medRxiv : the preprint server for health sciences最新文献

Objectives: Informal caregivers play an indispensable role in and are often the sole source of care for older adults in low and middle-income settings worldwide. Intensive informal care predicts mortality and morbidity among caregivers in higher-income settings. However, there is limited evidence from poorer settings, including Africa countries, where caregiving is shared widely, including across generations. We therefore investigated caregivers' health status in rural South Africa.

Methods: We conducted quantitative interviews with all household members and all non-household caregivers aged ≥12 (n=1012) of 106 older adults in rural Mpumalanga, South Africa. We used multivariable regression with care-recipient random intercepts to assess the relationships between four caregiving characteristics and both self-reported chronic conditions and self-reported health status, considering how caregiver age moderated each association.

Results: Over half of all caregivers reported at least one chronic health condition, despite half being aged under 40. Caregivers self-reporting the worst health status provided high hours of care. However, caregivers' health status was not significantly associated with weekly care quantity or history of caring. Those aged ≥40 who reported being a main caregiver had 52% increased odds of reporting poorer health compared to other same-aged carers (95% confidence interval: 0.99, 2.35), while having more chronic conditions was associated with being expected to act as a sole caregiver more often among caregivers aged ≤39.

Discussion: Greater caring responsibilities for older adults were not consistently associated with caregivers' health in a setting where poor health is common, and caregiving is spread widely. Longitudinal data is necessary to unpack possible explanations of these findings, and to determine whether intensive caregiving speeds downward health trajectories for carers.

The rapid advancements in magnetic resonance imaging (MRI) technology have precipitated a new paradigm wherein cross-modality data translation across diverse imaging platforms, field strengths, and different sites is increasingly challenging. This issue is particularly accentuated when transitioning from 3 Tesla (3T) to 7 Tesla (7T) MRI systems. This study proposes a novel solution to these challenges using generative adversarial networks (GANs)-specifically, the CycleGAN architecture-to create synthetic 7T images from 3T data. Employing a dataset of 1112 and 490 unpaired 3T and 7T MR images, respectively, we trained a 2-dimensional (2D) CycleGAN model, evaluating its performance on a paired dataset of 22 participants scanned at 3T and 7T. Independent testing on 22 distinct participants affirmed the model's proficiency in accurately predicting various tissue types, encompassing cerebral spinal fluid, gray matter, and white matter. Our approach provides a reliable and efficient methodology for synthesizing 7T images, achieving a median Dice of 6.82%,7,63%, and 4.85% for Cerebral Spinal Fluid (CSF), Gray Matter (GM), and White Matter (WM), respectively, in the testing dataset, thereby significantly aiding in harmonizing heterogeneous datasets. Furthermore, it delineates the potential of GANs in amplifying the contrast-to-noise ratio (CNR) from 3T, potentially enhancing the diagnostic capability of the images. While acknowledging the risk of model overfitting, our research underscores a promising progression towards harnessing the benefits of 7T MR systems in research investigations while preserving compatibility with existent 3T MR data. This work was previously presented at the ISMRM 2021 conference (Diniz, Helmet, Santini, Aizenstein, & Ibrahim, 2021).

Adenoid cystic carcinoma (ACC) is a rare and lethal malignancy that originates in secretory glands of the head and neck. A prominent molecular feature of ACC is the overexpression of the proto-oncogene MYB. ACC has a poor long-term survival due to its high propensity for recurrence and protracted metastasis. Currently, clinical technologies lack the efficiency to distinguish patient prognosis prior to its redevelopment. We hypothesize that metastatic ACC can be detected by monitoring tumor-specific MYB expression in patients' blood. We developed a quantitative polymerase chain reaction (qPCR) assay for MYB transcripts and screened blood samples from four patient cohorts: no history or evidence of ACC (n=23), past history of ACC and no evidence of disease (NED) for greater than three years (n=15), local ACC (n=6), and metastatic ACC (n=5). Our assay detected significantly elevated levels of MYB transcripts in the metastatic ACC cohort (p < 0.01). Receiver operating characteristic (ROC) curves comparing metastatic to NED and metastatic to local disease were significant, with p values < 0.0001 and 0.0008, respectively. Single-cell RNA sequencing (scRNA-seq) of blood from metastatic ACC identified a cluster of circulating tumor cells (CTCs) expressing MYB. Here, we report a sensitive, cost-effective, and minimally invasive diagnostic test that leverages tumor-specific signatures to screen for metastatic ACC disease, potentially enhancing detection earlier than the current clinical standard.

In this paper, we leveraged Large Language Models(LLMs) to accelerate data wrangling and automate labor-intensive aspects of data discovery and harmonization. This work promotes interoperability standards and enhances data discovery, facilitating AI-readiness in biomedical science with the generation of Common Data Elements (CDEs) as key to harmonizing multiple datasets. Thirty-one studies, various ontologies, and medical coding systems served as source material to create CDEs from which available metadata and context was sent as an API request to 4th-generation OpenAI GPT models to populate each metadata field. A human-in-the-loop (HITL) approach was used to assess quality and accuracy of the generated CDEs. To regulate CDE generation, we employed ElasticSearch and HITL to avoid duplicate CDEs and instead, added them as potential aliases for existing CDEs. The generated CDEs are foundational to assess the interoperability potential of datasets by determining how many data set column headers can be correctly mapped to CDEs as well as quantifying compliance with permissible values and data types. Subject matter experts reviewed generated CDEs and determined that 94.0% of generated metadata fields did not require manual revisions. Data tables from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and the Global Parkinson's Genetic Program (GP2) were used as test cases for interoperability assessments. Column headers from all test cases were successfully mapped to generated CDEs at a rate of 32.4% via elastic search.The interoperability score, a metric for dataset compatibility to CDEs and other connected datasets, based on relevant criteria such as data field completeness and compliance with common harmonization standards averaged 53.8 out of 100 for test cases. With this project, we aim to automate the most tedious aspects of data harmonization, enhancing efficiency and scalability in biomedical research while decreasing activation energy for federated research.

Background: Postoperative delirium may be mediated by perioperative systemic- and neuro-inflammation. By inhibiting the pro-inflammatory actions of plasmin, tranexamic acid (TXA) may decrease postoperative delirium. To explore this hypothesis, we modified an ongoing randomised trial of TXA, adding measures of postoperative delirium, cognitive function, systemic cytokines, and astrocyte activation.

Methods: Adults undergoing elective posterior lumbar fusion randomly received intraoperative intravenous TXA (n=43: 10 mg kg^-1 loading dose, 2 mg kg^-1 h^-1 infusion) or Placebo (n=40). Blood was collected pre- and at 24 h post-operatively (n=32) for biomarkers of systemic inflammation (cytokines) and astrocyte activation (S100B). Participants had twice daily delirium assessments using the 3-minute diagnostic interview for Confusion Assessment Method (n=65). Participants underwent 4 measures of cognitive function preoperatively and during post-discharge follow-up.

Results: Delirium incidence in the TXA group (7/32=22%) was not significantly less than in the Placebo group (11/33=33%); P=0.408, absolute difference=11%, relative difference=33%, effect size = -0.258 (95% CI -0.744 to 0.229). In the Placebo group (n=16), delirium severity was associated with the number of instrumented vertebral levels (P=0.001) and with postoperative interleukin -8 and -10 concentrations (P=0.00008 and P=0.005, respectively) and these associations were not significantly modified by TXA. In the Placebo group, delirium severity was associated with S100B concentration (P=0.0009) and the strength of the association was decreased by TXA (P=0.002).

Conclusions: A potential 33% relative decrease in postoperative delirium incidence justifies an adequately powered clinical trial to determine if intraoperative TXA decreases delirium in adults undergoing lumbar fusion.

Cross-linguistic studies with healthy individuals are vital, as they can reveal typologically common and different patterns while providing tailored benchmarks for patient studies. Nevertheless, cross-linguistic differences in narrative speech production, particularly among speakers of languages belonging to distinct language families, have been inadequately investigated. Using a picture description task, we analyze cross-linguistic variations in connected speech production across three linguistically diverse groups of cognitively normal participants-English, Chinese (Mandarin and Cantonese), and Italian speakers. We extracted 28 linguistic features, encompassing phonological, lexico-semantic, morpho-syntactic, and discourse/pragmatic domains. We utilized a semi-automated approach with Computerized Language ANalysis (CLAN) to compare the frequency of production of various linguistic features across the three language groups. Our findings revealed distinct proportional differences in linguistic feature usage among English, Chinese, and Italian speakers. Specifically, we found a reduced production of prepositions, conjunctions, and pronouns, and increased adverb use in the Chinese-speakers compared to the other two languages. Furthermore, English participants produced a higher proportion of prepositions, while Italian speakers produced significantly more conjunctions and empty pauses than the other groups. These findings demonstrate that the frequency of specific linguistic phenomena varies across languages, even when using the same harmonized task. This underscores the critical need to develop linguistically tailored language assessment tools and to identify speech markers that are appropriate for aphasia patients across different languages.

The study conducted a comprehensive genome-wide analysis of differential 5mC and 5hmC modifications at both CpG and non-CpG sites in postmortem orbitofrontal neurons from 25 PTSD cases and 13 healthy controls. It was observed that PTSD patients exhibit a greater number of differential 5hmC sites compared to 5mC sites. Specifically, individuals with PTSD tend to show hyper-5mC/5hmC at CpG sites, particularly within CpG islands and promoter regions, and hypo-5mC/5hmC at non-CpG sites, especially within intragenic regions. Functional enrichment analysis indicated distinct yet interconnected roles for 5mC and 5hmC in PTSD. The 5mC marks primarily regulate cell-cell adhesion processes, whereas 5hmC marks are involved in embryonic morphogenesis and cell fate commitment. By integrating published PTSD findings from central and peripheral tissues through multi-omics approaches, several biological mechanisms were prioritized, including developmental processes, HPA axis regulation, and immune responses. Based on the consistent enrichment in developmental processes, we hypothesize that if epigenetic changes occur during early developmental stages, they may increase the risk of developing PTSD following trauma exposure. Conversely, if these epigenetic changes occur in adulthood, they may influence neuronal apoptosis and survival mechanisms.

Background: Digital remote patient monitoring (RPM) enables longitudinal care outside traditional healthcare settings, especially in the vulnerable period after hospitalizations, with broad coverage of the service by payers. We sought to evaluate patterns of RPM service availability at US hospitals and the association of these services with 30-day readmissions for two key cardiovascular conditions, heart failure (HF) and acute myocardial infarction (AMI).

Methods: We used contemporary national data from the American Hospital Association (AHA) Annual Survey to ascertain US hospitals offering RPM services for post-discharge or chronic care and used census-based county-level data to define the characteristics of the communities they serve. We linked these with hospitals' benchmarked risk-standardized relative performance on readmissions (excess readmission ratio [ERR]) from CMS Hospital Quality Reports (2018-2022). We used mixed-effects multivariable regression to examine the association between RPM services at hospitals and hospital characteristics-adjusted ERR for HF and AMI.

Results: There were 2,754 hospitals with CMS quality report data. Over five years of the study, there was a 38.3% increase in the number of hospitals offering RPM services, rising from 952 (42.0%) hospitals in 2018 to 1,237 (58.1%) in 2022. However, the availability of RPM services varied across different hospital groups with smaller, non-teaching hospitals, particularly those serving rural, low-income communities, and those located in the South, were less likely to offer RPM services. There was a consistent association between the availability of RPM services and better risk-standardized readmission performance for HF, with lower ERR for hospitals offering RPM compared with those not offering RPM (absolute difference, -0.016 [-0.023, -0.009], standardized difference, 24.6%, p<0.001). However, no such association was observed for AMI (-0.007 [-0.016, 0.002], standardized difference, 10.3%, p=0.19).

Conclusions: In this national study of US hospitals, there has been a large increase in the availability of RPM services but with large variation among hospitals, with lower availability in hospitals serving low-income and rural communities. RPM services were associated with lower hospital readmission rates for HF but not AMI.

Objectives: To develop and compare prediction models for 30-day and 1-year mortality in Heart failure with preserved ejection fraction (HFpEF) using EHR data, utilizing both traditional and machine learning (ML) techniques.

Background: HFpEF represents 1 in 2 heart failure patients. Predictive models in HFpEF, specifically those derived from electronic health record (EHR) data, are less established.

Methods: Using MIMIC-IV EHR data from 2008-2019, patients aged ≥ 18 years admitted with a primary diagnosis of HFpEF were identified using ICD-9 and 10 codes. Demographics, vital signs, prior diagnoses, and lab data were extracted. Data was partitioned into 80% training, 20% test sets. Prediction models from seven model classes (Support Vector Classifier (SVC), Logistic Regression, Lasso Regression, Elastic Net, Random Forest, Histogram-based Gradient Boosting Classifier (HGBC), and XGBoost) were developed using various imputation and oversampling techniques with 5-fold cross-validation. Model performance was compared using several metrics, and individual feature importance assessed using SHapley Additive exPlanations (SHAP) analysis.

Results: Among 3910 hospitalizations for HFpEF, 30-day mortality was 6.3%, and 1-year mortality was 29.2%. Logistic regression performed well for 30-day mortality (Area Under the Receiver operating characteristic curve (AUC) 0.83), whereas Random Forest (AUC 0.79) and HGBC (AUC 0.78) for 1-year mortality. Age and NT-proBNP were the strongest predictors in SHAP analyses for both outcomes.

Conclusion: Models derived from EHR data can predict mortality after HFpEF hospitalization with comparable performance to models derived from registry or trial data, highlighting the potential for clinical implementation.

Binge drinking is a relatively common pattern of alcohol use among youth with normative frequency trajectories peaking in emerging and early adulthood. Frequent binge drinking is a critical risk factor for not only the development of alcohol use disorders (AUDs) but also increased odds of alcohol-related injury and death, and thus constitutes a significant public health concern. Changes in binge drinking across development are strongly associated with changes in impulsive personality traits (IPTs) which have been hypothesized as intermediate phenotypes associated with genetic risk for heavy alcohol use and AUD. The current study sought to examine the extent to which longitudinal changes in binge drinking and intoxication frequency across adolescence and early adulthood are associated with genetic influences underlying dual-systems IPTs (i.e., top-down [lack of self-control] and bottom-up [sensation seeking and urgency] constructs) alongside genetic risk for alcohol consumption and AUD. Associations were tested using conditional latent growth curve polygenic score (PGS) models in three independent longitudinal samples (N=10,554). Results suggested consistent significant and independent associations across all samples between sensation seeking PGSs and model intercepts (i.e., higher frequency of binge drinking at first measurement occasion) and alcohol consumption PGSs and model slopes (i.e., steeper increases toward peak binge drinking frequency). Urgency PGSs were not significantly associated with changes in binge drinking or intoxication frequency. Collectively, these findings highlight the role of unique but correlated IPT and alcohol-specific genetic factors in the emergence and escalation of binge drinking during adolescence and early adulthood.