Enhancement of Large Language Models' Performance in Diabetes Education: Retrieval-Augmented Generation Approach.

BACKGROUND: Large language models (LLMs) demonstrated advanced performance in processing clinical information. However, commercially available LLMs lack specialized medical knowledge and remain susceptible to generating inaccurate information. Given the need for self-management in diabetes, patients commonly seek information online. We introduce the RISE framework and evaluate its performance in enhancing LLMs to provide accurate responses to diabetes-related inquiries. OBJECTIVE: This study aimed to evaluate the potential of RISE framework, an information retrieval and augmentation tool, to improve the LLM's performance to accurately and safely respond to diabetes-related inquiries. METHODS: The RISE, an innovative retrieval augmentation framework, comprises four steps: Rewriting Query, Information Retrieval, Summarization, and Execution. Using a set of 43 common diabetes-related questions, we evaluated three base LLMs (GPT-4, Anthropic Claude 2, Google Bard) and their RISE-enhanced versions. Assessments were conducted by clinicians for accuracy and comprehensiveness, and by patients for understandability. RESULTS: The integration of RISE significantly improved the accuracy and comprehensiveness of responses from all three based LLMs. On average, the percentage of accurate responses increased by 12% (122 - 107/129) with RISE. Specifically, the rates of accurate responses increased by 7% (42 - 39/43) for GPT-4, 19% (39 - 31/43) for Claude 2, and 9% (41 - 37/43) for Google Bard. The framework also enhanced response comprehensiveness, with mean scores improving by 0.44. Understandability was also enhanced by 0.19 on average. Data collection was conducted from Sept. 30, 2023, to Feb. 05, 2024. CONCLUSIONS: RISE significantly improves LLMs' performance in responding to diabetes-related inquiries, enhancing accuracy, comprehensiveness, and understandability. These improvements have crucial implications for RISE's future role in patient education and chronic illness self-management, which contributes to relieving medical resource pressures and raising public awareness of medical knowledge.

Intraocular mRNA delivery with endogenous MmPEG10-based virus-like particles.

Virus-like particles (VLP) are a promising tool for intracellular gene delivery, yet their potential in ocular gene therapy remains underexplored. In this study, we bridged this knowledge gap by demonstrating the successful generation and application of vesicular stomatitis virus glycoprotein (VSVG)-pseudotyped mouse PEG10 (MmPEG10)-VLP for intraocular mRNA delivery. Our findings revealed that PEG10-VLP can efficiently deliver GFP mRNA to adult retinal pigment epithelial cell line-19 (ARPE-19) cells, leading to transient expression. Moreover, we showed that MmPEG10-VLP can transfer SMAD7 to inhibit epithelial-mesenchymal transition (EMT) in RPE cells effectively. In vivo experiments further substantiated the potential of these vectors, as subretinal delivery into adult mice resulted in efficient transduction of retinal pigment epithelial (RPE) cells and GFP reporter gene expression without significant immune response. However, intravitreal injection did not yield efficient ocular expression. We also evaluated the transduction characteristics of MmPEG10-VLP following intracameral delivery, revealing transient GFP protein expression in corneal endothelial cells without significant immunotoxicities. In summary, our study established that VSVG pseudotyped MmPEG10-based VLP can transduce mitotically inactive RPE cells and corneal endothelial cells in vivo without triggering an inflammatory response, underscoring their potential utility in ocular gene therapy.

Single-cell chromatin accessibility and transcriptomic characterization of Behcet's disease.

Behect's disease is a chronic vasculitis characterized by complex multi-organ immune aberrations. However, a comprehensive understanding of the gene-regulatory profile of peripheral autoimmunity and the diverse immune responses across distinct cell types in Behcet's disease (BD) is still lacking. Here, we present a multi-omic single-cell study of 424,817 cells in BD patients and non-BD individuals. This study maps chromatin accessibility and gene expression in the same biological samples, unraveling vast cellular heterogeneity. We identify widespread cell-type-specific, disease-associated active and pro-inflammatory immunity in both transcript and epigenomic aspects. Notably, integrative multi-omic analysis reveals putative TF regulators that might contribute to chromatin accessibility and gene expression in BD. Moreover, we predicted gene-regulatory networks within nominated TF activators, including AP-1, NF-kB, and ETS transcript factor families, which may regulate cellular interaction and govern inflammation. Our study illustrates the epigenetic and transcriptional landscape in BD peripheral blood and expands understanding of potential epigenomic immunopathology in this disease.

Insights gained from single-cell RNA analysis of murine endothelial cells in aging hearts.

Aging is the strongest risk factor for cardiovascular disease, with progressive decline in the function of vascular endothelial cells (ECs) with age. Systematic analyses of the effects of aging on different cardiac EC types remain limited. Here, we constructed a scRNA atlas of EC transcriptomes in young and old mouse hearts. We identified 10 EC subclusters. The multidimensionally differential genes (DEGs) analysis across different EC clusters shows molecular changes with aging, showing the increase in the overall inflammatory microenvironment and the decrease in angiogenesis and cytoskeletal support capacity of aged ECs. And we performed an in-depth analysis of 3 special ECs, Immunology, Proliferating and Angiogenic. The Immunology EC seems highly associated with some immune regulatory functions, which decline with aging at different degrees. Analysis of two types of neovascular ECs, Proliferating, Angiogenic, implied that Angiogenic ECs can differentiate into multiple EC directions after initially originating from proliferating ECs. And aging leads to a decrease in the ability of vascular angiogenesis and differentiation. Finally, we summarized the effects of aging on cell signaling communication between different EC clusters. This cardiac EC atlas offers comprehensive insights into the molecular regulations of cardiovascular aging, and provides new directions for the prevention and treatment of age-related cardiovascular disease.

Assessing the effects of aging on the renal endothelial cell landscape using single-cell RNA sequencing.

Endothelial cells (ECs) with senescence-associated secretory phenotypes (SASP) have been identified as a key mechanism of aging that contributes to various age-related kidney diseases. In this study, we used single-cell RNA sequencing (scRNA-seq) to create a transcriptome atlas of murine renal ECs and identify transcriptomic changes that occur during aging. We identified seven different subtypes of renal ECs, with glomerular ECs and angiogenic ECs being the most affected by senescence. We confirmed our scRNA-seq findings by using double immunostaining for an EC marker (CD31) and markers of specialized EC phenotypes. Our analysis of the dynamics of capillary lineage development revealed a chronic state of inflammation and compromised glomerular function as prominent aging features. Additionally, we observed an elevated pro-inflammatory and pro-coagulant microenvironment in aged glomerular ECs, which may contribute to age-related glomerulosclerosis and renal fibrosis. Through intercellular communication analysis, we also identified changes in signaling involved in immune regulation that may contribute to a hostile microenvironment for renal homeostasis and function. Overall, our findings provide new insights into the mechanisms of aging in the renal endothelium and may pave the way for the discovery of diagnostic biomarkers and therapeutic interventions against age-related kidney diseases.

Integrative Single-Cell RNA-Seq and ATAC-Seq Analysis of Mouse Corneal Epithelial Cells.

Purpose: Corneal epithelial homeostasis is maintained by coordinated gene expression across distinct cell populations, but the gene regulatory programs underlying this cellular diversity remain to be characterized. Here we applied single-cell multi-omics analysis to delineate the gene regulatory profile of mouse corneal epithelial cells under normal homeostasis. Methods: Single cells isolated from the cornea epithelium (with marginal conjunctiva) of adult mice were subjected to scRNA-seq and scATAC-seq using the 10×Genomics platform. Cell types were clustered by the graph-based visualization method uniform manifold approximation and projection and unbiased computational informatics analysis. The scRNA-seq and scATAC-seq datasets were integrated following the integration pipeline described in ArchR and Seurat. Results: We characterized diverse corneal epithelial cell types based on gene expression signatures and chromatin accessibility. We found that cell type-specific accessibility regions were mainly located at distal regions, suggesting essential roles of distal regulatory elements in determining corneal epithelial cell diversity. Trajectory analyses revealed a continuum of cell state transition and higher coordination between transcription factor (TF) motif accessibility and gene expression during corneal epithelial cell differentiation. By integrating transcriptomic and chromatin accessibility analysis, we identified cell type-specific and shared gene regulation programs. We also uncovered critical TFs driving corneal epithelial cell differentiation, such as nuclear factor I (NFI) family members, Rarg, Elf3. We found that nuclear factor-κB (NF-κB) family members were positive TFs in limbal cells and some superficial cells, but they were involved in regulating distinct biological processes. Conclusions: Our study presents a comprehensive gene regulatory landscape of mouse cornea epithelial cells, and provides valuable foundations for future investigation of corneal epithelial homeostasis in the context of cornea pathologies and regenerative medicine.

Assessing the effects of aging on the liver endothelial cell landscape using single-cell RNA sequencing.

Endothelial cell (EC) function declines with age and contributes to the development of many vascular-related disease processes. Currently, the effects of aging on the molecular regulatory mechanisms of liver ECs have not been fully elucidated. Here, we employed single-cell RNA sequencing to map the transcriptome of ECs and analyzed their relationship with aging. We identified 8 different EC subtypes, interestingly, 2 of which were specially expressed in aged mice ECs namely aged capillary ECs (Aged ECs) and pro-inflammation capillary ECs (Proinfla.ECs). Double immunostaining for an EC marker (Cd31) and a marker of these specialized EC phenotypes confirmed the single-cell RNA sequencing data. Gene ontology analysis revealed that Aged ECs and Proinfla.ECs were associated with inflammatory response. Then we found that liver proliferating capillary ECs (Prolife.ECs) were most affected by senescence. Single-cell transcript analysis suggests that Prolife.ECs and angiogenic capillary ECs may form a poor microenvironment that promotes angiogenesis and tumorigenesis. Pseudo-temporal trajectories revealed that Prolife.ECs have different differentiation pathways in young and aged mice. In aged mice, Prolife.ECs could specifically differentiate into an unstable state, which was mainly composed of angiogenic capillary ECs. Intercellular communication revealed inflammatory activation in old group. Overall, this work compared the single-cell RNA profiles of liver ECs in young and aged mice. These findings provide a new insight into liver aging and its molecular mechanisms, and further exploration of Aged ECs and Proinfla.ECs may help to elucidate the molecular mechanisms associated with senescence.

Chromatin accessibility analysis reveals regulatory dynamics and therapeutic relevance of Vogt-Koyanagi-Harada disease.

The barrier to curing Vogt-Koyanagi-Harada disease (VKH) is thought to reside in a lack of understanding in the roles and regulations of peripheral inflammatory immune cells. Here we perform a single-cell multi-omic study of 166,149 cells in peripheral blood mononuclear cells from patients with VKH, profile the chromatin accessibility and gene expression in the same blood samples, and uncover prominent cellular heterogeneity. Immune cells in VKH blood are highly activated and pro-inflammatory. Notably, we describe an enrichment of transcription targets for nuclear factor kappa B in conventional dendritic cells (cDCs) that governed inflammation. Integrative analysis of transcriptomic and chromatin maps shows that the RELA in cDCs is related to disease complications and poor prognosis. Ligand-receptor interaction pairs also identify cDC as an important predictor that regulated multiple immune subsets. Our results reveal epigenetic and transcriptional dynamics in auto-inflammation, especially the cDC subtype that might lead to therapeutic strategies in VKH.