Zhejiang da xue xue bao. Yi xue ban = Journal of Zhejiang University. Medical sciences最新文献

Objectives: To systematically evaluate the performance of generative artificial intelligence (GenAI) models, DeepSeek-V3 and the Qwen3 series, in the differential diagnosis of weight loss.

Methods: A search was conducted in the PubMed database for all case reports published in the American Journal of Case Reports between January 1, 2012 and June 2, 2025, containing the term "weight loss" in the title or abstract. Two senior general practitioners independently verified and assessed whether each case met the diagnostic criteria for weight loss (emaciation). Cases that did not meet these criteria, had incomplete information, or fell within the scope of clearly defined specialized diagnoses and treatments were excluded. The remaining cases were then compiled into standardized clinical case summaries. These summaries were presented to DeepSeek-V3 and the Qwen3 series models (Qwen3-235B-A22B, Qwen3-30B-A3B, and Qwen3-32B) to generate ranked lists of the top 10 differential diagnoses. The models were not specifically fine-tuned for this task. Sensitivity, precision, and F1-score were used to evaluate performance. Intergroup comparisons were performed using McNemar's test and Cochran's Q test.

Results: A total of 87 case were analyzed. For DeepSeek-V3, the sensitivity for Top1, Top5, and Top10 diagnoses was 26.44%, 56.32%, and 65.52%, respectively, with corresponding precision values of 26.44%, 11.26%, and 6.55%. For Qwen3-235B-A22B, the sensitivity values were 21.84%, 43.68%, and 59.77%, with corre-sponding precision values of 21.84%, 8.74%, and 5.98%. DeepSeek-V3 demonstrated significantly better performance than Qwen3-235B-A22B in sensitivity, precision, and F1-score at the Top5 level (P=0.043). Among the Qwen3 series models, Qwen3-235B-A22B showed the best performance in sensitivity, precision, and F1-score for the Top1 diagnosis, outperforming Qwen3-32B and Qwen3-30B-A3B. However, the differences among the three Qwen3 models across all diagnostic levels were not statistically significant (all P>0.05).

Conclusions: Domestic GenAI models exhibit a characteristic of "breadth over precision" in the differential diagnosis of weight loss, with DeepSeek-V3 performing better at key diagnostic levels. Although the sensitivity and precision for the top-ranked diagnosis require improvement, these models can serve as effective clinical decision support tools, broadening the diagnostic perspectives of general practitioners. They may hold significant application value in the management of undifferentiated diseases.

Objectives: Neddylation is an important post-translational modification involved in various physiological and pathological processes, but its regulatory role in B cell-mediated immune responses remains unclear. B cells exhibit different response to thymus dependent antigens and thymus independent antigens, making it significant to explore the role of neddylation in this process.

Methods: B cell-specific Ube2m knockout mice CD19^cre/+Ube2m^fl/fl were constructed using the Cre-LoxP system, with littermate Ube2m^fl/fl (Ctrl) mice as controls. Mice were immunized with 4 antigens (thymus dependent antigens: NP-CGG and DNP-KLH, thymus independent antigens: NP-LPS and DNP-Ficoll), respectively. Serum levels of NP/DNP specific antibodies (IgM, IgG1, IgG2a, IgG2b, IgG2c, IgG3) were detected by ELISA, and the proportions of germinal center B cells and plasma cells in the spleen and bone marrow were analyzed by flow cytometry. Statistical analysis was used to evaluate the differences in immune responses.

Results: Compared with the Ctrl mice, CD19^cre/+Ube2m^fl/fl mice showed significantly reduced levels of NP/DNP specific IgG1、IgG2a、IgG2b、IgG2c、IgG3 after thymus independent antigens (NP-LPS, DNP-Ficoll) immunization, while IgM levels were mostly unchanged, and the reduced proportions of plasma cells and germinal center B cells in the bone marrow or spleen were observed. In thymus dependent antigens (NP-CGG, DNP-KLH) immunization models, CD19^cre/+Ube2m^fl/fl mice exhibited significantly decreased levels of IgM, IgG1, IgG2a, IgG2b, IgG2c, IgG3, with differences in the timing and magnitude of antibody reduction induced by different antigens. A significant reduction in the proportions of plasma cells and NP specific germinal center B cells in the spleen were also observed.

Conclusions: Deficiency of B cell-intrinsic neddylation extensively impairs immune responses to both thymus independent antigens and thymus dependent antigens in mice, including early primary immune responses, Th1-type cellular immune responses, and Th2-type humoral immune responses, with antigen-dependent regulatory effects. This study provides experimental evidence for understanding the role of neddylation in B cell mediated immune responses.

As integral components of the immune response, the activating receptor NKG2D and its ligands, the UL16 binding protein family ( ULBPs ), play a pivotal role in autoimmune pathology. However, their specific functions exhibit multidimensional regulatory features. This review proposes a three-dimensional regulatory framework of "ligand supply, ligand fate, and receptor modulation" to dissect the role of this pathway in different diseases. The diversity of "ligand supply" governs the initiation of the signal, its "fate" determines the scope and nature of its effects, and "receptor modulation" ultimately sets the activation threshold of effector cells. By integrating and analyzing research advancements in diseases such as Systemic Lupus Erythematosus, Rheumatoid Arthritis, Type 1 Diabetes, and Multiple Sclerosis, we demonstrate how this framework can systematically elucidate the specific mechanisms of the NKG2D axis in each condition. This provides an analytical perspective for understanding its multifaceted functions across different disease contexts. A deeper understanding of this dynamic regulatory network offers new avenues for unveiling the pathogenesis of autoimmune diseases and for developing more precise therapeutic strategies.

Drug discovery is confronted with formidable challenges including high costs, significant risks, and prolonged timelines, necessitating disruptive technologies capable of systematically improving efficiency, enabling accurate predictions, and mitigating failure risks. Artificial intelligence (AI) agent-an emerging intelligent paradigm powered by large language models-hold immense potential to transform the entire drug development pipeline. Their core capability lies in performing autonomous reasoning, planning, and tool utilization oriented toward complex scientific objectives, thereby integrating and orchestrating multiple research stages and transitioning AI from a mere "tool" to an "active collaborator". On one hand, through knowledge integration and hypothesis generation, AI agents can identify underexplored targets and novel mechanisms. On the other hand, they can automate complex tasks such as molecular design, optimization, and synthesis planning, and close the loop from virtual design to physical experimentation by interfacing with automated experimental platforms. Moreover, AI agents are evolving toward higher-level paradigms, including the development of integrated drug design platforms and general-purpose biomedical agents. This review systematically outlines the core architecture of AI agents, highlights their applications across key stages of drug development, and discusses current limitations along with future directions, with the aim of providing a reference for researchers in related fields.

To address the core limitations of conventional insulin therapy-namely, delayed glycemic control and the frequent risk of hypoglycemia-the development of glucose-responsive insulin delivery systems capable of dynamically sensing blood glucose levels and releasing insulin on demand represents a pivotal strategy. Based on their underlying sensing mechanisms, these systems are primarily classified into three categories: those utilizing glucose oxidase, glucose-binding molecules, and phenylboronic acid. From the standpoint of administration routes, injectable and transdermal delivery are the two main approaches for glucose-responsive insulin. Injectable glucose-responsive insulin delivery systems are highly compatible with existing clinical treatment, primarily relying on glucose-responsive carriers to regulate the insulin release rate and enabling stable and efficient bioavailability. Transdermal glucose-responsive insulin delivery systems utilize glucose-responsive microneedle arrays to penetrate the skin stratum corneum and precisely regulate the insulin release rate, allowing for sufficient insulin delivery under almost painless conditions. This review systematically summarizes recent advances in both injectable and transdermal glucose-responsive insulin delivery systems, with a focus on the carrier construction strategies, glucose-responsive release mechanisms, and evolutionary pathways of preparation techniques. It also highlights the contributions of these systems toward improving glucose-responsive performance, therapeutic safety, biocompatibility, and patient adherence. Furthermore, challenges and future prospects for clinical translation are discussed. It is anticipated that this overview will provide valuable perspectives for further research and development in this field.

Spinal cord injury (SCI) is a major cause of motor disability. Epidural electrical stimulation (EES) has emerged as a promising neuromodulation technique and has been extensively investigated in recent years for promoting functional recovery after SCI. This case report describes a patient who sustained a thoracic SCI due to a fall from height. Following surgical intervention and four months of conventional rehabilitation, the patient's lower limb muscle strength remained at grade C on the American Spinal Injury Association (ASIA) Impairment Scale. After implantation of a closed-loop spinal neural interface for EES, the patient underwent a regimen of continuous stimulation with spatially targeted configurations and multimodal closed-loop rehabilitation training. This intervention led to progressive recovery, including the achievement of voluntary single-joint movements, independent standing, and assisted walking. At the 16-week follow-up (July 2025), the patient's ASIA grade improved from C to D. Improvements were also noted in sensory and autonomic functions alongside motor recovery. The overall rehabilita-tion outcomes were markedly better than those typically reported for chronic-phase SCI patients in the literature. Importantly, no implant-related infections or neurological complications were observed. These results indicate that the combination of closed-loop spinal neural interface-mediated EES and structured rehabilitation holds considerable potential as a therapeutic strategy.

Elbow orthoses offer a straightforward mechanical approach to rehabilitation following elbow joint injuries. The convergence of artificial intelligence with these orthoses has given rise to elbow rehabilitation robots, designed to address the personalized rehabilita-tion requirements of patients with diverse injury profiles. A typical elbow rehabilitation robot system comprises four key components: a mechanical rigid structure, an actuation system, bionic sensors, and integrated software. The rigid structure, analogous to the human skeletal system, includes linkage assemblies and gear transmission systems. The actuation system, mimicking muscles and ligaments, generates and modulates the necessary forces and torques for movement, employing actuators such as pneumatic, elastic (e.g., series elastic actuators), and cable-driven types. Bionic sensors, serving as the robot's perceptual interface, encompass photoelectric encoders, force/torque sensors, electromyo-graphic (EMG) signal sensors, and temperature sensors. The software system-encompassing control algorithms and machine learning models-functions as the "neural center," analogous to the brain and spinal cord, and is responsible for intelligent decision-making and motion control. The core technological achievement lies in seamlessly integrating hardware and software to enable precise tracking of elbow joint kinematics and adaptive modulation of assistive forces based on real-time human-robot interaction. This integration supports multiple training modalities-including passive, assistive, active, and resistive modes-to deliver safe, tailored, and intelligent rehabilitation support across different recovery phases. By harnessing technologies like bio-inspired design, precise impedance control, EMG-based assistance, and virtual reality integrated task training, these robotic systems improve training comfort, assistance accuracy, and patient adherence, potentially reducing post-injury disability rates. This review outlines the current state of research in elbow rehabilitation robotics, details the key system components and primary training modalities, discusses clinical demands and future development trends, and aims to offer insights for the further refinement of rehabilitation robotic systems.

Objectives: To investigate the association between sodium-glucose linked transporter (SGLT) 2 inhibitors use and functional outcomes at discharge in diabetic patients with acute ischemic stroke (AIS).

Methods: A retrospective cohort study was conducted using data from the CASE-Ⅱ registry (NCT04487340) for patients admitted between January 2022 and November 2024. Diabetic patients with AIS were categorized as SGLT2 inhibitor users or non-users (control group). The primary outcome was a favorable functional outcome, defined as a modified Rankin Scale (mRS) score of 0-2 at discharge. Propensity score matching (PSM) was performed in a 1∶2 ratio. Multivariate logistic regression models were used to assess the association between SGLT2 inhibitor use and outcomes, adjusting for potential confounders. Stratified analyses were performed based on pre-existing diabetes versus newly diagnosed diabetes during the hospitalization.

Results: The study included 9220 patients (mean age 67±11 years; 39.4% female), of whom 1915 (20.8%) were SGLT2 inhibitor users; after PSM there were 1912 cases user group and 3683 cases in non-user group. Before PSM, the proportion of patients with an mRS score of 0-2 at discharge was significantly higher in the SGLT2 inhibitor group than that in the control group (79.8% vs. 69.2%, P<0.01). After multivariable adjustment, SGLT2 inhibitor use remained significantly associated with higher proportion of mRS 0-2 (adjusted OR=1.24, 95%CI: 1.08-1.42, P<0.01) and mRS 0-3 (adjusted OR=1.62, 95%CI: 1.32-1.99, P<0.01) at discharge. Results were consistent after PSM. In stratified analyses, SGLT2 inhibitor use was associated with improved functional outcomes in both patients with pre-existing diabetes and those with newly diagnosed diabetes.

Conclusions: Among diabetic patients with AIS, SGLT2 inhibitor use is independently associated with better functional outcomes at discharge, showing a potential benefit for short-term neurological recovery.

Objectives: To investigate the relationship between hematocrit (HCT) and low resting oxygen saturation in a high-altitude population.

Methods: This retrospective cohort study included 3075 Tibetan adult inpatients who underwent elective non-cardiac, non-thoracic surgery at Tibet Autonomous Region People's Hospital between January 2023 and October 2024. Multivariate logistic regression was used to assess the association between HCT and low resting oxygen saturation (SpO₂<88%). Restricted cubic splines were employed to model non-linear relationships, and piecewise regression was used to identify inflection points.

Results: Multivariate logistic regression analysis showed that elevated HCT was an independent risk factor for low oxygen saturation in the patients. For every 5-percentage-point increase in HCT, the risk of low oxygen saturation increased by 12% (P<0.01). The restricted cubic spline model revealed a significant U-shaped association between HCT and the risk of low SpO₂ (P<0.01). The optimal HCT threshold for the total population was 42.8%. When HCT was below 42.8%, each 5% increase in HCT was associated with a 22% reduction in the risk of low oxygen saturation in the patients (OR=0.78, 95%CI: 0.62-0.98, P<0.05). When HCT was 42.8% or above, the risk increased by 27.2% (OR=1.27, 95%CI: 1.13-1.43, P<0.01). Stratified analysis indicated that the optimal HCT threshold was 40.4% for female patients and 46.6% for male patients.

Conclusions: There is a U-shaped association between HCT and low resting oxygen saturation in the Tibetan population. An HCT level exceeding the population-specific threshold suggests that the adaptive erythropoietic response to high-altitude hypoxia may transition from being compensatory to potentially detrimental.

Objectives: To analyze the distribution of Y chromosome azoospermia factor (AZF) microdeletions and their association with testicular development in male pediatric patients with congenital reproductive system diseases.

Methods: A prospective cohort study was conducted on pediatric patients admitted to the Department of Urology of Shanghai Children's Hospital from November 2021 to December 2023. The observation group included boys with hypospadias, cryptorchidism, or disorders of sex development (DSD), while the control group comprised boys with phimosis, indirect inguinal hernia, or hydrocele. Blood samples were collected for AZF microdeletion analysis using multiplex PCR to detect 15 sequence-tagged sites. Testicular ultrasound was performed to record testicular position and volume. Propensity score matching (PSM) was used to balance the groups. After matching, testicular volume differences were assessed. Stratified analyses compared testicular volume among children with AZF microdeletions, the control group, and children without AZF microdeletions in the observation group.

Results: A total of 493 children were enrolled (observation group: 463; control group: 30). The observation group consisted of 372 cases of hypospadias, 71 cases of cryptorchidism, and 20 cases of DSD. No Y chromosome microdeletions were detected in the control group. Four boys in the observation group had AZF microdeletions: one with cryptorchidism (AZFc+AZFd), one with isolated hypospadias (AZFc), and two with DSD (one with AZFb+AZFc+AZFd and one with AZFa). Ultraso-nography was performed for 888 testes. After PSM, testicular volume was significantly smaller in the observation group than that in the control group (P<0.01). Stratified analysis revealed a trend towards smaller testicular volumes in children with AZF microdeletions compared to the other two groups.

Conclusions: The prevalence of Y chromosome microdeletions is higher in male children with congenital reproductive system diseases compared to the general population, particularly in those with DSD. Hypospadias, cryptorchidism, DSD, and AZF microdeletions may be associated with delayed testicular development in these children.