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

Yang Ni, Fanli Jia

Artificial intelligence (AI)-enabled digital interventions, including Generative AI (GenAI) and Human-Centered AI (HCAI), are increasingly used to expand access to digital psychiatry and mental health care. This PRISMA-ScR scoping review maps the landscape of AI-driven mental health (mHealth) technologies across five critical phases: pre-treatment (screening/triage), treatment (therapeutic support), post-treatment (remote patient monitoring), clinical education, and population-level prevention. We synthesized 36 empirical studies implemented through early 2024, focusing on Large Language Models (LLMs), machine learning (ML) models, and autonomous conversational agents. Key use cases involve referral triage, empathic communication enhancement, and AI-assisted psychotherapy delivered via chatbots and voice agents. While benefits include reduced wait times and increased patient engagement, we address recurring challenges like algorithmic bias, data privacy, and human-AI collaboration barriers. By introducing a novel four-pillar framework, this review provides a comprehensive roadmap for AI-augmented mental health care, offering actionable insights for researchers, clinicians, and policymakers to develop safe, effective, and equitable digital health interventions.

Maysam Riyadh Mohammed Hussein Alaasam, Raffat Hilal Abboodi, Hayder Nadhim Mohsin Al-Khayyat

Background. Subclinical hypothyroidism is a common endocrine problem in Iraq. It is diagnosed through high serum levels of thyroid-stimulating hormone (TSH) and normal serum free thyroid hormones. Glycated hemoglobin (HbA1c) is used as a marker for long-term glycemic control, but its reliability in the context of subclinical hypothyroidism remains unclear. This study purposed to evaluate glycated hemoglobin levels and insulin resistance in nondiabetic persons with subclinical hypothyroidism. Materials and methods. The study involved 200 participants, a hundred nondiabetic persons with subclinical hypothyroidism, and a hundred healthy matched controls. History was obtained from the participants, complete general and systemic examinations were performed, and weight, height, and blood pressure were recorded. Blood tests for hormonal assessment include free triiodothyronine (fT3), free thyroxine (fT4), TSH, and serum insulin. Glycated hemoglobin was measured, and HOMA-IR was used to evaluate insulin resistance. Results. No significant differences were found in the levels of fT3 and fT4 in both groups. The mean fT3 was 1.59 ± 0.20 nmol/L in the case group and 1.72 ± 1.20 nmol/L in the control group (p = 0.28). The mean fT4 was 76.8 ± 2.7 nmol/L in the case group and 77.21 ± 0.10 nmol/L in the control group (p = 0.13). Significant differences were shown in the level of TSH between the two groups (p < 0.0001). There was a substantial difference in HbA1c in both groups; the mean was 4.87 ± 0.50 % in the case group and 4.12 ± 0.80 % in the control group (p < 0.0001). Likewise, HOMA-IR shows a significant difference between the two groups; the mean was 2.65 ± 1.70 in the case group and 1.31 ± 2.80 in the control (p = 0.0001). Conclusions. Elevated HbA1c and increased insulin resistance in subclinical hypothyroidism patients suggest a potential link between thyroid dysfunction and glucose metabolism dysregulation. These findings highlight the need to carefully interpret HbA1c levels in subclinical hypothyroidism when assessing glycemic status.

Chaoming Hu, Yichi Zhang, Yao Wu et al.

BackgroundThe gut microbiota plays a key role in regulating bone homeostasis. Our previous work demonstrated that the novel organic selenium compound β-trifluoroethoxy dimethyl selenide (4aa alleviates osteoporosis; however, its mechanism remains unclear.MethodThe cytotoxicity of 4aa in osteoblast (MC3T3-E1) and osteoclast precursor (RAW264.7) cells was evaluated using CCK-8 assays. Ovariectomized (OVX) and sham-operated mice were treated with various concentrations of 4aa for 8 weeks, including a subgroup pretreated with antibiotics (ABX) to deplete the gut microbiota. Femoral bone structure was assessed by micro-computed tomography (micro-CT), osteoclast numbers were quantified, gut microbial composition was analyzed via 16S rRNA sequencing, and fecal metabolites were profiled using LC-MS/MS.Results4aa concentrations below 20 μM were non-cytotoxic to MC3T3-E1 and RAW264.7 cells. In vivo, 4aa significantly improved femoral bone mass and trabecular microarchitecture in OVX mice. Gut microbiota analysis revealed increased relative abundances of Dubosiella, Akkermansia, and Bacillus spp following 4aa administration. Metabolomic profiling identified marked alterations in citronellal, tyrosol, kaempferol, leukotriene D4, clomipramine, and phenol sulfate level. Moreover, 4aa elevated butyric acid levels and reduced the accumulation of α-ketoisovaleric acid (α-KIV), contributing to the inhibition of osteoclast differentiation.Conclusion4aa prevents estrogen deficiency-induced bone loss by modulating gut microbial composition and function. These findings support the therapeutic of 4aa as a microbiota-targeted therapeutic strategy for osteoporosis management.

Jiayu Chen, Haochen Peng, Yawen Shao et al.

Jee Hee Yoo, Ji Yoon Kim, Jae Hyeon Kim

We evaluated the effectiveness of the predictive low-glucose suspend (PLGS) algorithm in the DIA:CONN G8. Forty people with type 1 diabetes mellitus (T1DM) who used a DIA:CONN G8 for at least 2 months with prior experience using pumps without and with PLGS were retrospectively analyzed. The objective was to assess the changes in time spent in hypoglycemia (percent of time below range [%TBR]) before and after using PLGS. The mean age, sensor glucose levels, glucose threshold for suspension, and suspension time were 31.1±22.8 years, 159.7±23.2 mg/dL, 81.1±9.1 mg/dL, and 111.9±79.8 min/day, respectively. Overnight %TBR <70 mg/dL was significantly reduced after using the algorithm (differences=0.3%, from 1.4%±1.5% to 1.1%±1.2%, P=0.045). The glycemia risk index (GRI) improved significantly by 4.2 (from 38.8±20.9 to 34.6±19.0, P=0.002). Using the PLGS did not result in a change in the hyperglycemia metric (all P>0.05). Our findings support the PLGS in DIA:CONN G8 as an effective algorithm to improve night-time hypoglycemia and GRI in people with T1DM.

Ha Young Kim, Jun Li, Ana Beatriz Solana et al.

Rare diseases represent the long tail of medical imaging, where AI models often fail due to the scarcity of representative training data. In clinical workflows, radiologists frequently consult case reports and literature when confronted with unfamiliar findings. Following this line of reasoning, we introduce RADAR, Retrieval Augmented Diagnostic Reasoning Agents, an agentic system for rare disease detection in brain MRI. Our approach uses AI agents with access to external medical knowledge by embedding both case reports and literature using sentence transformers and indexing them with FAISS to enable efficient similarity search. The agent retrieves clinically relevant evidence to guide diagnostic decision making on unseen diseases, without the need of additional training. Designed as a model-agnostic reasoning module, RADAR can be seamlessly integrated with diverse large language models, consistently improving their rare pathology recognition and interpretability. On the NOVA dataset comprising 280 distinct rare diseases, RADAR achieves up to a 10.2% performance gain, with the strongest improvements observed for open source models such as DeepSeek. Beyond accuracy, the retrieved examples provide interpretable, literature grounded explanations, highlighting retrieval-augmented reasoning as a powerful paradigm for low-prevalence conditions in medical imaging.

Weike Zhao, Chaoyi Wu, Yanjie Fan et al.

Rare diseases affect over 300 million individuals worldwide, yet timely and accurate diagnosis remains an urgent challenge. Patients often endure a prolonged diagnostic odyssey exceeding five years, marked by repeated referrals, misdiagnoses, and unnecessary interventions, leading to delayed treatment and substantial emotional and economic burdens. Here we present DeepRare, a multi-agent system for rare disease differential diagnosis decision support powered by large language models, integrating over 40 specialized tools and up-to-date knowledge sources. DeepRare processes heterogeneous clinical inputs, including free-text descriptions, structured Human Phenotype Ontology terms, and genetic testing results, to generate ranked diagnostic hypotheses with transparent reasoning linked to verifiable medical evidence. Evaluated across nine datasets from literature, case reports and clinical centres across Asia, North America and Europe spanning 14 medical specialties, DeepRare demonstrates exceptional performance on 3,134 diseases. In human-phenotype-ontology-based tasks, it achieves an average Recall@1 of 57.18%, outperforming the next-best method by 23.79%; in multi-modal tests, it reaches 69.1% compared with Exomiser's 55.9% on 168 cases. Expert review achieved 95.4% agreement on its reasoning chains, confirming their validity and traceability. Our work not only advances rare disease diagnosis but also demonstrates how the latest powerful large-language-model-driven agentic systems can reshape current clinical workflows.

Kailas Dayanandan, Brejesh Lall

Clinical systems operate in safety-critical environments and are not intended to function autonomously; however, they are currently designed to replicate clinicians' diagnoses rather than assist them in the diagnostic process. To enable better supervision of system-generated diagnoses, we replicate radiologists' systematic approach used to analyze chest X-rays. This approach facilitates comprehensive analysis across all regions of clinical images and can reduce errors caused by inattentional blindness and under reading. Our work addresses a critical research gap by identifying difficult-to-diagnose diseases for clinicians using insights from human vision, enabling these systems to serve as an effective "second pair of eyes". These improvements make the clinical imaging systems more complementary and combine the strengths of human and machine vision. Additionally, we leverage effective receptive fields in deep learning models to present machine-generated diagnoses with sufficient context, making it easier for clinicians to evaluate them.

Mutong Chen, Mutong Chen, Baizhi Li et al.

BackgroundA rapid increase in the prevalence of diabetes is an urgent public health concern among older adults, especially in developing countries such as China. Despite several studies on lifestyle factors causing diabetes, sleep, a key contributor, is understudied. Our study investigates the association between night sleep duration and diabetes onset over a 7-year follow-up to fill information gaps.MethodA population-based cohort study with 5437 respondents used 2011–2018 China Health and Retirement Longitudinal Study data. Using self-reported night sleep duration from the 2011 baseline survey, information on new-onset diabetes was collected in follow-up surveys. Baseline characteristics of participants with vs. without new-onset diabetes were compared using Chi-square and Mann-Whitney U tests. Multivariable Cox regression models estimated the independent relationship between night sleep and new-onset diabetes. The addictive Cox regression model approach and piece-wise regression described the nonlinear relationship between night sleep and new-onset diabetes. Subgroup analysis was also performed by age, gender, body measurement index, dyslipidemia, drinking status, smoking, hypertension, and afternoon napping duration.Result549 respondents acquired diabetes during a median follow-up of 84 months. After controlling for confounders, night sleep duration was substantially linked with new-onset diabetes in the multivariable Cox regression model. The risk of diabetes is lower for respondents who sleep longer than 5 hours, except for those who sleep over 8 hours [5.1–6h Hazard ratios (HR) [95% confidence intervals (CI)] = 0.71 (0.55, 0.91); 6.1–7h HR = 0.69 (0.53, 0.89); 7.1–8h HR = 0.58 (0.45, 0.76)]. Nonlinear connections were delineated by significant inflection points at 3.5 and 7.5 hours, with a negative correlation observed only between these thresholds. With one hour more night sleep, the risk of diabetes drops 15%. BMI and dyslipidemia were identified as modifiers when only consider the stand linear effect of sleep duration on diabetes.ConclusionThis study establishes a robust association between night sleep and new-onset diabetes in middle-aged and older Chinese individuals within the 3.5–7.5-hour range, offering a foundation for early glycemic management interventions in this demographic. The findings also underscore the pivotal role of moderate night sleep in preventing diabetes, marking a crucial juncture in community medical research.

Jonas Dippel, Niklas Prenißl, Julius Hense et al.

While previous studies have demonstrated the potential of AI to diagnose diseases in imaging data, clinical implementation is still lagging behind. This is partly because AI models require training with large numbers of examples only available for common diseases. In clinical reality, however, only few diseases are common, whereas the majority of diseases are less frequent (long-tail distribution). Current AI models overlook or misclassify these diseases. We propose a deep anomaly detection approach that only requires training data from common diseases to detect also all less frequent diseases. We collected two large real-world datasets of gastrointestinal biopsies, which are prototypical of the problem. Herein, the ten most common findings account for approximately 90% of cases, whereas the remaining 10% contained 56 disease entities, including many cancers. 17 million histological images from 5,423 cases were used for training and evaluation. Without any specific training for the diseases, our best-performing model reliably detected a broad spectrum of infrequent ("anomalous") pathologies with 95.0% (stomach) and 91.0% (colon) AUROC and generalized across scanners and hospitals. By design, the proposed anomaly detection can be expected to detect any pathological alteration in the diagnostic tail of gastrointestinal biopsies, including rare primary or metastatic cancers. This study establishes the first effective clinical application of AI-based anomaly detection in histopathology that can flag anomalous cases, facilitate case prioritization, reduce missed diagnoses and enhance the general safety of AI models, thereby driving AI adoption and automation in routine diagnostics and beyond.

Mingyu Derek Ma, Chenchen Ye, Yu Yan et al.

The integration of Artificial Intelligence (AI), especially Large Language Models (LLMs), into the clinical diagnosis process offers significant potential to improve the efficiency and accessibility of medical care. While LLMs have shown some promise in the medical domain, their application in clinical diagnosis remains underexplored, especially in real-world clinical practice, where highly sophisticated, patient-specific decisions need to be made. Current evaluations of LLMs in this field are often narrow in scope, focusing on specific diseases or specialties and employing simplified diagnostic tasks. To bridge this gap, we introduce CliBench, a novel benchmark developed from the MIMIC IV dataset, offering a comprehensive and realistic assessment of LLMs' capabilities in clinical diagnosis. This benchmark not only covers diagnoses from a diverse range of medical cases across various specialties but also incorporates tasks of clinical significance: treatment procedure identification, lab test ordering and medication prescriptions. Supported by structured output ontologies, CliBench enables a precise and multi-granular evaluation, offering an in-depth understanding of LLM's capability on diverse clinical tasks of desired granularity. We conduct a zero-shot evaluation of leading LLMs to assess their proficiency in clinical decision-making. Our preliminary results shed light on the potential and limitations of current LLMs in clinical settings, providing valuable insights for future advancements in LLM-powered healthcare.

Akeem Temitope Otapo, Alice Othmani, Ghazaleh Khodabandelou et al.

The integration of Artificial Intelligence (AI) and the Internet of Medical Things (IoMT) in healthcare, through Machine Learning (ML) and Deep Learning (DL) techniques, has advanced the prediction and diagnosis of chronic diseases. AI-driven models such as XGBoost, Random Forest, CNNs, and LSTM RNNs have achieved over 98\% accuracy in predicting heart disease, chronic kidney disease (CKD), Alzheimer's disease, and lung cancer, using datasets from platforms like Kaggle, UCI, private institutions, and real-time IoMT sources. However, challenges persist due to variations in data quality, patient demographics, and formats from different hospitals and research sources. The incorporation of IoMT data, which is vast and heterogeneous, adds complexities in ensuring interoperability and security to protect patient privacy. AI models often struggle with overfitting, performing well in controlled environments but less effectively in real-world clinical settings. Moreover, multi-morbidity scenarios especially for rare diseases like dementia, stroke, and cancers remain insufficiently addressed. Future research should focus on data standardization and advanced preprocessing techniques to improve data quality and interoperability. Transfer learning and ensemble methods are crucial for improving model generalizability across clinical settings. Additionally, the exploration of disease interactions and the development of predictive models for chronic illness intersections is needed. Creating standardized frameworks and open-source tools for integrating federated learning, blockchain, and differential privacy into IoMT systems will also ensure robust data privacy and security.

Yomna E. Dean, Sameh Samir Elawady, Wangpan Shi et al.

Abstract Background and Aim Low serum Vitamin D levels have been associated with diabetic nephropathy (DN). Our study aimed to analyse the serum levels of vitamin D in patients suffering from DN and the subsequent changes in serum vitamin D levels as the disease progresses. Methods PubMed, Embase, SCOPUS and Web of Science were searched using keywords such as ‘25 hydroxyvitamin D’ and ‘diabetic nephropathy’. We included observational studies that reported the association between the serum 25 hydroxy vitamin D levels and diabetic nephropathy without restriction to age, gender, and location. R Version 4.1.2 was used to perform the meta‐analysis. The continuous outcomes were represented as mean difference (MD) and standard deviation (SD) and dichotomous outcomes as risk ratios (RR) with their 95% confidence interval (CI). Results Twenty‐three studies were included in our analysis with 7722 patients. Our analysis revealed that vitamin D was significantly lower in diabetic patients with nephropathy than those without nephropathy (MD: −4.32, 95% CI: 7.91–0.74, p‐value = .0228). On comparing diabetic patients suffering from normoalbuminuria, microalbuminuria, or macroalbuminuria, we found a significant difference in serum vitamin D levels across different groups. Normoalbuminuria versus microalbuminuria showed a MD of −1.69 (95% CI: −2.28 to −1.10, p‐value = .0002), while microalbuminuria versus macroalbuminuria showed a MD of (3.75, 95% CI: 1.43–6.06, p‐value = .0058), proving that serum vitamin D levels keep declining as the disease progresses. Notwithstanding, we detected an insignificant association between Grade 4 and Grade 5 DN (MD: 2.29, 95% CI: −2.69–7.28, p‐value = .1862). Conclusion Serum Vitamin D levels are lower among DN patients and keep declining as the disease progresses, suggesting its potential benefit as a prognostic marker. However, on reaching the macroalbuminuria stage (Grades 4 and 5), vitamin D is no longer a discriminating factor.

Gebrehiwot Lema Legese, Getahun Asres, Shitaye Alemu et al.

BackgroundPoor glycemic control is one of the most determinant factors for type 2 diabetes-related morbidity and mortality. The proportion of type 2 diabetes mellitus with poor glycemic control remains high. Yet evidences on factors contributing to poor glycemic control remain scarce. The aim of this study is to identify determinants of poor glycemic control among type 2 diabetes mellitus patients at a diabetes mellitus clinic in University of Gondar Comprehensive Specialized Hospital, Northwest Ethiopia.MethodsA hospital-based case-control study was conducted from June to September 2020. Using convenience sampling techniques, a total of 90 cases and 90 controls with type 2 diabetes were recruited. The data were entered into Epidata version 4.6.0.2 and analyzed by Stata version 14. A multivariable logistic regression analysis was performed to assess the association between independent variables and glycemic control. Both 95% CI and p-value&lt;0.05 were used to determine the level and significance, respectively.ResultsThe mean age ( ± standard deviations) for the cases and controls were 57.55± 10.42 and 61.03± 8.93% respectively. The determinants of poor glycemic control were age (Adjusted odd ratio (AOR)= 0.08; 95% CI= 0.02-0.33), inadequate physical exercise (AOR = 5.05; 95% CI = 1.99-11.98), presence of comorbidities (AOR = 5.50; 95% CI = 2.06-14.66), non-adherence to anti-diabetes medications (AOR= 2.76; 95% CI= 1.19-6.40), persistent proteinuria (AOR=4.95; 95% CI=1.83-13.36) and high-density lipoprotein less than 40 mg/dl (AOR=3.08; 95% CI= 1.30-7.31).ConclusionsAge less than 65 years, inadequate physical exercise, presence of comorbidities, non-adherence to anti-diabetes medications, persistent proteinuria, and high-density lipoprotein less than 40 mg/dl were the determinants of poor glycemic control. Therefore, targeted educational and behavioral modification programs on adequate exercise and medication adherence should be routinely practiced. Furthermore, early guideline-based screening and treatment of comorbidities and complications is required to effectively manage diabetes mellitus.

O.Z. Lishchuk, Н.I. Suslyk, A.M. Urbanovych

Medullary thyroid cancer (MTC) accounts for 5–10 % of all thyroid cancers. Most cases (75 %) are sporadic, but the proportion of patients with MTC and a familial predisposition syndrome is the highest among those with any hereditary cancer syndrome (about 25 %), and this possibility should be considered when examining a patient with MTC. Familial syndromes include multiple endocrine neoplasia (MEN) 2A, MEN 2B, and non-MEN familial MTC (familial MTC). Familial MTC syndromes occur in approximately one case per 30,000 of the population. Reduced penetrance and clinical variability are well-established features of many monogenic disorders, particularly phenotypes associated with the RET proto-oncogene. They require an individual assessment of the pathogenic effects and clinical significance of any identified new sequence of the RET va­riant as prerequisites for individual genetic counseling and planning of medical monitoring and treatment. Diagnostic criteria for the disease have been outlined, molecular and genetic aspects have been discussed, and the determination of treatment and further observation has been addressed. The article describes a clinical case of a rare variant of MTC. Treatment of this pathology with radical extrafascial thyroidectomy is under consideration. Given the ineffectiveness of radioiodine and chemotherapy, the main task in the treatment of MTC is early diagnosis, radical surgical intervention, and active monitoring aimed at early detection of disease recurrence. When planning prophylactic thyroidectomy, it is recommended to focus on the stratification of the level of RET gene mutations and the timing of prophylactic thyroidectomy proposed by the American Thyroid Association. The introduction of molecular genetic research into clinical practice for the purpose of diagnosing MTC allows for the objective assessment of the genetic lineage of the disease within a biological family. A timely diagnosis of MTC makes it possible to prescribe an adequate treatment at the stage of preclinical manifestations of the disease, which can significantly increase the quality and duration of life.

Hong Cheng, Dan Chen, Hui Gao

BackgroundSome studies have investigated the association between vitamin D levels and precocious puberty (PP) but with limited sample sizes and inconsistent conclusions across studies.MethodsUntil July 2022, a comprehensive electronic search of works of literature was conducted in MEDLINE, Web of Science, and CNKI (Chinese National Knowledge Infrastructure). A systematic review and meta-analysis of 15 case-control studies with 2145 cases and 2063 controls was conducted to explore the relationship between vitamin D and PP. Stratified analyses by year of publication, country, diagnosis category of PP, child’s sex, and methods of 25(OH)D test were conducted.ResultsThere was a negative correlation between 25(OH)D concentrations and PP in all study populations (SMD = -1.046, 95%CI = -1.366, -0.726). The pooled SMD remained significant in Chinese studies (SMD = -1.113, 95%CI = -0.486, -0.741), studies published before or after 2018 (SMD = -0.9832 and -1.185, 95%CI = -2.044, -1.133 and -1.755, -0.726), studies with female children (SMD = -1.114, 95%CI = -1.446, -0.781), and studies using electrochemiluminescence to detect 25(OH)D (SMD = -0.999, 95%CI = -1.467, -0.531). Vitamin D deficiency also increased the risk of PP (OR = 1.531, 95%CI = 1.098, 2.134). Unfortunately, heterogeneity was high in all analyses, and there was some publication bias.ConclusionThis systematic review and meta-analysis demonstrated an association between vitamin D and precocious puberty. We recommend more high-quality studies, especially prospective cohort studies with big sample sizes or some randomized controlled intervention trials, to validate the reliability of the results.

Haizhao Liu, Min Cao, Yutong Jin et al.

IntroductionKidney stone disease (KS) is a complicated disease with an increasing global incidence. It was shown that Bushen Huashi decoction (BSHS) is a classic Chinese medicine formula that has therapeutic benefits for patients with KS. However, its pharmacological profile and mechanism of action are yet to be elucidated.MethodsThe present study used a network pharmacology approach to characterize the mechanism by which BSHS affects KS. Compounds were retrieved from corresponding databases, and active compounds were selected based on their oral bioavailability (≥30) and drug-likeness index (≥0.18). BSHS potential proteins were obtained from the Traditional Chinese Medicine Systems Pharmacology (TCMSP) database, whereas KS potential genes were obtained from GeneCards and OMIM, TTD, and DisGeNET. Gene ontology and pathway enrichment analysis were used to determine potential pathways associated with genes. The ingredients of BSHS extract were identified by the ultra‐high‐performance liquid chromatography coupled with quadrupole orbitrap mass spectrometry (UHPLC-Q/Orbitrap MS). The network pharmacology analyses predicted the potential underlying action mechanisms of BSHS on KS, which were further validated experimentally in the rat model of calcium oxalate kidney stones.ResultsOur study found that BSHS reduced renal crystal deposition and improved renal function in ethylene glycol(EG)+ammonium chloride(AC)-induced rats, and also reversed oxidative stress levels and inhibited renal tubular epithelial cell apoptosis in rats. BSHS upregulated protein and mRNA expression of E2, ESR1, ESR2, BCL2, NRF2, and HO-1 in EG+AC-induced rat kidney while downregulating BAX protein and mRNA expression, consistent with the network pharmacology results.DiscussionThis study provides evidence that BSHS plays a critical role in anti-KS via regulation of E2/ESR1/2, NRF2/HO-1, and BCL2/BAX signaling pathways, indicating that BSHS is a candidate herbal drug for further investigation in treating KS.

Shunjie Jia, Jian Liao, Yucheng Wang et al.

PurposeNephrolithiasis is thought to be a risk factor for osteoporosis, but data assessing if osteoporosis predisposes to the risk of nephrolithiasis are lacking. The present study aims to investigate whether patients with nephrolithiasis have a prominently higher prevalence of osteoporosis than the controls and vice versa via a cumulative analysis.MethodsFour databases were used to detect the eligible studies. We calculated the relative risk (RR) with a 95% confidence interval (CI) to assess the combined effect. The methodologies for conducting this study followed the PRISMA guidelines and were registered in the PROSPERO (ID: CRD42023395875),ResultsNine case-control or cohort studies with a total of 454,464 participants were finally included. Combined results indicated that there was a significantly higher prevalence of osteoporosis in patients with nephrolithiasis as compared to the general population without nephrolithiasis (overall RR from six studies= 1.204, 95%CI: 1.133 to 1.28, P&lt; 0.001; heterogeneity: I2 = 34.8%, P= 0.162). Conversely, osteoporosis was significantly correlated to an increased risk of nephrolithiasis as compared to the controls without osteoporosis (overall RR from four studies= 1.505, 95%CI: 1.309 to 1.731, P&lt; 0.001; I2 = 89.8%, P&lt; 0.001). Sensitivity analysis on the two categories validated the above findings. No significant publication bias was identified in this study.ConclusionsThe present study highlighted a significantly high prevalence of osteoporosis in patients with nephrolithiasis and vice versa. This reciprocal association reminded the clinicians to conduct a regular follow-up assessment when managing patients with nephrolithiasis or osteoporosis, especially for the elderly.Systematic review registrationhttps://www.crd.york.ac.uk/PROSPERO/#searchadvanced, identifier CRD42023395875.

Chihcheng Hsieh, Isabel Blanco Nobre, Sandra Costa Sousa et al.

This study investigates the effects of including patients' clinical information on the performance of deep learning (DL) classifiers for disease location in chest X-ray images. Although current classifiers achieve high performance using chest X-ray images alone, our interviews with radiologists indicate that clinical data is highly informative and essential for interpreting images and making proper diagnoses. In this work, we propose a novel architecture consisting of two fusion methods that enable the model to simultaneously process patients' clinical data (structured data) and chest X-rays (image data). Since these data modalities are in different dimensional spaces, we propose a spatial arrangement strategy, spatialization, to facilitate the multimodal learning process in a Mask R-CNN model. We performed an extensive experimental evaluation using MIMIC-Eye, a dataset comprising modalities: MIMIC-CXR (chest X-ray images), MIMIC IV-ED (patients' clinical data), and REFLACX (annotations of disease locations in chest X-rays). Results show that incorporating patients' clinical data in a DL model together with the proposed fusion methods improves the disease localization in chest X-rays by 12\% in terms of Average Precision compared to a standard Mask R-CNN using only chest X-rays. Further ablation studies also emphasize the importance of multimodal DL architectures and the incorporation of patients' clinical data in disease localization. The architecture proposed in this work is publicly available to promote the scientific reproducibility of our study (https://github.com/ChihchengHsieh/multimodal-abnormalities-detection)

Zoe Fowler, Kiran Kokilepersaud, Mohit Prabhushankar et al.

This paper presents a novel approach to active learning that takes into account the non-independent and identically distributed (non-i.i.d.) structure of a clinical trial setting. There exists two types of clinical trials: retrospective and prospective. Retrospective clinical trials analyze data after treatment has been performed; prospective clinical trials collect data as treatment is ongoing. Typically, active learning approaches assume the dataset is i.i.d. when selecting training samples; however, in the case of clinical trials, treatment results in a dependency between the data collected at the current and past visits. Thus, we propose prospective active learning to overcome the limitations present in traditional active learning methods and apply it to disease detection in optical coherence tomography (OCT) images, where we condition on the time an image was collected to enforce the i.i.d. assumption. We compare our proposed method to the traditional active learning paradigm, which we refer to as retrospective in nature. We demonstrate that prospective active learning outperforms retrospective active learning in two different types of test settings.