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

Jing Bai, Jirui Cai, Zhiwei Huang et al.

BackgroundCommunity screening programs increasingly use World Health Organization (WHO) cardiovascular disease (CVD) risk charts to identify individuals at high predicted 10-year risk. The atherogenic index of plasma (AIP), derived from triglycerides (TG) and high-density lipoprotein cholesterol (HDL-C), may capture atherogenic dyslipidemia and support pragmatic risk stratification.MethodsWe conducted a cross-sectional analysis of baseline data from the China Health Evaluation And risk Reduction through nationwide Teamwork (ChinaHEART) community screening program in Luohe, China. Among 6,860 screened participants, 6,702 with complete data for AIP computation, WHO risk classification, and prespecified covariates were included. The outcome was the WHO CVD risk chart-defined predicted 10-year CVD high-risk category (high risk: ≥20%), rather than adjudicated or incident CVD events. AIP was calculated as log10(TG [mmol/L]/HDL-C [mmol/L]) and modeled as both a continuous and categorical exposure; spline models tested nonlinearity, and ROC analyses evaluated discrimination and derived a Youden-index cutoff. In addition, we performed an explainable machine-learning pipeline for CVD high-risk prediction using LASSO logistic regression for feature selection (AIP forced-in), followed by a random forest classifier and SHAP-based interpretation.ResultsOf 6,860 screened participants, 6,702 were included in the analytic sample (median age 58 years; 38% men). The WHO CVD risk chart-defined predicted 10-year CVD high-risk category was present in 1,440 (21%) participants and was more frequent in the high-AIP group than in the low-AIP group. Higher AIP was associated with higher odds of CVD high-risk status. Restricted cubic splines supported a non-linear association. Discrimination was modest for AIP alone (AUC 0.557) and improved in adjusted models (AUC 0.650). In the machine-learning pipeline (LASSO + random forest), the random forest model achieved an AUC of 0.792, and SHAP analyses ranked LDL-C and history of hypertension as the strongest contributors, with AIP remaining among the top predictive features.ConclusionIn this community-based ChinaHEART population, higher AIP was non-linearly associated with the WHO CVD risk chart-defined predicted 10-year CVD high-risk category. Although AIP alone had limited discrimination, it may serve as a simple adjunct marker to triage individuals for intensified risk assessment in primary-care screening settings.

Rongjing Song, Wei Zhao, Shasha Li et al.

BackgroundThyroid eye disease (TED) is a sight-threatening autoimmune disorder with limited effective therapies. Teprotumumab, an insulin-like growth factor-1 receptor inhibitor, has emerged as a promising treatment. However, a comprehensive synthesis of its efficacy and safety across randomized trials remains limited.MethodsA systematic review and meta-analysis of randomized controlled trials (RCTs) comparing teprotumumab with placebo in TED was conducted. Primary outcomes included proptosis response, overall response, change in proptosis, diplopia response, achievement of a Clinical Activity Score (CAS) ≤1, changes in Graves’ ophthalmopathy–specific quality-of-life questionnaire (GO-QOL) scores and safety outcomes. Pooled risk ratios (RRs) and mean differences (MDs) with 95% confidence intervals (CIs) were calculated using random-effects models.ResultsSeven RCTs involving 438 participants were included. Teprotumumab significantly improved all efficacy outcomes: proptosis response (RR, 6.87; 95% CI, 3.32 to 14.24), overall response (RR, 7.82; 95% CI, 3.36 to 18.18), reduction in proptosis (MD, -2.46 mm; 95% CI, -2.96 to -1.96), diplopia response (RR, 1.85; 95% CI, 1.28 to 2.68), CAS ≤1 (RR, 3.39; 95% CI, 2.41 to 4.78) and increase in GO-QOL overall score (MD, 10.87; 95% CI, 9.91 to 11.83). Safety analysis indicated elevated risks of hyperglycemia (RR, 2.82; 95% CI, 1.08 to 7.37), muscle spasms (RR, 3.83; 95% CI, 1.97 to 7.43), dry skin (RR, 6.54; 95% CI, 1.52 to 28.09), and hearing impairment (RR, 3.74; 95% CI, 1.26 to 11.13).ConclusionsTeprotumumab provides substantial, consistent benefits in improving proptosis, diplopia, disease activity and GO-QOL in TED. Clinicians should monitor for adverse events, particularly hyperglycemia and hearing impairment. These findings reinforce teprotumumab as a pivotal therapeutic option and support balanced risk-benefit evaluation.

James L. Li

Background/Objectives: Endocrine hormones play critical roles in regulating physiological processes, and previous studies have reported their associations with psychiatric disorders. Levels of endocrine hormones and the risk of developing psychiatric disorders are influenced by both genetic and non-genetic factors. However, the shared genetic basis underlying these associations remains largely unexplored. This study aims to dually evaluate the genetic correlations among endocrine hormones, including thyroid and sex hormones, as well as between endocrine hormone metrics and psychiatric disorders to identify potential shared genetic architectures. Methods: We obtained genome-wide association study summary statistics for six thyroid hormone metrics, three sex hormone metrics, and ten psychiatric disorders from predominantly European-ancestry populations. Genetic correlations were computed using linkage disequilibrium score regression after harmonizing variant data to ensure consistency across studies. Results: Significant genetic correlations were observed among thyroid and sex hormone metrics, indicating a strong shared genetic basis. Sex hormones exhibited multiple genetic correlations with psychiatric disorders, including negative correlations between sex hormone-binding globulin and attention-deficit hyperactivity disorder (ADHD) (<i>p</i> = 3.95 × 10⁻¹²) and major depressive disorder (<i>p</i> = 4.67 × 10⁻⁵), and positive genetic correlations with anorexia nervosa (<i>p</i> = 2.86 × 10⁻¹²) and schizophrenia (<i>p</i> = 2.00 × 10⁻⁴). Testosterone and estradiol had negative genetic correlations with ADHD and major depressive disorder, while testosterone had positive genetic correlations with anorexia nervosa and schizophrenia. Although thyroid hormone metrics did not exhibit Bonferroni-significant genetic correlations, nominal associations were observed, such as a negative genetic correlation between thyroid-stimulating hormone and major depressive disorder (<i>p</i> = 2.33 × 10⁻²). Conclusions: These findings suggest a shared genetic basis between endocrine hormones and psychiatric disorders, particularly for sex hormones. Future studies leveraging larger, more diverse populations are warranted to validate and extend the genetic correlations observed in this study.

Yuqi Tang, Jing Yu, Zichang Su et al.

Clinical diagnosis begins with doctor-patient interaction, during which physicians iteratively gather information, determine examination and refine differential diagnosis through patients' response. This dynamic clinical-reasoning process is poorly represented by existing LLM benchmarks that focus on static question-answering. To mitigate these gaps, recent methods explore dynamic medical frameworks involving interactive clinical dialogues. Although effective, they often rely on limited, contamination-prone datasets and lack granular, multi-level evaluation. In this work, we propose ClinDEF, a dynamic framework for assessing clinical reasoning in LLMs through simulated diagnostic dialogues. Grounded in a disease knowledge graph, our method dynamically generates patient cases and facilitates multi-turn interactions between an LLM-based doctor and an automated patient agent. Our evaluation protocol goes beyond diagnostic accuracy by incorporating fine-grained efficiency analysis and rubric-based assessment of diagnostic quality. Experiments show that ClinDEF effectively exposes critical clinical reasoning gaps in state-of-the-art LLMs, offering a more nuanced and clinically meaningful evaluation paradigm.

Luis H. John, Jan A. Kors, Jenna M. Reps et al.

Background: Existing clinical prediction models often represent patient data using features that ignore the semantic relationships between clinical concepts. This study integrates domain-specific semantic information by mapping the SNOMED medical term hierarchy into a low-dimensional hyperbolic space using Poincaré embeddings, with the aim of improving lung cancer onset prediction. Methods: Using a retrospective cohort from the Optum EHR dataset, we derived a clinical knowledge graph from the SNOMED taxonomy and generated Poincaré embeddings via Riemannian stochastic gradient descent. These embeddings were then incorporated into two deep learning architectures, a ResNet and a Transformer model. Models were evaluated for discrimination (area under the receiver operating characteristic curve) and calibration (average absolute difference between observed and predicted probabilities) performance. Results: Incorporating pre-trained Poincaré embeddings resulted in modest and consistent improvements in discrimination performance compared to baseline models using randomly initialized Euclidean embeddings. ResNet models, particularly those using a 10-dimensional Poincaré embedding, showed enhanced calibration, whereas Transformer models maintained stable calibration across configurations. Discussion: Embedding clinical knowledge graphs into hyperbolic space and integrating these representations into deep learning models can improve lung cancer onset prediction by preserving the hierarchical structure of clinical terminologies used for prediction. This approach demonstrates a feasible method for combining data-driven feature extraction with established clinical knowledge.

Jean Seo, Hyunkyung Lee, Gibaeg Kim et al.

Safety is a paramount concern in clinical chatbot applications, where inaccurate or harmful responses can lead to serious consequences. Existing methods--such as guardrails and tool calling--often fall short in addressing the nuanced demands of the clinical domain. In this paper, we introduce TACOS (TAxonomy of COmprehensive Safety for Clinical Agents), a fine-grained, 21-class taxonomy that integrates safety filtering and tool selection into a single user intent classification step. TACOS is a taxonomy that can cover a wide spectrum of clinical and non-clinical queries, explicitly modeling varying safety thresholds and external tool dependencies. To validate our taxonomy, we curate a TACOS-annotated dataset and perform extensive experiments. Our results demonstrate the value of a new taxonomy specialized for clinical agent settings, and reveal useful insights about train data distribution and pretrained knowledge of base models.

Pedro Marques, Inês Sapinho, Márta Korbonits

AbstractObjectivePseudoacromegaly encompasses conditions with features of acromegaly/gigantism, but no growth hormone (GH) or insulin‐like growth factor‐1 (IGF‐1) excess. We aimed to review published pseudoacromegaly cases evaluated due to clinical suspicion of acromegaly.Design/PatientsPubMed/Medline search was conducted to identify reported pseudoacromegaly cases, which were systematically reviewed to ensure they met eligibility criteria: (1) presentation suggestive of acromegaly; (2) acromegaly excluded based on normal GH, IGF‐1 and/or GH suppression on oral glucose tolerance test (OGTT‐GH); (3) diagnosis of the pseudoacromegaly condition was established. Data were retrieved from each case and analysed collectively.ResultsOf 76 cases, 47 were males, mean ages at presentation and at first acromegaloid symptoms were 28 ± 16 and 17 ± 10 years, respectively. Most common conditions were pachydermoperiostosis (47%) and insulin‐mediated pseudoacromegaly (IMP) (24%). Acromegaloid facies (75%) and acral enlargement (80%) were the most common features. Measurement of random GH was reported in 65%, IGF‐1 in 79%, OGTT‐GH in 51%. GH excess was more frequently excluded based on two tests (53%). Magnetic resonance imaging (MRI) was performed in 30 patients, with pituitary adenoma or hyperplasia being reported in eight and three patients, respectively. Investigations differed between cases managed by endocrine and non‐endocrine specialists, the former requesting more often IGF‐1, OGTT‐GH and pituitary MRI.ConclusionsPseudoacromegaly is a challenging entity that may be encountered by endocrinologists. Pachydermoperiostosis and IMP are the conditions most often mimicking acromegaly. Adequate assessment of GH/IGF‐1 is crucial to exclude acromegaly, which may be better performed by endocrinologists. Pituitary incidentalomas are common and require careful judgement to prevent unnecessary pituitary surgery.

Xueliang Lu, Xueliang Lu, Xu Wang et al.

ObjectivesIn-depth understanding of osteonecrosis of femoral head (ONFH) has revealed that degeneration of the hip cartilage plays a crucial role in ONFH progression. However, the underlying molecular mechanisms and susceptibility to environmental factors in hip cartilage that contribute to ONFH progression remain elusive.MethodsWe conducted a multiomics study and chemical−gene interaction analysis of hip cartilage in ONFH. The differentially expressed genes (DEGs) involved in ONFH progression were identified in paired hip cartilage samples from 36 patients by combining genome-wide DNA methylation profiling, gene expression profiling, and quantitative proteomics. Gene functional enrichment and pathway analyses were performed via Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses. Functional links between proteins were discovered through protein−protein interaction (PPI) networks. The ONFH-associated chemicals were identified by integrating the DEGs with the chemical−gene interaction sets in the Comparative Toxicogenomics Database (CTD). Finally, the DEGs, including MMP13 and CHI3L1, were validated via quantitative real-time PCR (qRT−PCR) and immunohistochemistry (IHC).ResultsTwenty-two DEGs were identified across all three omics levels in ONFH cartilage, 16 of which were upregulated and six of which were downregulated. The collagen-containing extracellular matrix (ECM), ECM structural constituents, response to amino acids, the relaxin signaling pathway, and protein digestion and absorption were found to be primarily involved in cartilage degeneration in ONFH. Moreover, ten major ONFH-associated chemicals were identified, including, benzo(a)pyrene, valproic acid, and bisphenol A.ConclusionOverall, our study identified several candidate genes, pathways, and chemicals associated with cartilage degeneration in ONFH, providing novel clues into the etiology and biological processes of ONFH progression.

Zihan Zhou, Ziyi Zeng, Wenhao Jiang et al.

As social changes accelerate, the incidence of psychosomatic disorders has significantly increased, becoming a major challenge in global health issues. This necessitates an innovative knowledge system and analytical methods to aid in diagnosis and treatment. Here, we establish the ontology model and entity types, using the BERT model and LoRA-tuned LLM for named entity recognition, constructing the knowledge graph with 9668 triples. Next, by analyzing the network distances between disease, symptom, and drug modules, it was found that closer network distances among diseases can predict greater similarities in their clinical manifestations, treatment approaches, and psychological mechanisms, and closer distances between symptoms indicate that they are more likely to co-occur. Lastly, by comparing the proximity d and proximity z score, it was shown that symptom-disease pairs in primary diagnostic relationships have a stronger association and are of higher referential value than those in diagnostic relationships. The research results revealed the potential connections between diseases, co-occurring symptoms, and similarities in treatment strategies, providing new perspectives for the diagnosis and treatment of psychosomatic disorders and valuable information for future mental health research and practice.

Guanchu Wang, Junhao Ran, Ruixiang Tang et al.

Despite the impressive capabilities of Large Language Models (LLMs) in general medical domains, questions remain about their performance in diagnosing rare diseases. To answer this question, we aim to assess the diagnostic performance of LLMs in rare diseases, and explore methods to enhance their effectiveness in this area. In this work, we introduce a rare disease question-answering (ReDis-QA) dataset to evaluate the performance of LLMs in diagnosing rare diseases. Specifically, we collected 1360 high-quality question-answer pairs within the ReDis-QA dataset, covering 205 rare diseases. Additionally, we annotated meta-data for each question, facilitating the extraction of subsets specific to any given disease and its property. Based on the ReDis-QA dataset, we benchmarked several open-source LLMs, revealing that diagnosing rare diseases remains a significant challenge for these models. To facilitate retrieval augmentation generation for rare disease diagnosis, we collect the first rare diseases corpus (ReCOP), sourced from the National Organization for Rare Disorders (NORD) database. Specifically, we split the report of each rare disease into multiple chunks, each representing a different property of the disease, including their overview, symptoms, causes, effects, related disorders, diagnosis, and standard therapies. This structure ensures that the information within each chunk aligns consistently with a question. Experiment results demonstrate that ReCOP can effectively improve the accuracy of LLMs on the ReDis-QA dataset by an average of 8%. Moreover, it significantly guides LLMs to generate trustworthy answers and explanations that can be traced back to existing literature.

Juan Miguel Lopez Alcaraz, Wilhelm Haverkamp, Nils Strodthoff

Background: Liver diseases present a significant global health challenge and often require costly, invasive diagnostics. Electrocardiography (ECG), a widely available and non-invasive tool, can enable the detection of liver disease by capturing cardiovascular-hepatic interactions. Methods: We trained tree-based machine learning models on ECG features to detect liver diseases using two large datasets: MIMIC-IV-ECG (467,729 patients, 2008-2019) and ECG-View II (775,535 patients, 1994-2013). The task was framed as binary classification, with performance evaluated via the area under the receiver operating characteristic curve (AUROC). To improve interpretability, we applied explainability methods to identify key predictive features. Findings: The models showed strong predictive performance with good generalizability. For example, AUROCs for alcoholic liver disease (K70) were 0.8025 (95% confidence interval (CI), 0.8020-0.8035) internally and 0.7644 (95% CI, 0.7641-0.7649) externally; for hepatic failure (K72), scores were 0.7404 (95% CI, 0.7389-0.7415) and 0.7498 (95% CI, 0.7494-0.7509), respectively. The explainability analysis consistently identified age and prolonged QTc intervals (corrected QT, reflecting ventricular repolarization) as key predictors. Features linked to autonomic regulation and electrical conduction abnormalities were also prominent, supporting known cardiovascular-liver connections and suggesting QTc as a potential biomarker. Interpretation: ECG-based machine learning offers a promising, interpretable approach for liver disease detection, particularly in resource-limited settings. By revealing clinically relevant biomarkers, this method supports non-invasive diagnostics, early detection, and risk stratification prior to targeted clinical assessments.

Amirali Hariri, Mina Mirian, Ali Zarrabi et al.

Type 2 Diabetes Mellitus (T2DM) has been the main category of metabolic diseases in recent years due to changes in lifestyle and environmental conditions such as diet and physical activity. On the other hand, the circadian rhythm is one of the most significant biological pathways in humans and other mammals, which is affected by light, sleep, and human activity. However, this cycle is controlled via complicated cellular pathways with feedback loops. It is widely known that changes in the circadian rhythm can alter some metabolic pathways of body cells and could affect the treatment process, particularly for metabolic diseases like T2DM. The aim of this study is to explore the importance of the circadian rhythm in the occurrence of T2DM via reviewing the metabolic pathways involved, their relationship with the circadian rhythm from two perspectives, lifestyle and molecular pathways, and their effect on T2DM pathophysiology. These impacts have been demonstrated in a variety of studies and led to the development of approaches such as time-restricted feeding, chronotherapy (time-specific therapies), and circadian molecule stabilizers.

J. A. Henry, I. Abdesselam, O. Deal et al.

IntroductionObesity affects cardiac geometry, causing both eccentric (due to increased cardiac output) and concentric (due to insulin resistance) remodelling. Following bariatric surgery, reversal of both processes should occur. Furthermore, epicardial adipose tissue loss following bariatric surgery may reduce pericardial restraint, allowing further chamber expansion. We investigated these changes in a serial imaging study of adipose depots and cardiac geometry following bariatric surgery.Methods62 patients underwent cardiac magnetic resonance (CMR) before and after bariatric surgery, including 36 with short-term (median 212 days), 37 medium-term (median 428 days) and 32 long-term (median 1030 days) follow-up. CMR was used to assess cardiac geometry (left atrial volume (LAV) and left ventricular end-diastolic volume (LVEDV)), LV mass (LVM) and LV eccentricity index (LVei – a marker of pericardial restraint). Abdominal visceral (VAT) and epicardial (EAT) adipose tissue were also measured.ResultsPatients on average had lost 21kg (38.9% excess weight loss, EWL) at 212 days and 36kg (64.7% EWL) at 1030 days following bariatric surgery. Most VAT and EAT loss (43% and 14%, p&lt;0.0001) occurred within the first 212 days, with non-significant reductions thereafter. In the short-term LVM (7.4%), LVEDV (8.6%) and LAV (13%) all decreased (all p&lt;0.0001), with change in cardiac output correlated with LVEDV (r=0.35,p=0.03) and LAV change (r=0.37,p=0.03). Whereas LVM continued to decrease with time (12% decrease relative to baseline at 1030 days, p&lt;0.0001), both LAV and LVEDV had returned to baseline by 1030 days. LV mass:volume ratio (a marker of concentric hypertrophy) reached its nadir at the longest timepoint (p&lt;0.001). At baseline, LVei correlated with baseline EAT (r=0.37,p=0.0040), and decreased significantly from 1.09 at baseline to a low of 1.04 at 428 days (p&lt;0.0001). Furthermore, change in EAT following bariatric surgery correlated with change in LVei (r=0.43,p=0.0007).ConclusionsCardiac volumes show a biphasic response to weight loss, initially becoming smaller and then returning to pre-operative sizes by 1030 days. We propose this is due to an initial reversal of eccentric remodelling followed by reversal of concentric remodelling. Furthermore, we provide evidence for a role of EAT contributing to pericardial restraint, with EAT loss improving markers of pericardial restraint.

Peter N. Taylor, Andrew Lansdown, Justyna Witczak et al.

Abstract Background Thyroid hormones are key determinants of health and well-being. Normal thyroid function is defined according to the standard 95% confidence interval of the disease-free population. Such standard laboratory reference intervals are widely applied in research and clinical practice, irrespective of age. However, thyroid hormones vary with age and current reference intervals may not be appropriate across all age groups. In this review, we summarize the recent literature on age-related variation in thyroid function and discuss important implications of such variation for research and clinical practice. Main text There is now substantial evidence that normal thyroid status changes with age throughout the course of life. Thyroid stimulating hormone (TSH) concentrations are higher at the extremes of life and show a U-shaped longitudinal trend in iodine sufficient Caucasian populations. Free triiodothyronine (FT3) levels fall with age and appear to play a role in pubertal development, during which it shows a strong relationship with fat mass. Furthermore, the aging process exerts differential effects on the health consequences of thyroid hormone variations. Older individuals with declining thyroid function appear to have survival advantages compared to individuals with normal or high-normal thyroid function. In contrast younger or middle-aged individuals with low-normal thyroid function suffer an increased risk of adverse cardiovascular and metabolic outcomes while those with high-normal function have adverse bone outcomes including osteoporosis and fractures. Conclusion Thyroid hormone reference intervals have differential effects across age groups. Current reference ranges could potentially lead to inappropriate treatment in older individuals but on the other hand could result in missed opportunities for risk factor modification in the younger and middle-aged groups. Further studies are now needed to determine the validity of age-appropriate reference intervals and to understand the impact of thyroid hormone variations in younger individuals.

W. Colin Duncan, Linda M. Nicol, Rosie O’Hare et al.

IntroductionPolycystic ovary syndrome (PCOS) seems to be associated with increased ovarian sympathetic nerve activity and in rodent models of PCOS reducing the sympathetic drive to the ovary, through denervation or neuromodulation, improves ovulation rate. We hypothesised that sympathetic nerves work with gonadotropins to promote development and survival of small antral follicles to develop a polycystic ovary phenotype.MethodsUsing a clinically realistic ovine model we showed a rich sympathetic innervation to the normal ovary and reinnervation after ovarian transplantation. Using needlepoint diathermy to the nerve plexus in the ovarian vascular pedicle we were able to denervate the ovary resulting in reduced intraovarian noradrenaline and tyrosine hydroxylase immunostained sympathetic nerves. We developed an acute polycystic ovary (PCO) model using gonadotrophin releasing hormone (GnRH) agonist followed infusion of follicle stimulating hormone (FSH) with increased pulsatile luteinising hormone (LH). This resulted in increased numbers of smaller antral follicles in the ovary when compared to FSH infusion suggesting a polycystic ovary.ResultsDenervation had no effect of the survival or numbers of follicles in the acute PCO model and did not impact on ovulation, follicular and luteal hormone profiles in a normal cycle.DiscussionAlthough the ovary is richly inervated we did not find evidence for a role of sympathetic nerves in ovarian function or small follicle growth and survival.

Lehan Wang, Weihang Dai, Mei Jin et al.

Optical Coherence Tomography (OCT) is a novel and effective screening tool for ophthalmic examination. Since collecting OCT images is relatively more expensive than fundus photographs, existing methods use multi-modal learning to complement limited OCT data with additional context from fundus images. However, the multi-modal framework requires eye-paired datasets of both modalities, which is impractical for clinical use. To address this problem, we propose a novel fundus-enhanced disease-aware distillation model (FDDM), for retinal disease classification from OCT images. Our framework enhances the OCT model during training by utilizing unpaired fundus images and does not require the use of fundus images during testing, which greatly improves the practicality and efficiency of our method for clinical use. Specifically, we propose a novel class prototype matching to distill disease-related information from the fundus model to the OCT model and a novel class similarity alignment to enforce consistency between disease distribution of both modalities. Experimental results show that our proposed approach outperforms single-modal, multi-modal, and state-of-the-art distillation methods for retinal disease classification. Code is available at https://github.com/xmed-lab/FDDM.

Brihat Sharma, Yanjun Gao, Timothy Miller et al.

Generative artificial intelligence (AI) is a promising direction for augmenting clinical diagnostic decision support and reducing diagnostic errors, a leading contributor to medical errors. To further the development of clinical AI systems, the Diagnostic Reasoning Benchmark (DR.BENCH) was introduced as a comprehensive generative AI framework, comprised of six tasks representing key components in clinical reasoning. We present a comparative analysis of in-domain versus out-of-domain language models as well as multi-task versus single task training with a focus on the problem summarization task in DR.BENCH (Gao et al., 2023). We demonstrate that a multi-task, clinically trained language model outperforms its general domain counterpart by a large margin, establishing a new state-of-the-art performance, with a ROUGE-L score of 28.55. This research underscores the value of domain-specific training for optimizing clinical diagnostic reasoning tasks.

Nikita Mehandru, Brenda Y. Miao, Eduardo Rodriguez Almaraz et al.

Recent developments in large language models (LLMs) have unlocked new opportunities for healthcare, from information synthesis to clinical decision support. These new LLMs are not just capable of modeling language, but can also act as intelligent "agents" that interact with stakeholders in open-ended conversations and even influence clinical decision-making. Rather than relying on benchmarks that measure a model's ability to process clinical data or answer standardized test questions, LLM agents should be assessed for their performance on real-world clinical tasks. These new evaluation frameworks, which we call "Artificial-intelligence Structured Clinical Examinations" ("AI-SCI"), can draw from comparable technologies where machines operate with varying degrees of self-governance, such as self-driving cars. High-fidelity simulations may also be used to evaluate interactions between users and LLMs within a clinical workflow, or to model the dynamic interactions of multiple LLMs. Developing these robust, real-world clinical evaluations will be crucial towards deploying LLM agents into healthcare.

Konrad Furmańczyk, Wojciech Niemiro, Mariola Chrzanowska et al.

We propose a new graphical model to describe the comorbidity of allergic diseases. We present our model in two versions. First, we introduce a generative model that correctly reflects the variables' causal relationship. Then we propose an approximation of the generative model by another misspecified model that is computationally more efficient and easily interpretable. We will focus on the misspecified version, which we consider more practical. We include in the model two directed graphs, one graph of known dependency between the main binary variables (diseases), and a second graph of the dependence between the occurrence of the diseases and their symptoms. In the model, we also consider additional auxiliary variables. The proposed model is evaluated on a cross-sectional multicentre study in Poland on the ECAP database (www.ecap.pl). An assessment of the stability of the proposed model was obtained using bootstrap and jackknife techniques.

Wenkui Ma, Hao Wu, Guangdong Li et al.

Melatonin is an indole-like neuroendocrine hormone. A large number of studies have shown that melatonin can improve production performance of ewes, but it is not clear in lambs. In this study, the growth and development of the 2-month-old lambs implanted with melatonin were monitored for 60 days. The results showed that the growth rate of body weight and body skew length of lambs with melatonin treatment were significantly improved compared to the controls. The similar results were also observed in red blood cell count, hematocrit, red blood cell volume distribution width, the levels of growth hormone, testosterone, immunoglobulin A, immunoglobulin M and albumin. In addition, the cross sectional area of muscle fibers and adipose cells of lambs with melatonin implantation were also significantly increased compared to the controls (P&lt;0.05). To further explore the potential mechanisms, the muscle and adipose tissue were selected for transcriptome sequencing. KEGG enrichment results showed that melatonin regulated the expression of genes related to apoptotic signaling pathway in muscle and adipocytes. Since the intestinal microbiota are involved in the nutritional balance and animal growth, the 16SrRNA sequencing related to the intestinal microbiota was also performed. The data indicated that the structural differences of fecal microflora mainly occur in the pathways of Cardiovascular disease, Excretory system and Signaling molecules and interaction. In brief, melatonin promotes the growth and development of lambs. The potential mechanisms may be that melatonin increased the growth hormone and testosterone mediated apoptosis signaling pathway and regulated intestinal microbial flora. Our results provide valuable information for melatonin to improve the production of sheep husbandry in the future.