Hasil untuk "Diseases of the endocrine glands. Clinical endocrinology"

Yu Cao, Huiting Chen, Hui Liu et al.

Background Type 2 diabetes mellitus (T2DM) is a major cause of declining renal function. Methods Temporal trends in T2DM-related chronic kidney disease (CKD-T2DM) incidence across 204 countries and territories from 1992 to 2021 were analyzed using data from the Global Burden of Disease 2021. The impact of macro-factors (demographic change, age, period, and birth cohort) on CKD-T2DM incidence trends was assessed using decomposition analyses and age-period-cohort modeling, highlighting opportunities to improve incidence and reduce regional disparities. Results In 2021, global CKD-T2DM incidence cases reached 2.01 million, a 150.92% increase since 1992, with population growth and aging contributing to 80% of this rise. The age-standardized incidence rate (ASIR) ranged from 15.09 per 100,000 in low sociodemographic index (SDI) regions to 23.07 in high SDI regions. China, India, the United States, and Japan have the most incidence cases, accounted for 69% of incidence cases globally. With 175 countries showing an increasing ASIR trend. Unfavorable trend in ASIR increase were generally found in most high-middle and middle SDI countries, such as China and Mexico (net drift=0.15% and 1.17%, per year). Age-period-cohort analyses indicated a high incidence risk near age 80, with worsening risks for recent periods and birth cohorts, except in high SDI areas. Conclusion The CKD-T2DM incidence burden continues to rise globally, with significant variations between countries, posing major global health implications. CKD-T2DM is largely preventable and treatable, warranting greater attention in global health policy, particularly for older populations and in low and middle SDI regions.

Jing Yue, Jing Yue, Jinhao Guo et al.

Diabetic nephropathy (DN) remains a leading cause of end-stage renal disease despite guideline-based therapy. Acupuncture has been explored as an adjunct or alternative approach. We reviewed preclinical and clinical studies (2010–2025) on acupuncture for DN, summarizing mechanisms, intervention models (acupuncture alone; with Chinese medicine; with Western medicine; triple therapy), renal outcomes, and safety. Across animal and human data, acupuncture modulates immune–inflammatory and metabolic pathways—including HMGB1/NLRP3/NF-κB, SIRT1/AMPK/PGC-1α, eNOS–NO–cGMP, and autophagy (ULK1–Beclin-1–LC3)—enhances antioxidant defenses (SOD↑, MDA/8-OHdG↓), protects podocytes, and improves microcirculation. Clinically, it is associated with reductions in proteinuria (24-h UP, UACR/UAER), improvements in renal function (Scr, BUN, eGFR), and better metabolic control and symptoms. Combined regimens (with Chinese or Western medicines) tend to yield faster or broader benefits, with no serious adverse events reported in the included studies. Evidence quality is limited by small sample sizes, single-center designs, short follow-up, heterogeneous endpoints, and incomplete safety reporting. Acupuncture shows multi-target, complementary effects for DN and may be integrated with standard care. High-quality, multicenter randomized controlled trials with standardized endpoints (e.g., proteinuria, eGFR slope), robust safety monitoring, and embedded mechanistic assessments are warranted.

Carlijn A Hoekx, Borja Martinez-Tellez, Maaike E Straat et al.

Objective: Inflammation contributes to the development of type 2 diabetes mellitus (T2DM). While South Asians are more prone to develop T2DM than Europids, the inflammatory phenotype of the South Asian population remains relatively unknown. Therefore, we aimed to investigate potential differences in circulating levels of inflammation-related proteins in South Asians compared with Europids with T2DM. Method: In this secondary analysis of three randomized controlled trials, relative plasma levels of 73 inflammation-related proteins were measured using an Olink Target Inflammation panel and the serum fibroblast growth factor 21 (FGF21) concentration using an ELISA kit in Dutch South Asians (n = 47) and Dutch Europids (n = 49) with T2DM. Results: Of the 73 inflammation-related proteins, the relative plasma levels of six proteins were higher (stem cell factor, caspase-8, C–C motif chemokine ligand 28, interferon-gamma, sulfotransferase 1A1 and cystatin D; q-value <0.05), while relative levels of six proteins were lower (FGF21, human fibroblast collagenase, interferon-8, C–C motif chemokine ligand 4, C–X–C motif chemokine ligand 6 and monocyte chemoattractant protein-1; q-value <0.05) in South Asians compared with Europids. Of these, the effect size of FGF21 was the largest, particularly in females. We validated this finding by assessing the FGF21 concentration in serum. The FGF21 concentration was indeed lower in South Asians compared with Europids with T2DM in both males (−42.2%; P < 0.05) and females (−58.5%; P < 0.001). Conclusion: Relative plasma levels of 12 inflammation-related proteins differed between South Asians and Europids with T2DM, with a significantly pronounced reduction in FGF21. In addition, the serum FGF21 concentration was significantly lower in South Asian males and females compared with Europids. Whether low FGF21 is an underlying cause or consequence of T2DM in South Asians remains to be determined. Clinical trial registration: ClinicalTrials.gov (NCT01761318, registration date 20-12-2012; NCT02660047, registration date 21-03-2018; and NCT03012113, registration date 06-01-2017).

Zahra Zarei, Elham Nejadsadeghi, Seyedeh Leila Dehghani et al.

This study aimed to evaluate the effect of remdesivir on liver tests in diabetic and non-diabetic COVID-19 patients through a multicenter study conducted in Southeast Iran. Therefore, 200 participants, comprising 98 patients with diabetes and 102 non-diabetic subjects, were assessed based on the Declaration of Helsinki, with proper inclusion, and exclusion criteria. Demographic data were collected using a detailed questionnaire and a clinical checklist documenting underlying health conditions, particularly diabetes and hypertension. Liver function tests measured key enzymes and substances, including aspartate aminotransferase (AST), alanine aminotransferase (ALT), alkaline phosphatase (ALP), and total bilirubin, and data analysis was performed using SPSS (Confidence Interval = 0.95, p ≤ 0.05). Key findings indicate a significant correlation between increasing age and diabetes prevalence, with older age groups exhibiting higher rates of diabetes. Gender analysis revealed a slight predominance of females among diabetic patients, while educational attainment appeared lower in this group, suggesting a potential link between education and diabetes incidence. In patients with diabetes, AST levels rose from 19.2 ± 2.1 U/L before treatment to 25.3 ± 3.1 U/L after treatment, while ALT levels increased from 18.1 ± 1.4 U/L to 23.5 ± 2.2 U/L. Non-diabetic patients showed less pronounced increases in liver enzymes, with AST rising from 28.7 ± 3.1 U/L to 13.2 ± 2.1 U/L after treatment and ALT changing from 18.6 ± 3.2 U/L to 19.6 ± 3.1 U/L. Health-related factors, particularly the prevalence of hypertension and obesity, were notably higher among patients with diabetes. Lifestyle behaviors, including smoking and physical activity levels, further distinguished the two groups, with patients with diabetes showing a higher smoking prevalence and a lower engagement in regular exercise. The impact of remdesivir treatment on liver function revealed significant increases in liver enzyme levels among patients with diabetes post-treatment, contrasting with stable liver function in non-diabetic patients. The study underscores the intricate relationship between diabetes, liver health, and COVID-19, emphasizing the importance of considering comorbidities in treatment and management strategies for diabetic patients.

Maroun Badr, Ghazwa El-Rabaa, Marianne Freiha et al.

Childhood obesity has emerged as a significant public health challenge, with profound consequences that negatively impact endocrine functions. Excess adiposity in children leads to dysregulation of various hormonal pathways, notably insulin resistance and hyperinsulinemia, the best-established endocrine changes in obesity. If insulin resistance is not adequately managed, it might precipitate type 2 diabetes. Another common finding among children with obesity is thyroid dysfunction. Some studies suggest that obesity may be associated with alterations in thyroid hormone levels, potentially leading to hypothyroidism, although the relationship is complex and not fully understood. Additionally, obesity affects the hypothalamic-pituitary-gonadal axis, resulting in precocious puberty, particularly in girls. Elevated leptin levels, a hormone produced by adipose tissue, can contribute to a paradoxical state of leptin resistance, further complicating metabolic processes and appetite regulation. Moreover, childhood obesity can result in increased secretion of cortisol, which may enhance the risk of developing metabolic syndrome and cardiovascular complications. The interplay between obesity and endocrine function also extends to growth patterns, where excess weight can lead to growth acceleration followed by potential short stature in adulthood due to early epiphyseal closure. Addressing the endocrine consequences of childhood obesity requires a comprehensive approach that includes prevention, early intervention, and management strategies tailored to this vulnerable population. Understanding these complex interactions is crucial for developing effective public health policies to mitigate the impact of obesity on endocrine health in children. By reviewing research, this work provides a comprehensive overview of the most relevant endocrine consequences of childhood obesity.

Li-Chin Chen, Ji-Tian Sheu, Yuh-Jue Chuang

Demographic attributes are universally present in electronic health records. They are the most widespread information across populations and diseases, and serve as vital predictors in clinical risk stratification and treatment decisions. Despite their significance, these attributes are often treated as auxiliaries in model design, with limited attention being paid to learning their representations. This study explored the development of a General Demographic Pre-trained (GDP) model as a foundational model tailored to demographic attributes, focusing on age and gender. The model is pre-trained and evaluated using datasets with diverse diseases and populations compositions from different geographic regions. The composition of GDP architecture was explored through examining combinations of ordering approaches and encoding methods to transform tabular demographic inputs into effective latent embeddings. Results demonstrate the feasibility of GDP to generalize across task, diseases, and populations. In detailed composition, the sequential ordering substantially improves model performance in discrimination, calibration, and the corresponding information gain at each decision tree split, particularly in diseases where age and gender contribute significantly to risk stratification. Even in datasets where demographic attributes hold relatively low predictive value, GDP enhances the representational importance, increasing their influence in downstream gradient boosting models. The findings suggest that foundation models for tabular demographic attributes offer a promising direction for improving predictive performance in healthcare applications.

Ke Chen, Haohan Wang

Modern disease classification often overlooks molecular commonalities hidden beneath divergent clinical presentations. This study introduces a transcriptomics-driven framework for discovering disease relationships by analyzing over 1300 disease-condition pairs using GenoMAS, a fully automated agentic AI system. Beyond identifying robust gene-level overlaps, we develop a novel pathway-based similarity framework that integrates multi-database enrichment analysis to quantify functional convergence across diseases. The resulting disease similarity network reveals both known comorbidities and previously undocumented cross-category links. By examining shared biological pathways, we explore potential molecular mechanisms underlying these connections-offering functional hypotheses that go beyond symptom-based taxonomies. We further show how background conditions such as obesity and hypertension modulate transcriptomic similarity, and identify therapeutic repurposing opportunities for rare diseases like autism spectrum disorder based on their molecular proximity to better-characterized conditions. In addition, this work demonstrates how biologically grounded agentic AI can scale transcriptomic analysis while enabling mechanistic interpretation across complex disease landscapes. All results are publicly accessible at github.com/KeeeeChen/Pathway_Similarity_Network.

Yiming Lei, Michael Nguyen, Tzu Chia Liu et al.

Chest X-rays play a pivotal role in diagnosing respiratory diseases such as pneumonia, tuberculosis, and COVID-19, which are prevalent and present unique diagnostic challenges due to overlapping visual features and variability in image quality. Severe class imbalance and the complexity of medical images hinder automated analysis. This study leverages deep learning techniques, including transfer learning on pre-trained models (AlexNet, ResNet, and InceptionNet), to enhance disease detection and classification. By fine-tuning these models and incorporating focal loss to address class imbalance, significant performance improvements were achieved. Grad-CAM visualizations further enhance model interpretability, providing insights into clinically relevant regions influencing predictions. The InceptionV3 model, for instance, achieved a 28% improvement in AUC and a 15% increase in F1-Score. These findings highlight the potential of deep learning to improve diagnostic workflows and support clinical decision-making.

Anna L. Trella, Kelly W. Zhang, Inbal Nahum-Shani et al.

Online reinforcement learning (RL) algorithms offer great potential for personalizing treatment for participants in clinical trials. However, deploying an online, autonomous algorithm in the high-stakes healthcare setting makes quality control and data quality especially difficult to achieve. This paper proposes algorithm fidelity as a critical requirement for deploying online RL algorithms in clinical trials. It emphasizes the responsibility of the algorithm to (1) safeguard participants and (2) preserve the scientific utility of the data for post-trial analyses. We also present a framework for pre-deployment planning and real-time monitoring to help algorithm developers and clinical researchers ensure algorithm fidelity. To illustrate our framework's practical application, we present real-world examples from the Oralytics clinical trial. Since Spring 2023, this trial successfully deployed an autonomous, online RL algorithm to personalize behavioral interventions for participants at risk for dental disease.

Qingshan Hou, Shuai Cheng, Peng Cao et al.

Representation learning offers a conduit to elucidate distinctive features within the latent space and interpret the deep models. However, the randomness of lesion distribution and the complexity of low-quality factors in medical images pose great challenges for models to extract key lesion features. Disease diagnosis methods guided by contrastive learning (CL) have shown significant advantages in lesion feature representation. Nevertheless, the effectiveness of CL is highly dependent on the quality of the positive and negative sample pairs. In this work, we propose a clinical-oriented multi-level CL framework that aims to enhance the model's capacity to extract lesion features and discriminate between lesion and low-quality factors, thereby enabling more accurate disease diagnosis from low-quality medical images. Specifically, we first construct multi-level positive and negative pairs to enhance the model's comprehensive recognition capability of lesion features by integrating information from different levels and qualities of medical images. Moreover, to improve the quality of the learned lesion embeddings, we introduce a dynamic hard sample mining method based on self-paced learning. The proposed CL framework is validated on two public medical image datasets, EyeQ and Chest X-ray, demonstrating superior performance compared to other state-of-the-art disease diagnostic methods.

Karthik S. Vedula, Annika Gupta, Akshay Swaminathan et al.

Large language models (LLMs) excel at clinical information extraction but their computational demands limit practical deployment. Knowledge distillation--the process of transferring knowledge from larger to smaller models--offers a potential solution. We evaluate the performance of distilled BERT models, which are approximately 1,000 times smaller than modern LLMs, for clinical named entity recognition (NER) tasks. We leveraged state-of-the-art LLMs (Gemini and OpenAI models) and medical ontologies (RxNorm and SNOMED) as teacher labelers for medication, disease, and symptom extraction. We applied our approach to over 3,300 clinical notes spanning five publicly available datasets, comparing distilled BERT models against both their teacher labelers and BERT models fine-tuned on human labels. External validation was conducted using clinical notes from the MedAlign dataset. For disease extraction, F1 scores were 0.82 (teacher model), 0.89 (BioBERT trained on human labels), and 0.84 (BioBERT-distilled). For medication, F1 scores were 0.84 (teacher model), 0.91 (BioBERT-human), and 0.87 (BioBERT-distilled). For symptoms: F1 score of 0.73 (teacher model) and 0.68 (BioBERT-distilled). Distilled BERT models had faster inference (12x, 4x, 8x faster than GPT-4o, o1-mini, and Gemini Flash respectively) and lower costs (85x, 101x, 2x cheaper than GPT-4o, o1-mini, and Gemini Flash respectively). On the external validation dataset, the distilled BERT model achieved F1 scores of 0.883 (medication), 0.726 (disease), and 0.699 (symptom). Distilled BERT models were up to 101x cheaper and 12x faster than state-of-the-art LLMs while achieving similar performance on NER tasks. Distillation offers a computationally efficient and scalable alternative to large LLMs for clinical information extraction.

Yun-Yi Zou, Xiang-bin Tang, Zhang-Lin Chen et al.

IntroductionRecent reports indicate that mitochondrial quality decreases during non-alcoholic fatty liver disease (NAFLD) progression, and targeting the mitochondria may be a possible treatment for NAFLD. Exercise can effectively slow NAFLD progression or treat NAFLD. However, the effect of exercise on mitochondrial quality in NAFLD has not yet been established.MethodsIn the present study, we fed zebrafish a high-fat diet to model NAFLD, and subjected the zebrafish to swimming exercise.ResultsAfter 12 weeks, swimming exercise significantly reduced high-fat diet-induced liver injury, and reduced inflammation and fibrosis markers. Swimming exercise improved mitochondrial morphology and dynamics, inducing upregulation of optic atrophy 1(OPA1), dynamin related protein 1 (DRP1), and mitofusin 2 (MFN2) protein expression. Swimming exercise also activated mitochondrial biogenesis via the sirtuin 1 (SIRT1)/ AMP-activated protein kinase (AMPK)/ PPARgamma coactivator 1 alpha (PGC1α) pathway, and improved the mRNA expression of genes related to mitochondrial fatty acid oxidation and oxidative phosphorylation. Furthermore, we find that mitophagy was suppressed in NAFLD zebrafish liver with the decreased numbers of mitophagosomes, the inhibition of PTEN-induced kinase 1 (PINK1) – parkin RBR E3 ubiquitin protein ligase (PARKIN) pathway and upregulation of sequestosome 1 (P62) expression. Notably, swimming exercise partially recovered number of mitophagosomes, which was associated with upregulated PARKIN expression and decreased p62 expression.DiscussionThese results demonstrate that swimming exercise could alleviate the effects of NAFLD on the mitochondria, suggesting that exercise may be beneficial for treating NAFLD.

Johnny Awwad, Johnny Awwad, Juan Garcia-Velasco et al.

Ning Ning, Juan Li, Wenwu Sun et al.

BackgroundNonthyroidal illness syndrome (NTIS) is a common endocrine dysfunction predicting unfavorable outcomes in critical illness. The objective of the study is to evaluate the association between different NTIS subtypes with outcomes in septic patients.MethodsSeptic patients in two Chinese academic centers from October 2012 and October 2022 are enrolled in analysis. Multivariable regressions are used to assess associations between NTIS and outcomes. Outcomes include in-hospital mortality, length of stay in hospital (LOS), non-invasive ventilation failure and weaning failure. Patients with NTIS are categorized into 4 types according to the different levels of FT4 and TSH. The association between different NTIS subtypes and mortality are further analyzed. Survival curve is plotted using the Kaplan–Meier method.ResultsAfter screening, a total of 1226 septic patients with complete thyroid hormones result are eventually enrolled. Among them, 520 (42.4%) patients are diagnosed as NTIS. In multivariable regression analysis, NTIS is independently associated with increased 30-days mortality (OR=1.759, CI 1.009-3.104, p=0.047), but has no association with 60-days mortality (OR=1.524, CI 0.893-2.618, p=0.123), 90-days mortality (OR=1.411, CI 0.831-2.408, p=0.203), LOS, non-invasive ventilation failure or weaning failure. In NTIS subtypes, NTIS patients with low FT3 and TSH levels, regardless of the FT4 values, have significantly higher mortality than euthyroid patients (30-days mortality, OR= 6.488, CI 1.546-27.808, p=0.01; 60-days mortality, OR=3.973, CI 1.006-15.579, p=0.046; 90-days mortality, OR=3.849, CI 0.977-15.088, p=0.051). This result is consistent in patients with low FT3 and FT4 levels, regardless of the TSH values (30-days mortality, OR=3.349, CI 1.402-7.957, p=0.006; 60-days mortality, OR= 2.594, CI 1.122-5.930, p=0.024; 90-days mortality, OR=2.55, CI 1.110-5.804, p=0.025). There is no survival difference between NTIS patients with low FT3 only and euthyroid patients. Survival plot shows the worst prognosis is in NTIS patients with low FT3, FT4 and TSH level.ConclusionsNTIS is frequent in sepsis. A reduction of FT3 together with FT4 or TSH, but not FT3 only, is associated with an increased risk of mortality.

Paulina Dabrowska‐Schlepp, Andreas Busch, Jin‐Song Shen et al.

Abstract Fabry disease (FD, OMIM 301500) is a rare X‐linked inherited lysosomal storage disorder associated with reduced activities of α‐galactosidase A (aGal, EC 3.2.1.22). The current standard of care for FD is based on enzyme replacement therapy (ERT), in which a recombinantly produced version of αGal is intravenously (iv) applied to Fabry patients in biweekly intervals. Though the iv application is clinically efficacious, periodical infusions are inconvenient, time‐ and resource‐consuming and they negatively impact the patients’ quality of life. Subcutaneous (sc) injection, in contrast, is an established route of administration for treatment of chronic conditions. It opens the beneficial option of self‐administration, thereby improving patients’ quality of life and at the same time reducing treatment costs. We have previously shown that Moss‐α‐Galactosidase (moss‐aGal), recombinantly produced in the moss Physcomitrium patens, is efficient in degrading accumulated Gb3 in target organs of murine model of FD and in the phase I clinical study, we obtained first efficacy evidence in human patients following single iv infusion. Here, we tested the efficacy of subcutaneous administration of moss‐aGal and compared it with the results observed following iv infusion in Fabry mice. The obtained findings demonstrate that subcutaneously applied moss‐aGal is correctly transported to target organs and efficacious in degrading Gb3 deposits there and thus suggest the possibility of using this route of administration for therapy of Fabry disease.

Sunjun Kweon, Junu Kim, Jiyoun Kim et al.

The development of large language models tailored for handling patients' clinical notes is often hindered by the limited accessibility and usability of these notes due to strict privacy regulations. To address these challenges, we first create synthetic large-scale clinical notes using publicly available case reports extracted from biomedical literature. We then use these synthetic notes to train our specialized clinical large language model, Asclepius. While Asclepius is trained on synthetic data, we assess its potential performance in real-world applications by evaluating it using real clinical notes. We benchmark Asclepius against several other large language models, including GPT-3.5-turbo and other open-source alternatives. To further validate our approach using synthetic notes, we also compare Asclepius with its variants trained on real clinical notes. Our findings convincingly demonstrate that synthetic clinical notes can serve as viable substitutes for real ones when constructing high-performing clinical language models. This conclusion is supported by detailed evaluations conducted by both GPT-4 and medical professionals. All resources including weights, codes, and data used in the development of Asclepius are made publicly accessible for future research. (https://github.com/starmpcc/Asclepius)

Shenghuan Sun, Gregory M. Goldgof, Atul Butte et al.

Generative models capable of capturing nuanced clinical features in medical images hold great promise for facilitating clinical data sharing, enhancing rare disease datasets, and efficiently synthesizing annotated medical images at scale. Despite their potential, assessing the quality of synthetic medical images remains a challenge. While modern generative models can synthesize visually-realistic medical images, the clinical validity of these images may be called into question. Domain-agnostic scores, such as FID score, precision, and recall, cannot incorporate clinical knowledge and are, therefore, not suitable for assessing clinical sensibility. Additionally, there are numerous unpredictable ways in which generative models may fail to synthesize clinically plausible images, making it challenging to anticipate potential failures and manually design scores for their detection. To address these challenges, this paper introduces a pathologist-in-the-loop framework for generating clinically-plausible synthetic medical images. Starting with a diffusion model pretrained using real images, our framework comprises three steps: (1) evaluating the generated images by expert pathologists to assess whether they satisfy clinical desiderata, (2) training a reward model that predicts the pathologist feedback on new samples, and (3) incorporating expert knowledge into the diffusion model by using the reward model to inform a finetuning objective. We show that human feedback significantly improves the quality of synthetic images in terms of fidelity, diversity, utility in downstream applications, and plausibility as evaluated by experts.

Eliza B. Geer

Enrique Cervantes-Pérez, Gabino Cervantes-Guevara, Astrid Villaseñor-Ochoa et al.

Hay condiciones que resultan en la obstrucción o lesión del sistema linfático y que llevan a una filtración de la linfa a las cavidades subyacentes, dando lugar al llenado del espacio peritoneal, condicionando la aparición de ascitis quilosa (AQ). Las causas principales en adultos son las lesiones quirúrgicas, obstrucción, compresión extrínseca del sistema linfático, infección, cirrosis, enfermedades inflamatorias y tumores benignos o malignos. Existen diversos métodos diagnósticos como la paracentesis, citología del líquido ascítico, biopsia peritoneal, laparoscopia, así como estudios de imagen para complementar el diagnóstico.

Wei Dai, Ziyao Zhang, Lixia Tian et al.

In neuroimaging analysis, fMRI can well assess the function changes for brain diseases with no obvious structural lesions. To date, most deep-learning-based fMRI studies have employed functional connectivity (FC) as the basic feature for disease classification. However, FC is calculated on time series of predefined regions of interest and neglects detailed information contained in each voxel. Another drawback of using FC is the limited sample size for the training of deep models. The low representation ability of FC leads to poor performance in clinical practice, especially when dealing with multimodal medical data involving multiple types of visual signals and textual records for brain diseases. To overcome this bottleneck problem in the fMRI feature modality, we propose BrainFormer, an end-to-end functional interaction learning method for brain disease classification with single fMRI volume. Unlike traditional deep learning methods that construct convolution and transformers on FC, BrainFormer learns the functional interaction from fMRI signals, by modeling the local cues within each voxel with 3D convolutions and capturing the global correlations among distant regions with specially designed global attention mechanisms from shallow layers to deep layers. Meanwhile, BrainFormer can deal with multimodal medical data including fMRI volume, structural MRI, FC features and phenotypic data to achieve more comprehensive brain disease diagnosis. We evaluate BrainFormer on five independent multi-site datasets on autism, Alzheimer's disease, depression, attention deficit hyperactivity disorder and headache disorders. The results demonstrate its effectiveness and generalizability for multiple brain diseases diagnosis with multimodal features. BrainFormer may promote precision of neuroimaging-based diagnosis in clinical practice and motivate future studies on fMRI analysis.