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

Mingchen Shao, Yuzhang Xie, Carl Yang et al.

Accurate extraction of Alzheimer's Disease and Related Dementias (ADRD) phenotypes from electronic health records (EHR) is critical for early-stage detection and disease staging. However, this information is usually embedded in unstructured textual data rather than tabular data, making it difficult to be extracted accurately. We therefore propose LLM-MINE, a Large Language Model-based phenotype mining framework for automatic extraction of ADRD phenotypes from clinical notes. Using two expert-defined phenotype lists, we evaluate the extracted phenotypes by examining their statistical significance across cohorts and their utility for unsupervised disease staging. Chi-square analyses confirm statistically significant phenotype differences across cohorts, with memory impairment being the strongest discriminator. Few-shot prompting with the combined phenotype lists achieves the best clustering performance (ARI=0.290, NMI=0.232), substantially outperforming biomedical NER and dictionary-based baselines. Our results demonstrate that LLM-based phenotype extraction is a promising tool for discovering clinically meaningful ADRD signals from unstructured notes.

Jiawen Kang, Kun Li, Dongrui Han et al.

Automated Alzheimer's Disease (AD) screening has predominantly followed the inductive paradigm of pattern recognition, which directly maps the input signal to the outcome label. This paradigm sacrifices construct validity of clinical protocol for statistical shortcuts. This paper proposes Agentic Cognitive Profiling (ACP), an agentic framework that realigns automated screening with clinical protocol logic across multiple cognitive domains. Rather than learning opaque mappings from transcripts to labels, the framework decomposes standardized assessments into atomic cognitive tasks and orchestrates specialized LLM agents to extract verifiable scoring primitives. Central to our design is decoupling semantic understanding from measurement by delegating all quantification to deterministic function calling, thereby mitigating hallucination and restoring construct validity. Unlike popular datasets that typically comprise around a hundred participants under a single task, we evaluate on a clinically-annotated corpus of 402 participants across eight structured cognitive tasks spanning multiple cognitive domains. The framework achieves 90.5% score match rate in task examination and 85.3% accuracy in AD prediction, surpassing popular baselines while generating interpretable cognitive profiles grounded in behavioral evidence. This work demonstrates that construct validity and predictive performance need not be traded off, charting a path toward AD screening systems that explain rather than merely predict.

Md Mezbahul Islam, John Michael Templeton, Masrur Sobhan et al.

Parkinson's disease (PD) is a chronic and complex neurodegenerative disorder influenced by genetic, clinical, and lifestyle factors. Predicting this disease early is challenging because it depends on traditional diagnostic methods that face issues of subjectivity, which commonly delay diagnosis. Several objective analyses are currently in practice to help overcome the challenges of subjectivity; however, a proper explanation of these analyses is still lacking. While machine learning (ML) has demonstrated potential in supporting PD diagnosis, existing approaches often rely on subjective reports only and lack interpretability for individualized risk estimation. This study proposes SCOPE-PD, an explainable AI-based prediction framework, by integrating subjective and objective assessments to provide personalized health decisions. Subjective and objective clinical assessment data are collected from the Parkinson's Progression Markers Initiative (PPMI) study to construct a multimodal prediction framework. Several ML techniques are applied to these data, and the best ML model is selected to interpret the results. Model interpretability is examined using SHAP-based analysis. The Random Forest algorithm achieves the highest accuracy of 98.66 percent using combined features from both subjective and objective test data. Tremor, bradykinesia, and facial expression are identified as the top three contributing features from the MDS-UPDRS test in the prediction of PD.

M. Puig-Domingo, M. Marazuela, A. Giustina

S Bilgic, B Akovalı

Youtu Wu, Dawei Wang, Yi Guo et al.

Abstract Purpose Sodium imbalance are common complications after transsphenoidal surgery (TSS) for pituitary neuroendocrine tumors (PitNETs). We characterized the types of sodium imbalance, identified risk factors for these disorders, and provided corresponding treatment advice. Methods We screened patients who had undergone TSS for PitNETs at a single center to identify those who did and did not (control) develop sodium imbalance. Disorders were classified using three groups, based mainly on the serum sodium level and degree of daily increase or decrease therein. We performed multivariable logistic regression analysis to identify risk factors among numerous variables (patient characteristics, third ventricle deformation, tumor volume, maximum tumor diameter, hydrocephalus, cerebrospinal fluid rhinorrhea, and pituitary target gland axes). Results The sample comprised 105 patients with and 129 patients without sodium imbalance. Logistic regression analysis showed that hydrocephalus [P = 0.0015, odds ratio (OR) = 7.112, 95% confidence interval (CI) 1.475–34.3], cerebrospinal fluid rhinorrhea (P < 0.001, OR = 4.62, 95% CI 2.372–9), and preoperative hypothalamus–pituitary–gonadal (HPG) axis hypofunction (P = 0.009, OR = 3.211, 95% CI 1.341–7.691) were independent risk factors sodium imbalance development after TSS. Compared with the control, risk factors differed among disorder groups. Conclusion This study showed that cerebrospinal fluid rhinorrhea, hydrocephalus, and preoperative HPG axis hypofunction are risk factors for sodium imbalance development after TSS for PitNETs. We divided sodium imbalances into three groups to guide treatment.

Amir Hossein Karimi, Peter YF Zeng, Matthew Cecchini et al.

Anaplastic thyroid carcinoma (ATC) is among the most daunting entities in clinical oncology. Large-scale genomic studies of thyroid cancer within the last decade have uncovered a distinct set of recurrent somatic alterations implicated in the development, aggressiveness, and treatment resistance of ATC. The sequence of events leading to the development of ATC commonly begins with a tumorigenic mutation that constitutively activates the mitogen-activated protein kinase (MAPK) pathway, giving rise to indolent entities such as well-differentiated papillary or follicular thyroid carcinomas. This is followed by recurring alterations that drive oncogenic properties such as enhanced proliferation, genomic instability, replicative immortality, and dedifferentiation, culminating in the emergence of highly aggressive ATC tumors. The truncal MAPK-activating events present therapeutic opportunities, as small molecule inhibitors against key components of this pathway are available. Indeed, genotype-guided targeting of the MAPK pathway is now the standard of care for subgroups of ATC patients, and further efforts exploring additional MAPK inhibitors and the combination of immune checkpoint blockade with MAPK inhibition are overcoming resistance to the current targeted therapies in the clinic and expanding our arsenal against this disease. In this review, we summarize the current understanding of the genomic landscape of ATC, discuss the biological and clinical ramifications of recurring aberrations, and provide an overview of the opportunities and challenges in the clinical management of this lethal malignancy.

Haoyu Yang, Sanjoy Dey, Pablo Meyer

Modeling disease progression through multiple stages is critical for clinical decision-making for chronic diseases, e.g., cancer, diabetes, chronic kidney diseases, and so on. Existing approaches often model the disease progression as a uniform trajectory pattern at the population level. However, chronic diseases are highly heterogeneous and often have multiple progression patterns depending on a patient's individual genetics and environmental effects due to lifestyles. We propose a personalized disease progression model to jointly learn the heterogeneous progression patterns and groups of genetic profiles. In particular, an end-to-end pipeline is designed to simultaneously infer the characteristics of patients from genetic markers using a variational autoencoder and how it drives the disease progressions using an RNN-based state-space model based on clinical observations. Our proposed model shows improvement on real-world and synthetic clinical data.

Yunguo Yu

Automated clinical ICD-10 coding is a high-impact healthcare task requiring a balance between coverage, precision, and safety. While neural approaches achieve strong performance, they suffer from hallucination-generating invalid or unsupported codes-posing unacceptable risks in safety-critical clinical settings. Rule-based systems eliminate hallucination but lack scalability and coverage due to manual knowledge base (KB) curation. We present Hybrid-Code v2, a neuro-symbolic framework that achieves zero Type-I hallucination by construction while maintaining competitive coverage and precision. The system integrates neural candidate generation with a symbolic KB verification layer that enforces validity constraints through multi-layer verification, including format, evidence grounding, negation detection, temporal consistency, and exclusion rules. In addition, we introduce an automated KB expansion mechanism that extracts and validates coding patterns from unlabeled clinical text, addressing the scalability limitations of rule-based systems. Evaluated on the MIMIC-III dataset against ClinicalBERT, BioBERT, rule-based systems, and GPT-4, Hybrid-Code v2 achieves 85% coverage, 92% precision, and 0% Type-I hallucination, outperforming rule-based systems by +40% coverage while eliminating hallucination observed in neural baselines (6-18%). The proposed architecture provides a formal safety guarantee for syntactic validity while preserving strong empirical performance. These results demonstrate that neuro-symbolic verification can enforce safety constraints in neural medical AI systems without sacrificing effectiveness, offering a generalizable design pattern for deploying trustworthy AI in safety-critical domains.

Ziyuan Huang, Vishaldeep Kaur Sekhon, Roozbeh Sadeghian et al.

Alzheimer's Disease Analysis Model (ADAM) is a multi-agent reasoning large language model (LLM) framework designed to integrate and analyze multimodal data, including microbiome profiles, clinical datasets, and external knowledge bases, to enhance the understanding and classification of Alzheimer's disease (AD). By leveraging the agentic system with LLM, ADAM produces insights from diverse data sources and contextualizes the findings with literature-driven evidence. A comparative evaluation with XGBoost revealed a significantly improved mean F1 score and significantly reduced variance for ADAM, highlighting its robustness and consistency, particularly when utilizing human biological data. Although currently tailored for binary classification tasks with two data modalities, future iterations will aim to incorporate additional data types, such as neuroimaging and peripheral biomarkers, and expand them to predict disease progression, thereby broadening ADAM's scalability and applicability in AD research and diagnostic applications.

Benjamin D. Ballyk, Ankit Gupta, Sujay Konda et al.

Data privacy is a critical challenge in modern medical workflows as the adoption of electronic patient records has grown rapidly. Stringent data protection regulations limit access to clinical records for training and integrating machine learning models that have shown promise in improving diagnostic accuracy and personalized care outcomes. Synthetic data offers a promising alternative; however, current generative models either struggle with time-series data or lack formal privacy guaranties. In this paper, we enhance a state-of-the-art time-series generative model to better handle longitudinal clinical data while incorporating quantifiable privacy safeguards. Using real data from chronic kidney disease and ICU patients, we evaluate our method through statistical tests, a Train-on-Synthetic-Test-on-Real (TSTR) setup, and expert clinical review. Our non-private model (Augmented TimeGAN) outperforms transformer- and flow-based models on statistical metrics in several datasets, while our private model (DP-TimeGAN) maintains a mean authenticity of 0.778 on the CKD dataset, outperforming existing state-of-the-art models on the privacy-utility frontier. Both models achieve performance comparable to real data in clinician evaluations, providing robust input data necessary for developing models for complex chronic conditions without compromising data privacy.

Yuanyuan Tang, Zhaoxia Yang, Jinyang Wen et al.

BackgroundRight cardiac chamber remodeling is widespread in patients with connective tissue disease (CTD). Serum uric acid (SUA) is considered a potential independent risk factor for cardiovascular disease, and elevated SUA levels are often observed in patients with CTD. The correlation between SUA levels and right cardiac chamber remodeling remains unclear. This study investigated the association of SUA with right cardiac chamber remodeling as assessed by cardiac magnetic resonance feature-tracking (CMR-FT) in CTD patients.Methods and resultsIn this cross-sectional study, a total of 104 CTD patients and 52 age- and sex-matched controls were consecutively recruited. All individuals underwent CMR imaging, and their SUA levels were recorded. The patients were divided into three subgroups based on the tertiles of SUA level in the present study. CMR-FT was used to evaluate the right atrial (RA) longitudinal strain and strain rate parameters as well as right ventricular (RV) global systolic peak strain and strain rate in longitudinal and circumferential directions for each subject. Univariable and multivariable linear regression analyses were used to explore the association of SUA with RV and RA strain parameters. Compared with the controls, the CTD patients showed significantly higher SUA levels but a lower RV global circumferential strain (GCS) and RA phasic strain parameters (all p &lt; 0.05), except the RA booster strain rate. RV GCS remained impaired even in CTD patients with preserved RV ejection fraction. Among subgroups, the patients in the third tertile had significantly impaired RV longitudinal strain (GLS), RV GCS, and RA reservoir and conduit strain compared with those in the first tertile (all p &lt; 0.05). The SUA levels were negatively correlated with RV GLS and RV GCS as well as with RA reservoir and conduit strain and strain rates (the absolute values of β were 0.250 to 0.293, all P &lt; 0.05). In the multivariable linear regression analysis, the SUA level was still an independent determinant of RA conduit strain (β = -0.212, P = 0.035) and RV GCS (β = 0.207, P = 0.019).ConclusionSUA may be a potential risk factor of right cardiac chamber remodeling and is independently associated with impaired RA conduit strain and RV GCS in CTD patients.

Chaoyu Pang, Yongzheng Li, Ming Shi et al.

BackgroundCXC chemokine receptor 4 (CXCR4) is associated with the progression and metastasis of numerous malignant tumors. However, its relationship with Gastroenteropancreatic Neuroendocrine Neoplasms Grade 3 (GEP-NENs G3) is unclear. The aim of this study was to characterize the expression of CXCR4 in GEP-NENS and to explore the clinical and prognostic value of CXCR4.MethodsThis study retrospectively collected clinical and pathological data from patients with GEP-NENs who receiving surgery in Qilu Hospital of Shandong University from January 2013 to April 2021, and obtained the overall survival of the patients based on follow-up. Immunohistochemistry (IHC) was performed on pathological paraffin sections to observe CXCR4 staining. Groups were made according to pathological findings. Kaplan-Meier (K-M) curve was used to evaluate prognosis. SPSS 26.0 was used for statistical analysis.Results100 GEP-NENs G3 patients were enrolled in this study. There was a significant difference in primary sites (P=0.002), Ki-67 index (P&lt;0.001), and Carcinoembryonic Antigen (CEA) elevation (P=0.008) between neuroendocrine tumor (NET) G3 and neuroendocrine carcinoma (NEC). CXCR4 was highly expressed only in tumors, low or no expressed in adjacent tissues (P&lt;0.001). The expression level of CXCR4 in NEC was significantly higher than that in NET G3 (P=0.038). The K-M curves showed that there was no significant difference in overall survival between patients with high CXCR4 expression and patients with low CXCR4 expression, either in GEP-NEN G3 or NEC (P=0.920, P=0.842. respectively).ConclusionDifferential expression of CXCR4 was found between tumor and adjacent tissues and between NET G3 and NEC. Our results demonstrated that CXCR4 can be served as a new IHC diagnostic indicator in the diagnosis and differential diagnosis of GEP-NENs G3. Further studies with multi-center, large sample size and longer follow-up are needed to confirm the correlation between CXCR4 expression level and prognosis.

Agena Suzuki, Akinori Hayashi, Satoshi Oda et al.

The beneficial effects of sodium-glucose cotransporter 2 (SGLT2) inhibitors in people with type 2 diabetes (T2D) and metabolic dysfunction-associated steatotic liver disease (MASLD) have been suggested in several reports based on serological markers, imaging data, and histopathology associated with steatotic liver disease. However, evidence regarding their long-term effects is currently insufficient. In this retrospective observational study, 34 people with T2D and MASLD, treated with SGLT2 inhibitors, were examined by proton density fat fraction derived by magnetic resonance imaging (MRI-PDFF) and other clinical data before, one year after the treatment. Furthermore, 22 of 34 participants underwent MRI-PDFF five years after SGLT2 inhibitors were initiated. HbA1c decreased from 8.9 ± 1.8% to 7.8 ± 1.0% at 1 year (p = 0.006) and 8.0 ± 1.1% at 5 years (p = 0.122). Body weight and fat mass significantly reduced from baseline to 1 and 5 year(s), respectively. MRI-PDFF significantly decreased from 15.3 ± 7.8% at baseline to 11.9 ± 7.6% (p = 0.001) at 1 year and further decreased to 11.3 ± 5.7% (p = 0.013) at 5 years. Thus, a 5-year observation demonstrated that SGLT2 inhibitors have beneficial effects on liver steatosis in people with T2D and MASLD.

Yuanyuan Zhang, Huaizhen Liu

Abstract Background Insulin resistance (IR) was a prominent feature commonly observed in individuals with type 2 diabetes mellitus (T2DM). T2DM Individuals often exhibited a concomitant presence of low-grade chronic inflammation. In this study conducted retrospectively, the aim was to investigate the connection between neutrophils-lymphocytes rate (NLR), monocytes-lymphocytes rate (MLR), platelets-lymphocytes rate (PLR) and IR, specifically among individuals with T2DM. Method This study encompassed a cohort of 405 individuals diagnosed with T2DM, comprising cases from January 2021 to November 2022. On the basis of whether there was IR or not, these sufferers were categorized into two cohorts, namely T2DM with IR group (292 cases) and T2DM without IR group (113 cases), as determined by a homeostasis model assessment-IR (HOMA-IR) value exceeding 2.0. Results The findings of this study demonstrated compelling evidence of distinct biomarker profiles between individuals with T2DM who had IR and those without IR. Specifically, the IR individuals displayed notably raise NLR, MLR, PLR, C reactive protein (CRP) and serum amyloid A (SAA). Additionally, there was a noticeable decrease in superoxide dismutase (SOD) levels. Furthermore, IR was negatively correlated with SOD values, while positive associations were found between IR and NLR, CRP, and SAA levels (p < 0.05). Moreover, a rise in NLR and PLR levels demonstrated an identical relationship with the prevalence of IR (p = 0.007, p = 0.025, separately). The Receiver operating characteristic (ROC) curve demonstrated that the areas under the curve (AUC) for NLR, MLR, PLR, CRP, SAA and SOD in predicting occurrence of IR in T2DM patients were 0.603, 0.575, 0.581, 0.644, 0.594 and 0.632 respectively, with sensitivity of 79.5%, 95.2%, 46.9%,54.1% (or 51.4), 47.6% (or 45.7%) and 98.6% and specificity of 37.2%, 19.5%, 69.9%, 69% (or 71.7%), 71.6% (or 73.5%) and 23% respectively. Conclusion Our findings support the notion that higher magnitude of NLR, PLR, MLR, CRP, and SAA values, corresponded to lower SOD levels, indicating a more severe degree of IR in T2DM patients. Additionally, NLR, PLR, MLR, CRP, SAA, and SOD demonstrated predictive potential for assessing IR. Regrettably, due to the retrospective nature of this study, it was not feasible to take a measurement the majority of inflammatory factors and reactive oxygen species (ROS).

Magnus F G Grøndahl, Jonatan I Bagger, Malte P Suppli et al.

Objective: In obesity and type 2 diabetes, hyperglucagonaemia may be caused by elevated levels of glucagonotropic amino acids due to hepatic glucagon resistance at the level of amino acid turnover. Here, we investigated the effect of exogenous glucagon on circulating amino acids in obese and non-obese individuals with and without type 2 diabetes. Design: This was a post hoc analysis in a glucagon infusion study performed in individuals with type 2 diabetes (n = 16) and in age, sex, and body mass index-matched control individuals without diabetes (n = 16). Each group comprised two subgroups of eight individuals with and without obesity, respectively. Methods: All participants received a 1-h glucagon infusion (4 ng/kg/min) in the overnight fasted state. Plasma amino acid concentrations were measured with frequent intervals. Results: Compared to the control subgroup without obesity, baseline total amino acid levels were elevated in the control subgroup with obesity and in the type 2 diabetes subgroup without obesity. In all subgroups, amino acid levels decreased by up to 20% in response to glucagon infusion, which resulted in high physiological steady-state glucagon levels (mean concentration: 74 pmol/L, 95% CI [68;79] pmol/L). Following correction for multiple testing, no intergroup differences in changes in amino acid levels reached significance. Conclusion: Obesity and type 2 diabetes status was associated with elevated fasting levels of total amino acids. The glucagon infusion decreased circulating amino acid levels similarly in all subgroups, without significant differences in the response to exogenous glucagon between individuals with and without obesity and type 2 diabetes. Significance statement The hormone glucagon stimulates glucose production from the liver, which may promote hyperglycaemia if glucagon levels are abnormally elevated, as is often seen in type 2 diabetes and obesity. Glucagon levels are closely linked to, and influenced by, the levels of circulating amino acids. To further investigate this link, we measured amino acid levels in individuals with and without obesity and type 2 diabetes before and during an infusion of glucagon. We found that circulating amino acid levels were higher in type 2 diabetes and obesity, and that glucagon infusion decreased amino acid levels in both individuals with and without type 2 diabetes and obesity. The study adds novel information to the link between circulating levels of glucagon and amino acids.

Kai Kang, Yuqi Gu

Current diagnosis and prognosis for Parkinson's disease (PD) face formidable challenges due to the heterogeneous nature of the disease course, including that (i) the impairment severity varies hugely between patients, (ii) whether a symptom occur independently or co-occurs with related symptoms differs significantly, and (iii) repeated symptom measurements exhibit substantial temporal dependence. To tackle these challenges, we propose a novel blockwise mixed membership model (BM3) to systematically unveil between-patient, between-symptom, and between-time clinical heterogeneity within PD. The key idea behind BM3 is to partition multivariate longitudinal measurements into distinct blocks, enabling measurements within each block to share a common latent membership while allowing latent memberships to vary across blocks. Consequently, the heterogeneous PD-related measurements across time are divided into clinically homogeneous blocks consisting of correlated symptoms and consecutive time. From the analysis of Parkinson's Progression Markers Initiative data (n=1,531), we discover three typical disease profiles (stages), four symptom groups (i.e., autonomic function, tremor, left-side and right-side motor function), and two periods, advancing the comprehension of PD heterogeneity. Moreover, we identify several clinically meaningful PD subtypes by summarizing the blockwise latent memberships, paving the way for developing more precise and targeted therapies to benefit patients. Our findings are validated using external variables, successfully reproduced in validation datasets, and compared with existing methods. Theoretical results of model identifiability further ensures the reliability and reproducibility of latent structure discovery in PD.

Xuanzhong Chen, Xiaohao Mao, Qihan Guo et al.

Generalist Large Language Models (LLMs), such as GPT-4, have shown considerable promise in various domains, including medical diagnosis. Rare diseases, affecting approximately 300 million people worldwide, often have unsatisfactory clinical diagnosis rates primarily due to a lack of experienced physicians and the complexity of differentiating among many rare diseases. In this context, recent news such as "ChatGPT correctly diagnosed a 4-year-old's rare disease after 17 doctors failed" underscore LLMs' potential, yet underexplored, role in clinically diagnosing rare diseases. To bridge this research gap, we introduce RareBench, a pioneering benchmark designed to systematically evaluate the capabilities of LLMs on 4 critical dimensions within the realm of rare diseases. Meanwhile, we have compiled the largest open-source dataset on rare disease patients, establishing a benchmark for future studies in this domain. To facilitate differential diagnosis of rare diseases, we develop a dynamic few-shot prompt methodology, leveraging a comprehensive rare disease knowledge graph synthesized from multiple knowledge bases, significantly enhancing LLMs' diagnostic performance. Moreover, we present an exhaustive comparative study of GPT-4's diagnostic capabilities against those of specialist physicians. Our experimental findings underscore the promising potential of integrating LLMs into the clinical diagnostic process for rare diseases. This paves the way for exciting possibilities in future advancements in this field.

Shengjun Zhu, Siyu Liu, Yang Li et al.

Coronavirus Disease 2019 (COVID-19), which emerged in 2019, has caused millions of deaths worldwide. Although effective vaccines have been developed to mitigate severe symptoms, certain populations, particularly the elderly and those with comorbidities, remain at high risk for severe outcomes and increased mortality. Consequently, early identification of the severity and clinical outcomes of the disease in these patients is vital to prevent adverse prognoses. Although traditional machine learning and deep learning models have been widely employed in this area, the potential of large language models (LLMs) remains largely unexplored. Our research focuses primarily on constructing specialized prompts and adopting multi-objective learning strategies. We started by selecting serological indicators that significantly correlate with clinical outcomes and disease severity to serve as input data for the model. Blood test samples often contain numerous missing values, and traditional models generally rely on imputation to handle these gaps in the data. In contrast, LLMs offer the advantage of robust semantic understanding. By setting prompts, we can explicitly inform the model when a feature's value is missing, without the need for imputation. For the multi-objective learning strategy, the model is designed to first predict disease severity and then predict clinical outcomes. Given that LLMs utilize both the input text and the generated tokens as input for generating the next token, the predicted severity is used as a basis for generating the clinical outcome. During the fine-tuning of the LLM, the two objectives influence and improve each other. Our experiments were implemented based on the ChatGLM model. The results demonstrate the effectiveness of LLMs in this task, suggesting promising potential for further development.

Francesco Pallotti, Marianna Pelloni, Stefano Colangelo et al.