22689 research outputs found
Sort by
DEVELOPING AND VALIDATING LESION PROTOCOLS FOR DEEP CEREBELLAR NUCLEI IN RATS
Get PDFThe deep cerebellar nuclei, particularly the fastigial and interposed nuclei, are increasingly recognized for their role in processing self-motion signals and potentially modulating hippocampal spatial computations. This thesis aimed to develop and validate lesion protocols in rats to investigate the deep cerebellar nuclei’s contribution
to path integration gain recalibration in the hippocampus. Two approaches were tested: NMDA-induced excitotoxic lesions, which offer focal ablation of neuronal populations
but limited spatial coverage, and anodal electrolytic lesions, which provided broader and more robust ablation of both nuclei.
Histological analysis confirmed accurate targeting, and behavioral testing showed that NMDA-lesioned rats maintained locomotor abilities with subtle gait impairments. These impairments were consistent with fastigial dysfunction and indicate that the animals remain suitable for downstream experiments. These protocols enable future studies using a virtual-reality dome system that manipulates visual feedback to induce
hippocampal gain recalibration. By combining targeted lesions with in vivo place cell recordings, this work sets the stage for identifying how cerebellar outputs shape hippocampal spatial representations and adaptive navigation
Blind Reconstruction of Astronomical Images from Multiple Short Exposures
Get PDFGround-based astronomical imaging is significantly affected by atmospheric turbulence, which leads to high-frequency changes in the short-exposure observations and considerable blurring over longer exposure times. The thesis addresses the challenge of reconstructing a high-fidelity sky image by integrating multiple short-exposure frames captured with a budget CMOS-based system. A comprehensive calibration pipeline, including bias, dark, and flat corrections, is developed to mitigate detector noise and fixed-pattern artifacts. Building on this, blind ImageMM is introduced as an alternating Majorization–Minimization (MM) framework that jointly estimates the latent scene and a distinct point-spread function (PSF) for each exposure without requiring pre-measured PSFs. The resulting multiplicative update rules preserve non-negativity and flux conservation and guarantee monotonic descent of the surrogate objective. On synthetic data, where PSFs are modeled as capital English letters encoding high-frequency structure, blind ImageMM recovers the exact PSF shapes and the latent image achieves a peak signal-to-noise ratio of 49.46 dB versus 39.98 dB on median coaddition. Applied to real short-exposure observations of a bright star Vega and the binary system Beta Cygni, the algorithm resolves close stellar companions, sharpens stellar cores, and suppresses background noise more effectively than conventional coaddition. The results demonstrate that blind ImageMM provides an empirically validated framework for high-quality reconstruction of short-exposure astronomical images with modest instrumentation
EVALUATION OF A DIGITAL HEALTH SOLUTION FOR CHRONIC DISEASE PATIENTS IN PEKING UNION MEDICAL COLLEGE HOSPITAL
Get PDFBackground: Peking Union Medical College Hospital (PUMCH) was founded in 1921 by the Rockefeller Foundation through its China Medical Board (CMB). [1] In response to the COVID-19 outbreak, a telemedicine/digital health solution (app) was introduced at PUMCH to facilitate physician-patient collaboration via online communication during the lockdown period.[1] This application transform into the online offline integrated service model to address the traditional offline model’s pain points, including restricted accessibility, high costs, and fragmented services. [3] [4] The implementation of digital health solutions represents a critical opportunity to explore their value in improving healthcare delivery, particularly for chronic disease patients whose treatment necessitates long-term communication and collaboration with physicians throughout the care continuum.[5] However, there remains an absence of widely adopted, standardized tools to assess patient experiences and outcomes associated with telemedicine interventions.[6]
Methods: A systematic review was conducted to identify and analyze the potential benefits of telemedicine for patients with chronic diseases. The Telemedicine Service Experience Questionnaire (TSEQ) was developed to evaluate the value. This questionnaire consists of 14 items across four dimensions. The scale’s internal consistency and validity were rigorously tested. The survey was administered to patients attending in-person consultations—some of whom had prior internet-based service experiences—at PUMCH between July and August 2021. Eligible participants were identified based on predefined inclusion criteria and completed the custom-designed questionnaire.
Results: A total of 3337 valid questionnaires were collected. Results showed significant enhancements in service efficiency, information guidance, doctor-patient communication, and overall experience scores among patients utilizing internet-based diagnosis and treatment services (all P<0.05). Patients in the oncology department had better medical experiences compared to patients in other chronic disease departments (all P<0.05). Regression analysis identified age, and treatment location (local or remote) as significant determinants of patient experience. Younger patients and those treated outside their local area, reported better medical experiences (all P<0.05).
Conclusion: The Findings from this survey demonstrate that PUMCH digital health solution significantly enhance the overall patient experience. A comparison between patients from chronic disease-related departments and those from other departments revealed no significant differences. The study identifies key factors influencing telemedicine experiences, like age, gender, and cross-regional healthcare utilization. Additionally, the study advocates for the integration of more comprehensive quantitative evaluation methods to assess the efficacy and efficiency of telemedicine interventions
Effects of Dietary Patterns and Sodium Intake on Blood Pressure Variability: Results from the DASH and DASH-Sodium Trials
Get PDFBackground
In observational studies, blood pressure (BP) variability has been independently associated with adverse cardiovascular disease outcomes. The Dietary Approaches to Stop Hypertension (DASH) diet and sodium reduction lower BP and cardiovascular risk, but their effects on BP variability (BPV) are uncertain.
Methods
This study assessed the effects of dietary patterns (DASH vs Control) and sodium intake (higher vs lower) on BPV, using both office and 24-hour ambulatory BP measurements in the DASH and DASH-Sodium trials. In primary analyses, week-to-week office BPV and 24-hour ambulatory BPV were quantified using variation independent of the mean (VIM). Between-group comparisons were performed using t-tests; interactive effects between dietary patterns and sodium intake were assessed using multivariate linear regression models.
Results
In pooled analyses of both trials, there was no difference in week-to-week systolic BPV (difference in systolic VIM: 0.49, 95% CI -0.05 to 1.03) or 24-hour systolic BPV (difference in systolic VIM: 0.37, 95% CI -0.13 to 0.87) between the DASH and Control diet groups. In the DASH-Sodium trial, VIM at higher and lower sodium levels did not differ (e.g. difference in VIM for week-to-week systolic BP: 0.31, 95% CI -0.10 to 0.72). No significant combined or interactive effects of diet and sodium level on BPV were observed.
Conclusion
While the DASH diet and reduced sodium intake lower absolute BP levels, these dietary interventions do not significantly affect BPV. These findings suggest that the BP-related benefits of the DASH diet and sodium reduction likely result from reducing absolute BP rather than reducing BPV
Autoencoder-based deep learning methods for single cell and spatial transcriptomics
No full textThe advent of single-cell and spatial transcriptomics has provided unprecedented resolution into cellular heterogeneity but introduced major computational challenges for data integration and interpretation. In this thesis, we develop and explore deep learning-based methods to address these challenges through the lens of autoencoder architectures.
In Chapter 1, we benchmark Variational Autoencoders (VAEs) against traditional linear decomposition methods, demonstrating that VAEs more effectively capture nonlinear biological variation in single-cell RNA-seq datasets. However, we also highlight inherent limitations in interpretability of VAE latent spaces and show through simulation studies that interpretability cannot be guaranteed without additional model constraints.
Building on these findings, Chapter 2 addresses the critical problem of cross-sample integration in spatial transcriptomics. Using the human dorsolateral prefrontal cortex dataset, we evaluate the performance of adversarial domain adaptation methods and show that standard domain classifier approaches may falter when scaling to multiple samples, motivating the need for more stable integration frameworks.
Finally, in Chapter 3, we introduce a novel autoencoder-based model that explicitly disentangles latent representations into common and dataset-specific components. We propose using the Sliced Wasserstein Distance as a stable and interpretable alternative to adversarial losses for multi-sample integration. Proof-of-concept experiments on synthetic datasets show that our approach successfully recovers and disentangles shared and residual signals, laying the groundwork for future applications to real-world biological data.
Overall, this thesis advances the application of deep learning for high-dimensional biomedical data analysis by proposing interpretable, stable, and scalable representation learning frameworks
A COMPARISON OF METHODS TO IDENTIFY INTERVENTION TARGETS TO REDUCE DISPARITIES ACROSS INTERSECTIONAL GROUPS USING CAUSAL DECOMPOSITION ANALYSIS
No full textHealth disparities across intersecting dimensions of race, gender, and socioeconomic status remain a persistent challenge in public health research and practice. Causal decomposition analysis (CDA) offers a principled framework to estimate the extent to which disparities could be reduced through hypothetical interventions on modifiable factors. However, when applied to intersectional populations, CDA faces practical limitations due to small sample sizes within strata, resulting in imprecise estimates and wide confidence intervals. This thesis proposes and evaluates three modeling strategies to address data sparsity in the context of intersectional causal decomposition: (1) fixed effects models, which include indicator variables for each group without borrowing strength; (2) Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA), which stabilizes estimates through partial pooling; and (3) MAIHDA models with interaction terms, which allow for synergistic effects between social identities.
Using data from the NIH All of Us Research Program (n = 22,289), we examined disparities in hypertension control across 4-, 6-, and 18-group intersectional configurations defined by race/ethnicity, gender, and class. We estimated three causal estimands—observed, reduced, and residual disparities—under each model, using g-computation to simulate counterfactual scenarios in which sleep disturbance, a modifiable risk factor, was equalized across groups. Across all stratifications, MAIHDA-based models produced more stable estimates than fixed effects models, particularly in small subgroups. MAIHDA with interaction terms was most sensitive to heterogeneity across intersectional strata, though it yielded wider variability in estimates.
Our results demonstrate that incorporating multilevel modeling into CDA enhances the interpretability and robustness of disparity estimates under data sparsity. While sleep-focused interventions showed limited potential for reducing disparities, the framework developed here can be extended to other intervention targets. This work underscores the value of integrating flexible, equity-aligned modeling strategies into causal inference to guide more effective public health interventions across diverse and intersecting populations
Iron Starvation Stress and Riboflavin in The Fungal Pathogen Candida albicans
Get PDFCandida albicans, a fungal pathogen, adapts to host-imposed iron (Fe) starvation during
infection by activating virulence-associated Fe acquisition pathways. In addition to activating
Fe uptake, C. albicans secretes redox active flavins during Fe-starvation for reasons that are
totally unknown. Both Fe-uptake pathways and flavin secretion are regulated in C. albicans
by a Fe-sensing transcription factor Sef1, but how Sef1 activates flavin secretion is not clear.
This study investigates the role of Sef1 in regulating riboflavin (vitamin B₂) biosynthesis and
secretion under Fe starvation and explores the physiological significance of this process.
Using qRT-PCR, we demonstrated here that Sef1 drives the upregulation of riboflavin
biosynthesis genes (RIB1, RIB3, RIB4) during Fe-starvation conditions, while genes
responsible for converting riboflavin into flavin cofactors (FMN1, FAD1) remain unaffected.
Using fluorescence assays, we confirmed that riboflavin secretion is Sef1-dependent and
specific to Fe-starved WT cells. We tested the hypothesis that redox active riboflavin will
help cells accumulate Fe from the environment by reducing extracellular Fe3+ to the more
soluble and cell accessible Fe2+ form. Our experiments used C. albicans cells exposed to
various sources of Fe3+ in the presence and absence of riboflavin, and the colorimetric
bathophenanthroline disulfonate (BPS) to test intracellular Fe levels. We discovered that
riboflavin can enhance Fe uptake from animal serum under anaerobic conditions, a
physiologically relevant environment for host infection, but not from ionic Fe3+ provided as
FeCl3. Two major Fe3+ binding molecules in serum include transferrin and ferritin, however
our studies indicate that neither serves as target for Fe uptake by riboflavin. The serum
ii
component targeted by riboflavin remains unidentified, suggesting involvement of the large
class of non-transferrin-bound iron (NTBI) molecules or heme. This work advances
understanding of fungal nutritional immunity and opens avenues for targeting riboflavin
pathways in antifungal strategies
DIFFERENTIAL MALARIA RISK IN A HIGH-TRANSMISSION SETTING: LONGITUDINAL EVIDENCE FROM COMMUNITY-LEVEL DATA IN NORTHERN ZAMBIA
Get PDFBackground:
Malaria remains a major public health threat, particularly in sub-Saharan Africa, where 95% of global malaria deaths occur. Despite intensive control efforts, regions like Nchelenge District in northern Zambia experience persistently high malaria transmission. Within these high-burden settings, some individuals are repeatedly infected while others remain uninfected, highlighting the need to understand the drivers of differential malaria risk.
Methods:
We conducted a longitudinal analysis of community-based surveillance data from Nchelenge District. Participants were followed monthly with rapid diagnostic tests (RDTs) and microscopy for Plasmodium falciparum. We compared individuals who tested positive in ≥50% of visits (“high-risk”) to those with ≤1 positive test (“low-risk”) and separately compared those who never tested positive (“never infected”) to individuals infected at least once (“ever infected”). Generalized Estimating Equations (GEE) with an exchangeable correlation structure were used to model associations between malaria risk and individual/household-level factors.
Results:
In multivariate analyses, children aged 5–15 years (OR = 10.77; 95% CI: 1.51–76.87; P = 0.02) and male sex (OR = 2.40; 95% CI: 1.11–5.22; P = 0.03) were significantly associated with high malaria risk. Living in a household with a thatched roof and incomplete adherence to insecticide-treated net (ITN) use showed increased odds of infection, though not statistically significant. When comparing never-infected individuals to those ever infected, residing in lakeside areas was associated with nearly three times the odds of remaining uninfected (OR = 2.95; 95% CI: 1.37–6.34; P < 0.01), while children aged 5–15 had significantly lower odds of being uninfected (OR = 0.38; 95% CI: 0.21–0.71; P < 0.01). Knowledge of malaria transmission and higher ITN adherence were protective in univariate models but lost statistical significance in multivariate analysis.
Conclusions:
Age, sex, geography, and household characteristics contribute to differential malaria risk in Nchelenge District, Zambia. These findings support the need for targeted interventions—particularly among school-aged children—and highlight the importance of understanding household and behavioral factors in sustaining malaria transmission
FAMILY HISTORY OF KIDNEY FAILURE, APOL-1 RISK VARIANTS, SOCIAL DETERMINANTS OF HEALTH, AND RISK OF CHRONIC KIDNEY DISEASE PROGRESSION: FINDINGS FROM THE CRIC STUDY
Get PDFKidney disease is often clustered within families, including Black families, and could be due in part to shared adverse social determinants of health (SDoH) and/or genetic factors. We examined the relationship between family history of kidney failure with SDoH and apolipoprotein L1 (APOL1) risk allele status, and the association between family history of kidney failure and chronic kidney disease (CKD) progression in people with CKD.
5,623 participants from Chronic Renal Insufficiency Cohort Study (CRIC), a longitudinal observational study, were used for this study. The exposure self-reported family history of kidney failure was defined as a first-degree relative treated for kidney failure with dialysis or transplantation. The outcome was CKD progression defined as incident end-stage kidney disease or 50% decline in estimated glomerular filtration rate (eGFR) from baseline. Logistic regression models were used to estimate adjusted odds ratios (aORs) of family history of kidney failure according to race-ethnicity/APOL1 risk allele status and SDoH. Cox proportional hazards models were used to assess the association of family history of kidney failure with the risk of CKD progression.
Among all participants (mean age 59.6 [SD 10.7] years; 43.7% female; 43.1% Black race), 948 (16.9%) reported a family history of kidney failure. Compared to White participants, Black participants were more likely to report a family history of kidney failure regardless of APOL1 status (aOR =2.25 (95% CI: 1.74-2.91) for 0 or 1 risk allele; and aOR=3.46 (95% CI: 2.39-5.02) for 2 risk alleles). Adverse SDoH, such as lower income and lower educational attainment, were positively associated with family history of kidney failure in crude models, but not in multivariable models. In prospective analysis, the family history of kidney failure was significantly associated with an increased risk of CKD progression in both crude (Hazard ratio [HR], 1.33 (95% CI: 1.19-1.49)) and multivariable models adjusting for demographics, APOL1 risk allele status, SDoH, and clinical factors (HR, 1.16 (95% CI: 1.02-1.33)).
These findings highlight the importance of collecting information on family history of kidney failure and further efforts to understand the reasons for familial aggregation of CKD
ASSESSING THE QUALITY OF LIFE OF BREAST CANCER PATIENTS UNDERGOING ADJUVANT ENDOCRINE THERAPY IN INNER MONGOLIA: SELECTING, TRANSLATING, AND VALIDATING A PATIENT-REPORTED OUTCOME MEASURE(PROM)
Get PDFBackground
Breast cancer (BC) remains one of the most prevalent cancers among women. The majority of BC cases are hormone receptor-positive, which requires long-term adjuvant endocrine therapy (AET). While AET effectively reduces recurrence and improves survival, it often causes side effects that may negatively impact patients' health-related quality of life (HRQoL) and result in the discontinuation of treatment. It is essential to assess HRQoL using patient-reported outcome measures (PROMs) to monitor treatment effects and ensure patient well-being. Nevertheless, research on PROMs for Mongolian-speaking BC patients undergoing AET has been limited.
Objective
This study aimed to identify the most appropriate PROM and assess HRQoL in Mongolian BC patients undergoing AET after culturally adapting and validating the selected PROM.
Methods
A systematic review was conducted following the COSMIN guidelines to identify PROMs used for BC patients receiving AET. The most suitable PROM was identified, translated into Mongolian, and then culturally adapted and validated. The evaluation of psychometric properties included assessments of reliability, validity, and responsiveness. Finally, an HRQoL assessment study was conducted to apply the validated PROM to Mongolian BC patients undergoing AET.
Results
The systematic review identified six PROMs, with the Functional Assessment of Cancer Therapy – Endocrine Symptoms (FACT-ES) and the European Organization for Research and Treatment of Cancer Breast Cancer-Specific Quality of Life Questionnaire (EORTC BR45/42) emerging as the most suitable instruments due to comprehensive coverage of AET-related symptoms and strong psychometric properties. The FACT-ES was selected for translation and cross-cultural validation. The cultural adaptation process ensured linguistic and conceptual equivalence, and psychometric validation confirmed good reliability and validity in Mongolian BC patients. The HRQoL assessment study demonstrated that AET significantly impacted various domains of HRQoL, highlighting the importance of tailored patient support strategies.
Conclusions
FACT-ES and EORTC BR45/42 are reliable and valid PROMs for assessing HRQoL in BC patients undergoing AET. The translation and cross-cultural adaptation of the FACT-ES offers the opportunity to assess the HRQoL of Mongolian BC patients on AET. Using the Mongolian FACT-ES will improve our understanding of the HRQoL impacts of AET in BC patients in Mongolia