Objective: To develop and validate a Transformer-based radiomics model for classifying lung adenocarcinoma subtypes from computed tomography imaging data. Methods: We retrospectively collected 289 computed tomography images of lung adenocarcinoma, including adenocarcinoma in situ, minimally invasive adenocarcinoma, and invasive adenocarcinoma. Correlation-based feature analysis was employed and identified 15 optimal radiomic features. A Transformer-based classification model incorporating multi-head attention and position-wise feed-forward Networks was subsequently constructed. Results: The proposed model achieved a training accuracy of 0.98, test accuracy of 0.914, training recall of 0.942, test recall of 0.874, training F1-score of 0.940, test F1-score of 0.871, training area under the curve of 0.99, and test area under the curve of 0.88. Conclusion: This Transformer-based radiomics model effectively classifies lung adenocarcinoma subtypes, aiding early screening, diagnosis, and personalized treatment strategies to improve patient prognosis.

In recent years, gait prediction has gradually become a cutting-edge research direction in the fields of biomechanics and artificial intelligence. Gait prediction technology, which analyzes an individual’s walking patterns to predict future changes, is crucial for the precision of rehabilitation and exoskeleton robot control. This paper reviews the recent research progress in the field of gait prediction, focusing on the multimodal information acquisition methods based on physical sensors and bioelectric signals, as well as the application of machine learning and deep learning algorithms in gait prediction. By analyzing different sensor data fusion strategies, the importance of multimodal information fusion for improving the accuracy of gait prediction is emphasized. Furthermore, this paper introduces the performance of traditional machine learning algorithms such as Support Vector Machine, Random Forest, and Back Propagation Neural Network, as well as deep learning models such as Long Short-Term Memory, Convolutional Neural Network, and Transformer in gait prediction, highlighting the advantages of deep learning in feature extraction and adaptability to complex scenarios. Finally, this paper explores future directions for the development of gait prediction technology, emphasizing improvements in timeliness, accuracy, and personalization to advance exoskeleton robotics and related fields.

Objective: Machine learning can automatically extract valuable insights from vast datasets, predict and classify diseases, and evaluate drug efficacy. To assess the effectiveness of machine learning algorithms in analyzing non-formed components in urine, real medical data were processed and annotated. Methods: Five models, including K-Nearest Neighbors, Decision Trees, Random Forests, Support Vector Machines, and Gaussian distributions,were constructed to quantitatively analyze 12 non-formed urine components, such as vitamin C, white blood cells, and urinary bilirubin. The efficacy of these models was then compared. Results: It was found that the RandomForest model outperformed others, achieving the lowest mean squared error, high recall rate, accuracy, and areaunder the curve. Conclusions: These findings indicate that machine learning offers significant potential for studying non-formed urine components, potentially enhancing the precision and effectiveness of disease detection andproviding valuable support for clinical decision-making.

Pulsed field ablation (PFA), an emerging treatment method for atrial fibrillation, has demonstrated significant potential in arrhythmia therapy. PFA employs high-intensity, short-duration electric fields to induce irreversible electroporation in myocardial cells, disrupting abnormal cardiac rhythms, and restoring normal heart function. This technique exhibits high success rates and low recurrence rates in animal studies and early clinical trials, offering advantages over traditional methods by reducing damage to adjacent structures such as the esophagus, phrenic nerve, and pulmonary veins. This review examines PFA’s application mechanisms, benefits, key operational parameters and the design and safety of related ablation devices. It emphasizes PFA’s potential to enhance both the efficacy and safety of atrial fibrillation treatment and explores future research directions and technological developments.

Forced-damped vibrations are pivotal in various medical applications, significantly contributing to the examination of tissue mechanical properties, development of medical devices, and understanding of biological systems’ complexities. These vibrations represent the dynamic behavior of systems subjected to external forces and damping, where an external force continues to act, and damping determines the rate of energy dissipation. Advanced exploration of damping properties has led to the creation of novel technologies and methods, enhancing our ability to probe and manipulate the complex mechanical dynamics of biological tissues.

Endoscopic submucosal dissection (ESD) is now considered the standard endoscopic resection technique for patients with early gastric cancer. ESD provides a higher rate of complete resection and a lower local recurrence rate. However, ESD results in larger and deeper ulcers, and post-ESD bleeding is a common complication. Bleeding after ESD cannot be completely avoided, especially in patients with large-sized gastric ulcers, those on anticoagulant therapy, and elderly patients. Most bleeding can be controlled by endoscopic hemostatic methods during the procedure, such as the application of hemostatic clips for ulcer closure and hemostatic powder for ulcer shielding. In addition, we also found the potential value of using new materials such as self-assembling peptides for hemostasis. This review first revisits the definition of endoscopic resection of the digestive tract. Then, we discuss post-ESD bleeding and the influence of risk factors such as the location, size, and depth of the surgical lesion, anticoagulant medication use, and the patient's age and lifestyle. Finally, we review the treatment methods for post-ESD bleeding, including intraoperative ulcer closure, ulcer shielding, and the application of thrombin and adrenaline injection.

GPT is demonstrating great potential in the medical field, especially in the diagnosis and treatment of colds. Its application not only improves the efficiency of diagnosis and treatment, but also enhances patient education and promotes the popularization and improvement of medical knowledge. In terms of cold diagnosis and treatment, GPT can automatically and intelligently analyze patients’ symptoms and quickly provide initial diagnostic suggestions, which is conducive to reducing the workload of doctors. Meanwhile, it can also generate easily understandable educational content to help patients gain a deeper understanding of their conditions, treatment plans and preventive measures. This personalized educational approach, combined with interactive learning and multi-channel dissemination, has greatly enhanced the health literacy of patients. GPT also plays an important role in the diagnosis and treatment of colds. Patients can obtain initial diagnostic suggestions by interacting with GPT to describe their symptoms, providing a strong reference for medical treatment. In addition, GPT can also recommend treatment plans based on the patient's condition, including medication, rest and dietary adjustments, etc. For patients with mild symptoms, GPT can also conduct remote monitoring, promptly alert them of changes in their condition, and provide management suggestions. However, the application of GPT in the medical field also faces challenges. Data privacy and security are the top priorities. It is essential to ensure the encryption and desensitization of patient data. Although GPT has certain application potential, its diagnostic accuracy still cannot be compared with that of experienced doctors. In addition, legal and ethical issues cannot be ignored. For instance, medical liability and informed consent of patients need to be further clarified. To address these challenges, it is necessary to enhance data protection, improve the diagnostic accuracy of the GPT model, and conduct reviews in combination with doctors’ experience. At the same time, relevant laws, regulations and ethical norms should be established and improved, key issues should be clarified, supervision and evaluation should be strengthened to ensure the compliant application of GPT in the medical field.

Key Words: generative pre-trained transformer GPT; artificial intelligence; cold

The center for new-quality productive forces in medicine is an institution dedicated to driving innovation and development in the medical field. The center will adopt cutting-edge new qualitative productive forces technologies and methods, such as artificial intelligence, the Internet of Things in medicine, metaverse in medicine and digital human medical GPT, metaverse technology and Internet of Things technology, while integrating new-quality productive forces quality control system, streamlining processes, strengthening supervision, strengthening information management and effect evaluation, and realizing the linkage between patients, general practitioners and medical center experts to improve the quality and efficiency of medical services. In addition, the center will focus on the cultivation and use of virtual and real talents, the management of medical platforms (including clinics and wards), and the management of social and economic efficiency. The vision of this center is to provide a homogeneous medical service platform with patient-centered, meta-medicine as the focus, special diseases as the starting point, and quality control as the guarantee, as well as a corresponding new-quality productive forces medical model, to help achieve the goal of the Healthy China 2030 plan with strong grassroots and wide coverage.

Key Words: artificial intelligence; Internet of Things in medicine; metaverse in medicine; digital human medical GPT; new quality productive forces

Against the backdrop of The Lancet’s international initiative to “bring death back to life,” and in response to the pressing issue of insufficient death education in China, this paper first reviews five representative cases over the past three years in which metaverse-based medical education technologies have empowered death education — namely, brain death diagnosis, breaking bad news, advance care planning and advance directives communication, end-of-life and postmortem care, and terminal communication. Guided by relevant European, American, and Chinese guidelines, the study further proposes four core modules of death education: cognition and values, emotions and psychology, communication and relationships, and culture and behavior. These are structured into a tiered and progressive three-phase framework spanning undergraduate, postgraduate, and continuing medical education. Finally, the paper elaborates on the empowerment pathways of metaverse medical education technologies — including VR, AR, MR, and AI-enhanced virtual patients — to promote the localization, systematization, and sustainable high-quality development of death education in China through collective academic effort.

Key Words: death education; metaverse medical education technology; tiered progression; empowerment pathways; localization

With the rapid development of information technology, the Metaverse, as an emerging concept of virtual worlds, is gradually showing immense potential in medical education and clinical training. Particularly in the clinical reasoning training of resident physicians, traditional training models face numerous challenges, such as limited clinical resources, complex cases, and restrictions on training time and venues. Metaverse technologies has opened new pathways for resident physician training. This study explores how to leverage metaverse-related technologies, such as virtual reality (VR), augmented reality (AR), and artificial intelligence (AI), to build an immersive and interactive clinical reasoning training platform. By simulating various complex clinical scenarios, resident physicians can engage in repeated training in a risk-free environment, enhancing their flexibility in clinical thinking and their ability to respond effectively. The metaverse platform offers real-time feedback and intelligent analysis to help esident physicians  quickly adjust learning strategies and optimize diagnostic and treatment thinking, and provides more practical opportunities for them.

Key Words: resident training; metaverse; cognitive training