Objective To summarize and analyze the efficacy and influencing factors of second allogeneic hematopoietic stem cell transplantation (allo-HSCT) for acute leukemia relapsing after the first allo-HSCT. Methods Clinical data of 17 patients with acute leukemia who underwent second allo-HSCT at Peking University First Hospital from January 2005 to December 2024 were retrospectively analyzed. Results Among the 17 patients, 7 achieved long-term disease-free survival after second transplantation. The median progression-free survival after successful second transplantation was 7 months (range 8 days to 69 months). The relapse fatality was 24%, and the transplant-related fatality was 35%. Conclusions Second transplantation is an effective treatment for relapsed and refractory acute leukemia, but the relapse fatality and transplant-related fatality remain high. Patient age, time of relapse after the first transplantation and disease status before second transplantation are all factors that affect the efficacy of second transplantation. Younger age, late relapse and complete remission of disease before second transplantation are all beneficial for long-term disease-free survival after second transplantation.

Childhood obesity has become a global public health issue. Current research indicates that early life obesogen exposure has emerged as a significant risk factor for childhood obesity. While obesogens have been confirmed to influence the development and progression of childhood obesity through mechanisms such as endocrine disruption and epigenetic programming, controversies remain regarding the establishment of causal relationships, assessment of combined exposures, and validation of transgenerational effects in humans. In recent years, novel approaches including multi-omics technologies, exposome-based analysis, and multigenerational cohort studies have integrated dynamic biomarker monitoring with analyses of social-environmental interactions, offering new perspectives and methodologies for constructing a systematic "exposure-mechanism-outcome" research framework. This article reviews literature from PubMed and Web of Science up to August 2025 on the association between early life obesogen exposure and childhood obesity, summarizing evidence on the health effects of early life obesogen exposure, major exposure pathways and internal exposure assessment, interactions and amplifying effects of social and environmental factors, as well as the biological mechanisms underlying obesogen action. It further examines current research frontiers and challenges, aiming to provide a theoretical foundation for early prevention and precision intervention of childhood obesity.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Objective: Machine learning (ML) has been reported to have better predictive capability than traditional statistical techniques. The aim of this study was to assess the efficacy of ML algorithms and logistic regression (LR) for predicting depressive symptoms during the COVID-19 pandemic. Methods: Analyses were carried out in a national cross-sectional study involving 21,916 participants. The ML algorithms in this study included random forest (RF), support vector machine (SVM), neural network (NN), and gradient boosting machine (GBM) methods. The performance indices were sensitivity, specificity, accuracy, precision, F1-score, and area under the receiver operating characteristic curve (AUC). Results: LR and NN had the best performance in terms of AUCs. The risk of overfitting was found to be negligible for most ML models except for RF, and GBM obtained the highest sensitivity, specificity, accuracy, precision, and F1-score. Therefore, LR, NN, and GBM models ranked among the best models. Conclusion Compared with ML models, LR model performed comparably to ML models in predicting depressive symptoms and identifying potential risk factors while also exhibiting a lower risk of overfitting.

Neuroscience is a significant frontier discipline within the natural sciences and has become an important interdisciplinary frontier scientific field. Brain is one of the most complex organs in the human body, and its structural and functional analysis is considered the “ultimate frontier” of human self-awareness and exploration of nature. Driven by the strategic layout of “China Brain Project”, Chinese scientists have conducted systematic research focusing on “understanding the brain, simulating the brain, and protecting the brain”. They have made breakthrough progress in areas such as the principles of brain cognition, mechanisms and interventions for brain diseases, brain-like computation, and applications of brain-machine intelligence technology, aiming to enhance brain health through biomedical technology and improve the quality of human life. Due to limited understanding and comprehension of neuroscience, there are still many important unresolved issues in the field of neuroscience, resulting in a lack of effective measures to prevent and protect brain health. Therefore, in addition to actively developing new generation drugs, exploring non pharmacological treatment strategies with better health benefits and higher safety is particularly important. Epidemiological data shows that, exercise is not only an indispensable part of daily life but also an important non-pharmacological approach for protecting brain health and preventing neurodegenerative diseases, forming an emerging research field known as motor neuroscience. Basic research in motor neuroscience primarily focuses on analyzing the dynamic coding mechanisms of neural circuits involved in motor control, breakthroughs in motor neuroscience research depend on the construction of dynamic monitoring systems across temporal and spatial scales. Therefore, high spatiotemporal resolution detection of movement processes and movement-induced changes in brain structure and neural activity signals is an important technical foundation for conducting motor neuroscience research and has developed a set of tools based on traditional neuroscience methods combined with novel motor behavior decoding technologies, providing an innovative technical platform for motor neuroscience research. The protective effect of exercise in neurodegenerative diseases provides broad application prospects for its clinical translation. Applied research in motor neuroscience centers on deciphering the regulatory networks of neuroprotective molecules mediated by exercise. From the perspectives of exercise promoting neurogenesis and regeneration, enhancing synaptic plasticity, modulating neuronal functional activity, and remodeling the molecular homeostasis of the neuronal microenvironment, it aims to improve cognitive function and reduce the incidence of Parkinson’s disease and Alzheimer’s disease. This has also advanced research into the molecular regulatory networks mediating exercise-induced neuroprotection and facilitated the clinical application and promotion of exercise rehabilitation strategies. Multidimensional analysis of exercise-regulated neural plasticity is the theoretical basis for elucidating the brain-protective mechanisms mediated by exercise and developing intervention strategies for neurological diseases. Thus,real-time analysis of different neural signals during active exercise is needed to study the health effects of exercise throughout the entire life cycle and enhance lifelong sports awareness. Therefore, this article will systematically summarize the innovative technological developments in motor neuroscience research, review the mechanisms of neural plasticity that exercise utilizes to protect the brain, and explore the role of exercise in the prevention and treatment of major neurodegenerative diseases. This aims to provide new ideas for future theoretical innovations and clinical applications in the field of exercise-induced brain protection.

ObjectiveTo evaluate the therapeutic effect and mechanism of tendon-management and patella-movement therapeutic manipulation in the treatment of patellofemoral arthritis based on imaging evaluation. MethodsTotally 126 patients with patellofemoral arthritis were recruited and divided into a treatment group and a control group according to a randomised numerical table. The control group received routine sodium hyaluronate injection once a week for a total of 5 times； the treatment group received tendon-management and patella-movement therapeutic manipulation three times a week for four weeks. We compared the Western Ontario and McMaster University osteoarthritis index score （WOMAC）， visual analogue scale （VAS）， imaging indicators including patellar external displacement distance， patellofemoral fit angle， lateral patellofemoral angle， and patellofemoral index， and overall effectiveness evaluation between the two groups before and one week after treatment. ResultsThe total effective rate of the treatment group （45/54， 83.33%） was significantly higher than that of the control group （36/54， 66.67%，P＜0.05）. One week after the end of treatment， the VAS scores and WOMAC scores of both groups were lower than those before treatment in the same group （P＜0.01）， and the patellofemoral index and patellofemoral fit angle of the treatment group decreased compared with that of the control group （P＜0.05）. Compared with the pre-treatment， the distance of patellar external displacement， patellofemoral index， and patellofemoral fit angle decreased in the treatment group 1 week after the end of treatment， and the patellofemoral fit angle decreased in the control group （P＜0.05）. ConclusionThe therapeutic manipulation of tendon-management and patella-movement can correct the degree of patellar external displacement， alleviate joint pain symptoms， improve joint function， and achieve the goal of treating patellofemoral arthritis.

ObjectiveBased on functional near-infrared spectroscopy (fNIRS), we investigated the brain activity characteristics of gross motor tasks in children with autism spectrum disorder (ASD) and motor dysfunctions (MDs) to provide a theoretical basis for further understanding the mechanism of MDs in children with ASD and designing targeted intervention programs from a central perspective. MethodsAccording to the inclusion and exclusion criteria, 48 children with ASD accompanied by MDs were recruited into the ASD group and 40 children with typically developing (TD) into the TD group. The fNIRS device was used to collect the information of blood oxygen changes in the cortical motor-related brain regions during single-handed bag throwing and tiptoe walking, and the differences in brain activation and functional connectivity between the two groups of children were analyzed from the perspective of brain activation and functional connectivity. ResultsCompared to the TD group, in the object manipulative motor task (one-handed bag throwing), the ASD group showed significantly reduced activation in both left sensorimotor cortex (SMC) and right secondary visual cortex (V2) (P<0.05), whereas the right pre-motor and supplementary motor cortex (PMC&SMA) had significantly higher activation (P<0.01) and showed bilateral brain region activity; in terms of brain functional integration, there was a significant decrease in the strength of brain functional connectivity (P<0.05) and was mainly associated with dorsolateral prefrontal cortex (DLPFC) and V2. In the body stability motor task (tiptoe walking), the ASD group had significantly higher activation in motor-related brain regions such as the DLPFC, SMC, and PMC&SMA (P<0.05) and showed bilateral brain region activity; in terms of brain functional integration, the ASD group had lower strength of brain functional connectivity (P<0.05) and was mainly associated with PMC&SMA and V2. ConclusionChildren with ASD exhibit abnormal brain functional activity characteristics specific to different gross motor tasks in object manipulative and body stability, reflecting insufficient or excessive compensatory activation of local brain regions and impaired cross-regions integration, which may be a potential reason for the poorer gross motor performance of children with ASD, and meanwhile provides data support for further unraveling the mechanisms underlying the occurrence of MDs in the context of ASD and designing targeted intervention programs from a central perspective.