22689 research outputs found
Sort by
The Impact of an Integrated Care Management Program: Bettering Health Outcomes and Reducing Care Dependency among Older Adults in Selected Cities of China
Get PDFChina faces unprecedented health and social care burden as the population ages fast, people live longer with higher prevalence of chronic diseases, frailty and care dependency. Fragmentation of services further increases such burden. Local evidence for implementing person-centered integrated care is urgently needed to advance practice and policies to achieve healthy ageing as part of the United National Decade of Healthy Ageing initiatives.
Both World Health Organization (WHO) and China government have identified integrated care as key to bettering health outcomes at lower costs and reduce/delay admission to hospital or nursing home. For further policy design, it is important to understand China’s readiness for implementing integrated care and its impact on quality of life in older age.
This is a retrospective study of data from the implementation pilot of WHO’s Integrated Care for Older People (ICOPE) program in China, a randomized controlled trial among community-dwelling older adults in Chaoyang District of Beijing. The ICOPE care pathways was localized and implemented through a network of trained primary health care providers as “integrated care managers”, while health outcomes and utilization of health care resources from baseline to 6-month follow-up were compared between the intervention group receiving full ICOPE program and the control group receiving standard available care.
Propensity score matching (PSM) was used to adjust for the imbalances in some baseline variables among participants due to impact of the COVID-19 pandemic situation in recruitment phase. After 1 to 1 PSM, participants in intervention and control groups (totally 938) had comparable baseline characteristics, demonstrated feasibility of implementing ICOPE with satisfaction by participants (97%-99%) and providers (92%-93%). All outcomes showed improvements after a 6-month intervention, while statistically significant Least-Squares Mean Differences (control-intervention) in vitality (MNA-SF, -0.21, 95% CI, -0.40--0.02), mobility (SPPB, -0.29, 95% CI, -0.44--0.14), and psychological health (GDS-5, 0.09, 95% CI, 0.03-0.14) were observed (p<0.05).
In conclusion, the study showed that it is feasible to localize and implement WHO’s ICOPE approach in regions with fragmented resources. Preliminary evidence supports its acceptance among key stakeholders and impact on health outcomes, which is helpful when planning for further piloting of the program
CHARACTERIZATION OF CAF TRANSCRIPTIONAL AND TRANSLATIONAL CHANGE IN RESPONSE TO IRRADIATION
Get PDFIonizing radiation nowadays has become a primary therapeutic modality in oncology, many cancer patients undergo this treatment at some point during their disease trajectory. This intervention acts sequentially, first by damaging the cell membrane and subsequently permeating into the cell where it exerts both direct and indirect effects. These actions collectively compromise DNA integrity, destabilizing the structure of the DNA double helix and culminating in cell death. While radiation is undeniably potent in its tumoricidal capabilities, its effects are not without consequence. One of the most notable repercussions of radiation therapy is its potential to foster the development of Cancer-Associated Fibroblasts (CAFs). Diverse stimuli, spanning from radiation to growth factors and cytokines, have been identified as potent inducers, compelling resting fibroblasts to assume CAF-specific phenotypic and functional attributes.
CAFs, once activated, are instrumental in playing a role in tumorigenesis. They churn out an array of growth factors, cytokines, and chemokines at elevated concentrations, which in tandem shape tumor immunity and the overarching inflammatory milieu.
Previous in-house data from Boehringer Ingelheim underscored the capacity of irradiation to activate CAFs. A direct correlation was observed between escalating irradiation doses and the expression levels of the marker GeneX, substantiating the direct role of irradiation in CAF activation. Our studies further delved into characterizing other upregulated genes and proteins post-radiation. RNA sequencing data elucidated that radiation-induced DNA damage leads to CAF cell cycle arrest within 8 to 24 hours. Subsequently, from 48 hours to 7 days, CAFs exhibit activation and an immunosuppressive gene profile. Notably, the cytokine GDF-15, secreted by CAFs, displayed upregulation post-radiation in both dose and time-dependent manners. GDF-15 is known to inhibit dendritic cell maturation; however, the effect of irradiated CAF on dendritic cell maturation warrants comprehensive exploration
IMPACT OF BEHAVIORAL HEALTH SERVICES IN AN URBAN INTENSIVE PRIMARY CARE UNIT
Get PDFProblem Statement
Patients considered the highest utilizers of healthcare account for an overwhelming portion of healthcare expenditures. These high-need high-cost (HNHC) patients are often burdened with multiple chronic conditions, compounded by mental and behavioral health issues. With needs unattended to, their struggles perpetuate the cycle of inappropriate utilization, resulting in ineffective personal and system-wide efforts to gain and maintain health. In an effort to meet these patients’ medical and behavioral health needs, one intensive primacy care unit (IPCU) targets HNHC patients and provides greater access using a multi-disciplinary team approach.
Method
A retrospective cohort study with a longitudinal design utilized a limited data set from an electronic health record between 2013-2020. Quantitative analyses were completed to explore the characteristics of patients who were enrolled into an urban intensive primary care unit and how this enrollment impacted diagnoses and utilization. T tests explored changes in mental and behavioral health diagnoses pre to post IPCU enrollment and utilization pre to post enrollment. Logistic regression analyses examined associations between patient characteristics and diagnoses pre and post IPCU enrollment. Linear regression models explored associations between mental and behavioral health diagnosis and utilization pre post enrollment as well as patient characteristics association with utilization pre and post enrollment.
Results
This sample (n=225) consisted of 65% females with a mean enrollment age of 44.8 who participated in this urban IPCU between the years 2014 -2020. Race was composed of 90% Black/African American (n=203) patients, 70% past or current smokers (n=158), and 88% not partnered in a relationship (n=199), with a mean BMI of 30.52. Ninety-six percent of patients had a mental, behavioral, and neurodevelopmental disorder diagnosis. These diagnoses increased from pre to post enrollment. ED utilization and hospital admissions decreased from pre to post enrollment, and outpatient office visits increased.
Conclusion
Results from this research suggest that an IPCU can be effective for improving care and utilization for HNHC patients. Patients benefit from the integration of a behavioral intervention in managing both their chronic medical and mental issues. Targeted screening into these intimate programs enables the multi-disciplinary team to give increased access and holistic patient-centered care
Adversarial Training and Computer Vision Methods for AI Robustness
No full textRecent advances in artificial intelligence (AI) have achieved remarkable success in various fields such as computer vision, natural language processing, and robotics. Driven by a vast amount of computing resources and training data, AI is reshaping human life. Despite their remarkable performance, the robustness issue of AI could pose serious challenges to their widespread deployment, especially in security-critical applications like autonomous driving and medicine. The most alarming issue is adversarial attacks, where malicious modifications can be made to the input data to mislead AI systems to make wrong predictions. In addition, AI systems trained in controlled and experimental settings often fail in real-world applications when the data and tasks differ from those used in the training phase.
To overcome these challenges, we focus on investigating and improving the robustness of AI using adversarial training and computer vision methods. The first part of this dissertation aims to develop robust and generalizable defenses for AI systems against adversarial attacks. We focus on two defense strategies: adversarial training, and pre-processing. The second part of this dissertation explores novel adversarial attacks based on diffusion models. We show that adversarial images that are natural and high-quality can fool state-of-the-art AI systems. The last part of this dissertation attempts to address some of the practical issues for deploying AI in the real world, including dealing with incomplete input data, and recognizing objects outside pre-defined categories
LITERACY INSTRUCTION IN A HIGH-ACHIEVING HIGH SCHOOL: DISCIPLINARY PERCEPTIONS, CHALLENGES, AND OPPORTUNITIES
Get PDFIn an affluent, high achieving high school in Western Pennsylvania, state-level exam scores have been diverging over the course of five years, with English Language Arts increasing and other disciplinary areas decreasing. Bronfenbrenner’s (1997) ecological systems theory served as the framework for a synthesis of research, which revealed that high socioeconomic status environments can have negative impacts on student reading achievement, high-achieving environments impose specific stressors on educators, and low levels of knowledge and self-efficacy lead to less literacy instruction across content areas. A needs assessment using a mixed methods sequential explanatory design confirmed low teacher beliefs, attitudes, and efficacy in literacy instruction across content areas combined with high value placed on using challenging texts. The study also revealed that teachers primarily used generalized reading strategies in their classrooms, infused more critical and disciplinary literacy than they perceived, and had positive views of literacy instruction if it did not negatively impact content instruction. After a review of interventions, using Mezirow’s (1997) transformative learning theory as a framework, the researcher determined that a comprehensive literacy coaching program, combining joint action and disciplinary models, would best fulfill the literacy needs of the context. The researcher then developed a literacy coaching guidebook, written in practitioner-friendly language, that combines approaches directly informed by the synthesis of literature and needs assessment
THE COST OF CARING: SECONDARY TRAUMATIC STRESS IN TEACHERS
No full textIndividuals who assist others in the context of their occupations are known as helping professionals. Such professionals may support individuals who have experienced traumatic events and merely lending an ear to the details of these traumatic experiences can make an individual susceptible to an insidious occupational hazard referred to as secondary traumatic stress. Research findings have well documented certain helping professions, such as social workers, paramedics, and first responders, are at risk for secondary traumatic stress. However, the research also demonstrates teachers have long been excluded under the consideration as helping professionals and have been left to suffer in silence. A mixed-method design was used to understand the prevalence and impact of secondary traumatic stress, burnout, compassion fatigue, and compassion satisfaction between intervention and control groups, utilizing a mixed-methods approach. One intervention group engaged in an online curriculum using the Building Personal Resilience curriculum from HeartMath™ only and the other intervention group received the curriculum and coaching on using the Inner Balance biofeedback machine. Findings suggest that the group who received biofeedback plus the curriculum had reduced total STS scores and reduced scores in all STS subscores of avoidance, intrusion, and arousal symptoms. HRV findings suggest positive trends in pre-post HRV indices in the biofeedback group
Randomized Data Splitting Leads to Inflated Performance for Seizure Forecasting Algorithms
Get PDFRecent advancements in machine learning (ML) have facilitated personalized treatments across various domains, particularly in healthcare. In this thesis, we present the application of seizure prediction in which we analyze electrographic signals to predict when a seizure is likely to occur. To begin, we outline Toast, a Python framework specifically designed for ML researchers to efficiently develop, train, and assess neural networks using patient-specific datasets. Unique to Toast are features like experiment management, simple network construction, and optimized data handling for time series data, setting it apart from existing libraries. We delve into the challenges faced by seizure forecasting algorithms in managing epilepsy, emphasizing the necessity of timely warnings and proactive interventions. Addressing the historical struggle of algorithms when applied to unseen, real-time data, we propose alternative data-splitting strategies to counter the naivety of traditional ML algorithms, which often assume data points are independent. Specifically, we introduce two alternative strategies---a time-based split and a semi-divided split---which provide more accurate representations of the underlying data compared to randomized sampling strategies and yield performance expected in deployed applications. Furthermore, we discuss model-agnostic approaches for tracking distributional shifts of input features and internal representations using techniques such as multivariate PCA reconstruction error, t-SNE visualizations, and optimal transport-based distance measurements across time and patients. We demonstrate the effectiveness of these measures in correlating with a model's performance on held-out data, thus highlighting their utility in predicting algorithmic robustness and generalization capabilities. Driven by the innovative use of modular implantable devices aimed at understanding and treating neural pathologies, lightweight, robust, and interpretable ML models customized for seizure forecasting hold promise in bridging two pivotal clinical phases: sensing and intervention within a closed-loop system
DEPARTMENT VS CENTRALIZED RESEARCH ADMINISTRATION ASSISTANCE
Get PDFAccording to the Higher Education Research and Development Expenditures, the 2020 cycle shows that average research expenditures increased by 3.3 percent which was 86.3 billion. The PI’s (Principal Investigators) administrative responsibilities increased in tandem with rising research expenditures, along with requirements from the funding agency and associated institutions. Principal Investigators often have countless administrative responsibilities related to managing awards. Principal Investigators frequently complain that either the department provides little to no research support, or that the only resource for assistance with award management is the Central Office/Office of Sponsored Programs. This study examines the research administrative assistance investigators received from the department to that they receive only from the Central Office relating to the award management
A Comparative Analysis of Urban Forest Canopy Coverage and Their Respective Urban Forestry Management Strategies
Get PDFThe purpose of this study was to review the impact of the management strategies of five different San Francisco Bay Area Cities to determine if there was a best strategy for retaining and growing urban tree canopy. This study utilized aerial imagery and ArcGIS to detect changes to land cover between 2016 and 2022. The change detection data was then visualized and compared between the five cities. This data was compared to the information found in Community Forest Master Plans and Municipal Codes that determined management strategy and funding level. The study found that programs that invested more dollars per tree, tended to be programs where cities took responsibility for maintaining trees. These programs produced maps with less visible loss than those cities that invested fewer dollars per tree. The cities and towns that required property owners to maintain trees also had more areas of loss. Overall, the study found that this methodology could be an essential first step to determining how to evaluate a city’s tree canopy further to produce the data to change management strategies to improve retention and growth
IDENTIFYING RISK FACTORS FOR COMMUNITY-ACQUIRED AND HOSPITAL-ACQUIRED BACTERIAL INFECTIONS ASSOCIATED WITH STILLBIRTHS AND EARLY NEONATAL DEATHS IN BANGLADESH
Get PDFAs of 2021, Bangladesh is ranked sixth globally with 63199 stillbirths annually (21 stillbirths per 1,000 total births). Maternal infections during pregnancy are important preventable risk factors in LMICs. However, there are limited studies due to the necessity of specialized procedures like fetal autopsies for understanding the relationship between infections and perinatal outcomes. Therefore, we aim to identify risk factors for stillbirths and deaths within 72 hours of birth (SB and Deaths < 72 hours of birth) with culture-confirmed bacterial infection using minimally invasive tissue sampling (MITS) in Bangladesh.
Methods:
Using the data from the Child Health and Mortality Prevention Surveillance in Bangladesh between September 2017 and January 2024, we did descriptive analysis for stillbirths and deaths within 72 hours of birth associated with bacterial infections and characterized exposures in the mother that could have led to colonization or infection and onward infection in the fetus.
Results:
Out of 744 SB and Deaths < 72 hours of birth with MITS performed, 109 (15%) were positive bacterial growth [Escherichia coli (14%), Acinetobacter baumannii (12%), Klebsiella pneumoniae (6%), Acinetobacter lwoffii (16%) and Staphylococcus aureus (6%). Although hospitalization in the past month before delivery was higher in culture-positive cases (77%) compared to culture-negative cases (73%). The use of unsafe drinking water sources (tube wells) was (79%) in culture-positive cases compared to culture-negative cases (68%). The distribution of newborns having symptoms of pneumonia was 4% in culture-positive and 1% in culture-negative cases. Monsoon and winter seasons had the highest occurrence 31-33% of the culture-positive cases compared to other seasons. The analysis was limited by substantial data ‘missing not at random’ across crucial variables limiting use of reliable statistical testing and interpretation of these differences.
Conclusion:
The bacteria identified in the culture are known to be associated with both hospital and community-acquired infections. Further detailed prospective cohort studies that include collecting maternal samples such as blood, urine, and vaginal swabs, as well as environmental samples and minimally invasive tissue sampling (MITS) from neonates are necessary to better understand the role of infections in adverse perinatal outcomes