Seoyoon Heo

Objective: To systematically review the evidence for orthotic interventions on pain, balance, and plantar pressure distribution in individuals with flexible flatfoot. Methods: Randomized and nonrandomized controlled trials of orthotic interventions for flexible flatfoot were included. Primary outcomes were pain, balance, and plantar pressure. MEDLINE via PubMed, EMBASE via Elsevier, and the Cochrane Library were searched from inception to October 16, 2025. All effect sizes were independently recomputed, and pooling required two or more compatible datasets. Results: Eight studies with 382 participants were included: six randomized controlled trials, one nonrandomized controlled trial, and one within‐subject repeated‐measures study. No domain met the pooling condition, so all previously reported summary estimates and heterogeneity statistics are withdrawn. Pain against a comparator without arch support rested on Yurt et al. ( n = 67): g = −1.02 [95% CI: −1.54, −0.49] on the trial's designated primary outcome, a 100 mm visual analog scale, but g = −0.28 on its Foot Function Index total. In Elsayed et al. ( n = 40) both arms wore the same orthosis, so g = −0.97 [−1.62, −0.33] estimates added exercise, not the device. Balance rested on Yalfani et al. ( n = 30): g = −0.86 [−1.59, −0.13]. In Khodaei et al. ( n = 19), plantar load was redistributed rather than reduced, moving from the heel and lesser metatarsals ( g = −1.93 to −1.09) to the medial midfoot (+1.76). Taspinar et al. were withdrawn as arithmetically irreproducible. Conclusions: Every domain rests on a single small study, so no general efficacy claim is supportable. The most defensible finding is mechanical: load shifts from the heel and lesser metatarsals to the medial midfoot, as a medial arch support is designed to do. The pain estimate is not robust to instrument choice within its own trial. GRADE certainty is very low for every outcome; certainty for plantar pressure began at low because the contributing study did not randomize allocation. The two nonrandomized studies were additionally appraised with ROBINS‐I, at critical and serious risk, respectively. Adequately powered randomized trials with a defined comparator and a pre‐specified primary outcome are required.

There was an error in the funding statement of the original publication

Background: Physical inactivity remains a major public health challenge, particularly for underserved populations lacking exercise facility access. AI-powered smartphone applications with real-time human pose estimation offer scalable solutions, but they lack rigorous clinical validation. Objective: This study validates the clinical efficacy of a 16-week on-device AI-driven resistance training program using MediaPipe pose estimation technology in young adults with limited facility access. Primary outcomes included muscular strength (1RM squat), body composition, functional movement (FMS), and cardiorespiratory fitness (VO₂max). Methods: A single-group pre–post study enrolled 216 participants (mean age 23.77 ± 4.02 years; 69.2% male), with 146 (67.6%) completing the protocol. Participants performed three 30 min weekly sessions of seven compound exercises delivered via a smartphone app providing real-time pose analysis (97.2% key point accuracy, 28.6 ms inference), multimodal feedback, and personalized progression using self-selected equipment. Results: Significant improvements across all domains: muscular strength (+4.39 kg 1RM squat, p < 0.001, d = 1.148), body fat (−2.92%, p < 0.001, d = −1.373), skeletal muscle mass (+2.19 kg, p < 0.001, d = 1.433), FMS (+0.29 points, p = 0.001, d = 0.285), and VO₂max (+1.82 mL/kg/min, p < 0.001, d = 0.917). Pose classification accuracy reached 95.8% vs. physiotherapist assessment (ICC = 0.94). Conclusions: This study provides the first clinical evidence that on-device AI pose estimation enables facility-independent resistance training with outcomes comparable to traditional programs. Unlike cloud-based systems, our lightweight model (28.6 ms inference) supports real-time mobile deployment, advancing accessible precision exercise medicine. Limitations include a single-arm design and gender imbalance, warranting future RCTs with diverse cohorts.

Generative artificial intelligence (AI) has introduced transformative paradigms into education, facilitating personalized, real-time, and interactive learning experiences. This study investigates the integration of AI-based textbooks within regular college curricula, with a specific focus on their application in ancillary engineering subjects that demand high precision. AI textbooks enable customized learning pathways, enhance student engagement through adaptive content, and provide educators with data-driven insights. Employing a mixed-methods approach, this research compares the academic performance and learning experiences of two groups: the Traditional Learning Group (TLG), which utilized printed materials; and the AI Learning Group (ALG), which employed a generative AI-powered textbook based on LLama 3.1. Over a 15-week semester, data were collected through pre- and post-tests, task evaluations, platform log analyses, and satisfaction surveys. The findings reveal no statistically significant differences between the two groups in quantitative measures such as academic achievement, learning time, and overall satisfaction. However, the qualitative assessments underscore the role of AI-based learning in supporting self-directed education, while also highlighting the critical challenges, including digital equity, algorithmic biases, and data privacy concerns. Furthermore, the study emphasizes the necessity of comprehensive educator training, the establishment of ethical frameworks, and the development of scalable implementation strategies to fully leverage the potential of AI textbooks. Future research should prioritize randomized controlled trials (RCTs) to evaluate the long-term impacts of AI-based educational tools and develop adaptive frameworks that balance technological advancements with the emotional and motivational dimensions of human-centered education. This pioneering research lays the groundwork for advancing generative AI in higher education and fostering an innovative and equitable learning ecosystem.

The increasing prevalence of machine learning (ML) and automated machine learning (AutoML) applications across diverse industries necessitates rigorous comparative evaluations of their predictive accuracies under various computational environments. The purpose of this research was to compare and analyze the predictive accuracy of several machine learning algorithms, including RNNs, LSTMs, GRUs, XGBoost, and LightGBM, when implemented on different platforms such as Google Colab Pro, AWS SageMaker, GCP Vertex AI, and MS Azure. The predictive performance of each model within its respective environment was assessed using performance metrics such as accuracy, precision, recall, F1-score, and log loss. All algorithms were trained on the same dataset and implemented on their specified platforms to ensure consistent comparisons. The dataset used in this study comprised fitness images, encompassing 41 exercise types and totaling 6 million samples. These images were acquired from AI-hub, and joint coordinate values (x, y, z) were extracted utilizing the Mediapipe library. The extracted values were then stored in a CSV format. Among the ML algorithms, LSTM demonstrated the highest performance, achieving an accuracy of 73.75%, precision of 74.55%, recall of 73.68%, F1-score of 73.11%, and a log loss of 0.71. Conversely, among the AutoML algorithms, XGBoost performed exceptionally well on AWS SageMaker, boasting an accuracy of 99.6%, precision of 99.8%, recall of 99.2%, F1-score of 99.5%, and a log loss of 0.014. On the other hand, LightGBM exhibited the poorest performance on MS Azure, achieving an accuracy of 84.2%, precision of 82.2%, recall of 81.8%, F1-score of 81.5%, and a log loss of 1.176. The unnamed algorithm implemented on GCP Vertex AI showcased relatively favorable results, with an accuracy of 89.9%, precision of 94.2%, recall of 88.4%, F1-score of 91.2%, and a log loss of 0.268. Despite LightGBM’s lackluster performance on MS Azure, the GRU implemented in Google Colab Pro displayed encouraging results, yielding an accuracy of 88.2%, precision of 88.5%, recall of 88.1%, F1-score of 88.4%, and a log loss of 0.44. Overall, this study revealed significant variations in performance across different algorithms and platforms. Particularly, AWS SageMaker’s implementation of XGBoost outperformed other configurations, highlighting the importance of carefully considering the choice of algorithm and computational environment in predictive tasks. To gain a comprehensive understanding of the factors contributing to these performance discrepancies, further investigations are recommended.

Background: The concept of virtual reality (VR)-based rehabilitation therapy for treating people with low back pain is of growing research interest. However, the effectiveness of such therapy for pain reduction in clinical settings remains controversial. Methods: The present study was conducted according to the reporting guidelines presented in the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement. We searched the PubMed, Embase, CENTRAL, and ProQuest databases for both published and unpublished papers. The Cochrane risk of bias tool (version 2) was used to evaluate the quality of the selected studies. GRADEprofiler software (version 3.6.4) was used to evaluate the level of evidence. We analyzed the included research results using RevMan software (version 5.4.1). Results: We included a total of 11 articles in the systematic review and meta-analysis, with a total of 1761 subjects. Having assessed the quality of these studies, the risk of bias was generally low with high heterogeneity. The results revealed a small to medium effect (standardized mean difference = ±0.37, 95% confidence interval: 0.75 to 0) based on evidence of moderate overall quality. Conclusion: There is evidence that treatment using VR improves patients’ pain. The effect size was small to medium, with the studies presenting evidence of moderate overall quality. VR-based treatment can reduce pain; therefore, it may help in rehabilitation therapy.

Previous studies reported that virtual reality (VR)-based exposure therapy (VRET) was a clinically beneficial intervention for specific phobias. However, among VRET, VR-based graded exposure therapy (VR-GET) is little known about its efficacy on posttraumatic stress disorder (PTSD) symptoms. Therefore, this meta-analysis investigated the effects of VR-GET for PTSD symptoms. A literature search yielded seven randomized controlled trials. The differences between conditions regarding the primary outcome of PTSD symptoms in the effect size of the individual study were calculated using Hedges’ g. The findings showed VR-GET showed a significantly larger effect size for PTSD symptoms (g = 1.100, p = 0.001), compared to controls. However, no significant difference between conventional VRET and controls was found for PTSD symptoms (g = −0.279, p = 0.970). These findings indicated the superiority of VR-GET for PTSD symptoms compared to controls, supporting the importance of immersive PTSD treatments. Nevertheless, the results need to be interpreted with caution due to the substantial number of military service personnel studies. Future trials, considering individually tailored scenarios in virtual environments to cover a wider range of trauma types, are required to investigate its evidence on treating PTSD.

Background: The concept of virtual reality (VR)-based rehabilitation therapy for treating people with low back pain is of growing research interest. However, the effectiveness of such therapy for pain reduction in clinical settings remains controversial. Objective: To assess the effect of VR-based treatment on pain reduction. Methods: The present study was conducted according to the reporting guidelines presented in the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement. We searched the PubMed, Embase, CENTRAL, and ProQuest databases for both published and unpublished papers. The Cochrane risk of bias tool (version 2) was used to evaluate the quality of the selected studies. GRADEprofiler software (version 3.6.4) was used to evaluate the level of evidence. We analyzed the included research results using RevMan software (version 5.4.1). Results: We included a total of 11 articles in the systematic review and meta-analysis, with a total of 1761 subjects. Having assessed the quality of these studies, the risk of bias was generally low with high heterogeneity. The results revealed a small to medium effect (standardized mean difference = - 0.37, 95% confidence interval: - 0.75 to 0) based on evidence of moderate overall quality. Conclusion: There is evidence that treatment using VR improves patients’ pain. The effect size was small to medium with the studies presenting evidence of moderate overall quality. VR-based treatment can reduce pain; therefore, it may help in rehabilitation therapy.Registration Details: PROSPERO CRD42021244051.Funding Information: None received.Declaration of Interests: The authors declare that they have no competing interests.

The purpose of this study was to classify ULTT videos through transfer learning with pre-trained deep learning models and compare the performance of the models. We conducted transfer learning by combining a pre-trained convolution neural network (CNN) model into a Python-produced deep learning process. Videos were processed on YouTube and 103,116 frames converted from video clips were analyzed. In the modeling implementation, the process of importing the required modules, performing the necessary data preprocessing for training, defining the model, compiling, model creation, and model fit were applied in sequence. Comparative models were Xception, InceptionV3, DenseNet201, NASNetMobile, DenseNet121, VGG16, VGG19, and ResNet101, and fine tuning was performed. They were trained in a high-performance computing environment, and validation and loss were measured as comparative indicators of performance. Relatively low validation loss and high validation accuracy were obtained from Xception, InceptionV3, and DenseNet201 models, which is evaluated as an excellent model compared with other models. On the other hand, from VGG16, VGG19, and ResNet101, relatively high validation loss and low validation accuracy were obtained compared with other models. There was a narrow range of difference between the validation accuracy and the validation loss of the Xception, InceptionV3, and DensNet201 models. This study suggests that training applied with transfer learning can classify ULTT videos, and that there is a difference in performance between models.