose pre-trained CNN models, specifically DenseNet201 and ResNet50, to classify ischemic stroke in MRI images, achieving high accuracy, recall, precision, and F1 scores on two datasets. The method involves feature extraction, selection, and classification using SVM with cross-validation. Despite promising results, challenges remain, including limited data, lengthy training times, and the need for models tailored to medical imaging. Future directions include data augmentation, broader application to other brain conditions, and exploration of CT images.
This collection provides a comprehensive overview of the latest findings and innovative methodologies in biomechanics research1,2,8,12,13,17,18,21,22,23,24,25,26,27,28,29,30,31 within the context of medicine. The articles contained herein offer unique insights into the fundamental principles underlying the mechanical behavior of biological tissues, medical imaging technology,3,4,5,6,7,9,10,11,14,15,16,19,20 the application of computational models in solving medical challenges, and their relevance in clinical practice, rehabilitation, and biomedical engineering. Each paper reflects the dedication of its authors to push the boundaries of knowledge in the realm of biomechanics. We hope that this special issue serves as a catalyst for further investigation, collaboration, and translation of scientific discoveries into practical solutions for improving human health.
