22691 research outputs found
Sort by
Interactive Evaluation of LLM-based Agents
Full text linkPersonality design plays an important role in chatbot development. From rule-based chatbots to LLM-based chatbots, evaluating the effectiveness of personality design has become more challenging due to the increasingly open-ended interactions. A recent popular approach uses self-report questionnaires to assess LLM-based chatbots' personality traits. However, such an approach has raised serious validity concerns: chatbot's “self-report” personality may not align with human perception based on their interaction. Can LLM-based chatbots “self-report” their personality? We created 500 chatbots with distinct personality designs and evaluated the validity of self-reported personality scales in LLM-based chatbot's personality evaluation. Our findings indicate that the chatbot's answers on human personality scales exhibit weak correlations with both user perception and interaction quality, which raises both criterion and predictive validity concerns of such a method. Further analysis revealed the role of task context and interaction in the chatbot's personality design assessment. We discuss the design implications for building contextualized and interactive evaluation of the chatbot's personality design
Molecular simulation studies of extrinsic and intrinsic mechanism of permeation block in eukaryotic potassium channels
Full text linkIon channels are an extremely important class of proteins. They are crucial to every aspect of life, including sensory perception, movement, thought and memory, and heartbeat. Potassium (K+) channels are the most abundant type of ion channel. Mutations in these proteins are linked to a wide array of neurological, cardiovascular, and metabolic diseases in humans. Research into the molecular mechanisms of these proteins is necessary to understand the origin of these diseases and to identify potential therapeutic strategies. This thesis is specifically focused on investigating intrinsic and extrinsic mechanisms of blocking ion permeation in three distinct K+ channels using Molecular dynamics simulations: human TMEM175, which is key in lysosomal homeostasis; Shaker from Drosophila melanogaster, a prototypical member of the family of eukaryotic voltage-gated K+ channels and important in neuronal function; and Kv2.1 from Rattus norvegicus, another member of the family of voltage-gated K+ channels and also crucial for neuronal function. Investigating the mechanism of inhibition for a known inhibitor, 4-aminopyridine (4-AP) of the non-canonical TMEM175, we learn that 4-AP is recognized by TMEM175 in both a dynamic and specific manner and inhibits permeation in an electrostatic as well as steric manner. We also investigate the recently solved structure of C-type inactivated Shaker and learn that dilation of the filter through repositioning of crucial neighboring residues causes inhibition of rapid ion permeation. Finally, we compare RY785, a potent and specific inhibitor of Kv2.1, to a well characterized inhibitor tetraethylammonium (TEA), and discover that while TEA occludes the pore, RY785 still allows for permeation indicating RY785 facilitates channel closing suggesting an allosteric mechanism of inhibition. The insights gained into mechanisms of intrinsic and extrinsic inhibition can not only aid in the development of more targeted therapeutic approaches and novel drugs for these specific K+ channels but can contribute to a broader understanding of the mechanisms of inhibition and inactivation in other K+ channels, as well as, ion channels in general
Compositional Neuro-Symbolic Reasoning over Natural Language
Full text linkMuch of classical artificial intelligence in the tradition of McCarthy (1959) pursued symbolic theorem proving over carefully hand-crafted ontologies of facts and rules. Due to challenges in scalability and expressivity, such reasoning has fallen out of favor in light of data-driven large language models (LLMs), which capture vast amounts of knowledge and reasoning power by learning patterns over natural language (NL) text corpora. Yet, aspects of the symbolic theorem proving paradigm – namely, mechanistic systematicity, groundedness, and logical underpinnings – are lost in most applications of LLMs, leading to common issues such as hallucinations. This thesis revisits theorem proving in the context of recent advances in LLM reasoning and knowledge retrieval, aiming to leverage the strengths of both the neural and symbolic traditions while addressing their respective pitfalls. I develop a neuro-symbolic entailment engine that reasons over natural language, building on the skeleton of a symbolic prover but leveraging neural models to reason flexibly and dynamically. The entailment engine is designed to systematically determine whether the natural language (NL) hypothesis is compositionally entailed by pieces of evidence provided in large and eadily available text corpora. Through its symbolic search, the engine constructs complex and interpretable entailment trees explaining how complex inferences follow from recursive multi-hop granular inferences, leveraging recent advances in information retrieval and modular reasoning with language models.
While “symbol pushing” engines are semi-guaranteed to arrive at proofs if they exist, the neural variant that I explore in this thesis does not have such a guarantee. NL is infinitely expressive, and by creating an NL proof search, we introduce numerous challenges regarding the kinds of inferences and search branches that are most viable to reason over for a given problem. Moreover, by using neural LLMs to generate inferences, we face challenges in getting them to reason robustly. This thesis explores ways to tackle this large and intractable search problem: using a noisy recent tool for reasoning (LMs) for mechanistic backward chaining proof search
MURINE GUT MICROBIOTA DYSBIOSIS VIA ENTERIC INFECTION MODULATES THE FOREIGN BODY RESPONSE TO A DISTAL BIOMATERIAL IMPLANT
Full text linkImplantable biomaterials and medical devices have transformed modern healthcare, but the long-term success of these implants is often limited by complications induced by the foreign body response (FBR). The FBR is an immune-mediated reaction triggered by any foreign material that is implanted in the body. Researchers have explored various material-based strategies to mitigate the FBR and enhance the long-term biocompatibility of medical implants. However, these approaches do not fully address the variability in FBR severity due to patient-to-patient heterogeneity. Specifically, the influence of host factors such as age, sex, and gut microbiota on the FBR remains largely unexplored.
In this doctoral dissertation, we investigate how gut microbiota dysbiosis via enteric infection can impact the immunological and fibrotic response to a distal biomaterial implant. We demonstrated that enteric infection induced systemic changes in immunity and distal tissue function. Subsequently, we established several experimental models coupling an enteric infection with a distal biomaterial implant, testing different variables such as time of infection, type of biomaterial, age of mice, method of implantation, and type of infection. Across all variables, we demonstrated that enteric infection impacts the immune response to a biomaterial implant, including implant-associated inflammation and immune cell infiltration. However, this largely did not translate to changes in fibrosis around the implant. Collectively, this work establishes that there is immune-mediated communication between the gut microbiota and a distal implant site, suggesting that the gut microbiota may be a potential therapeutic target to mitigate the negative immunological consequences of the FBR
FOREIGN LANGUAGE TRAINING IN THE U.S. DEPARTMENT OF STATE: AN EXOTIC SKILL THAT WAS HARD TO OBTAIN IS NOW A MONSTER THAT IS HARD TO CONTAIN
Full text linkFew Americans know what diplomats even do, which means that even fewer Americans know how they are trained. The people could be forgiven, though- In the year 1906, The Department of State sent the same ambassador to China as the Congo because ambassadors were assigned alphabetically. In the 1950s, things were only marginally better: the Secretary of State, the U.S.’s top diplomat, did not know how many embassy workers in Paris spoke French. Knowing that, one would find it hard to take seriously the Foreign Service Institute, the school that trains America’s diplomatic and consular workforce… despite its best efforts. In reality, the FSI has been on a path of constant improvement since its very inception. After all, it is indeed the place where the officials in the State Department learned their lessons. Some problems in the Department of State are as modern as the times; but others are as old as the nation itself. Diplomats in the United States today are still going out into the field without the skills they need. Some of them do not even speak the language of the country to which they are assigned. Worse still- those at the top of the Department of State have all the knowledge they need to solve the problem; they just lack the budget and the manpower to get it done. In 2024, the Foreign Service Institute was the closest it has ever been to solving the problems within itself, and its Department as a whole… New budget cuts in 2025 threaten this progress, but the school that trains diplomats is showing peerless resolve.
This thesis tells the history of language training and occupational education for diplomats and other foreign affairs personnel. In that history, corruption is exposed, the civil service is professionalized, standards are introduced, and not only is the development of the Foreign Service Institute delivered in a way never before seen in academia; but the ongoings of FSI today in 2025 are written on paper for the first time, ever
GENETIC ANALYSIS OF SOX2 FUNCTION AND CONTRIBUTION TO HAIR CELL REGENERATION IN ZEBRAFISH
Full text linkHearing loss is an escalating global health issue, with projections indicating that over 70 million people will experience some degree of hearing impairment by 2040. While not life-threatening, hearing loss significantly diminishes the quality of life and leads to broader societal challenges, including social isolation, limited employment opportunities, and cognitive decline.
The most common cause of permanent hearing loss is damage to inner ear hair cells (HCs), which are mechanosensory receptors responsible for converting sound vibrations into electrical signals that the brain interprets. In humans, damage to HCs is irreversible leading to deafness and/or balance disorders.
In contrast, non-mammalian species, such as zebrafish, can regenerate HCs after damage or loss throughout life. Due to the evolutionary conservation of inner ear auditory and vestibular organs, the zebrafish is an excellent model organism to study regeneration mechanisms with potential relevance to human therapies. Previous studies have underscored the crucial role of Sox family transcription factors in this process. This study focuses on Sox2, a transcription factor known for its involvement in early development, and its role in reprogramming supporting cells into progenitor cells that give rise to hair cells in zebrafish—a process absent in mammals.
To investigate the role of Sox2 in HC regeneration, we generated sox2 knockouts (KO) using CRISPR-Cas9 gene editing, and validated these KOs using fluorescent PCR, fragment analysis, and Sanger sequencing. Phenotypic analysis of sox2-/- fish revealed developmental abnormalities, including the absence of swim bladders, curved tails, and uncoordinated swimming. The sox2 knockouts also exhibited lethality 12 days post-fertilization. Despite these systematic defects, HC development appeared normal. Moreover, observed no significant defects in HC regeneration following copper-sulfate induced ototoxic damage. Even after conducting two rounds of HC ablation, sox2-/- larvae exhibited robust HC regeneration comparable to wild-type controls.
These findings challenge the presumed requirement of Sox2 for HC regeneration in zebrafish and suggest the presence of compensatory or redundant regulatory mechanisms that sustain regeneration in its absence. Ongoing analyses aim to identify these mechanisms and provide deeper insight into the transcriptional networks governing inner ear HC regeneration
Deciphering Mitochondrial Roles of XIAP During Sublethal and Lethal Cellular Stress
No full textMitochondria play central roles in determining cell fate, balancing between support of survival and triggering of apoptosis in response to cellular stress. The X-linked inhibitor of apoptosis (XIAP) is best known for suppressing caspase activity, yet emerging evidence suggests it also directly regulates mitochondrial stress responses. This thesis sought to investigate how XIAP influences the decision between cell death and survival across varying severities of cellular insults that are sensed by mitochondria. Using immunoblotting and fluorescence microscopy I compared mitochondrial integrity, and apoptosis- and autophagy-related parameters in wild-type and XIAP-deficient cancer cells subjected to controlled stress conditions. My findings reveal insult- and time-dependent differential effects of XIAP deficiency on cellular survival and apoptotic signaling cascades in cancer cells and support a model in which XIAP acts as a molecular switch, sensing the level of mitochondrial damage and steering cells toward apoptosis or survival accordingly. Overall, this work highlights a previously underappreciated dimension of XIAP function, offering new insights into how cells integrate mitochondrial damage signals
Blind Reconstruction of Astronomical Images from Multiple Short Exposures
Full text linkGround-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
CHANGES AND FACTORS AFFECTING INPATIENT SELF-REPORTED QUALITY OF CARE EXPERIENCE IN CHINESE HOSPITALS
Full text linkThis nationwide study investigates patient self-reported quality of care experience—a multidimensional construct integrating subjective perceptions and objective evaluations of healthcare services—within China’s evolving hospital system (2016–2023). Guided by three research aims about patient experience (1. longitudinal trends; 2. relationship with institutional & individual factors; 3. relationship with the People-Centered Care (PCC)), we analyzed 834,546 inpatient records from tertiary, secondary, and community hospitals using multivariate regression and proportional odds models to address hierarchical data structures.
Our key findings reveal the overall patient experience in China in the past few years(2016–2023) is improving (by satisfaction, recognition, recommendations). These improvements likely reflect systemic enhancements in care coordination and PCC implementation across hospital tiers. Our research find out that tertiary hospitals showing more stable improving trends, while secondary hospitals experienced a pre-2020 better performance, and post-2020 declines. General hospitals performed more stable than specialized hospitals. Private hospitals had significantly higher per than public hospitals pre-2020. However, the difference became insignificant(at 5% level) after 2020. Individual factors, especially age and income, have a significant impact on patient experience.
Our study reveals that institutional, personal and process factors collectively influence the patient experience outcome. These findings advocate for composite intervention strategies addressing both systemic constraints and population-specific care expectations.
The study operationalizes PCC through five attributes: continuity (coordinated care pathways), information sharing (transparent communication), enhanced access (reduced barriers), effectiveness (perceived outcomes), and respect (dignity preservation), and confirms the transformative potential of PCC in enhancing patient experience, with care continuity emerging as the most influential attribute.
The findings propose three strategic priorities to sustain care quality improvements: (1) Structural reinforcement of tiered systems through PCC-driven specialization of secondary hospitals in chronic care, supported by digital referral protocols and financial incentives for hierarchical care adherence; (2) Equity-focused resource allocation via tiered insurance reimbursement models and specialist outreach to rural facilities; (3) Technology-augmented process optimization leveraging AI-driven triage systems and real-time patient feedback platforms to standardize dignity-preserving practices. These priorities collectively address systemic fragmentation while operationalizing PCC’s five attributes—as interdependent levers for equitable health system transformation
EVALUATING IMPLEMENTATION OF ANTENATAL MULTIPLE MICRONUTRIENT VERSUS IRON-FOLIC ACID SUPPLEMENTATION WITHIN THE UNRWA HEALTH CARE SYSTEM IN JORDAN
Full text linkMaternal micronutrient deficiencies are prevalent among pregnant Palestine refugee women across the Middle East. Trials demonstrate that antenatal multiple micronutrient supplementation (MMS) reduces adverse birth outcomes versus iron-folic acid (IFA). WHO recommends antenatal MMS accompanied by rigorous research. Accordingly, the United Nations Relief and Works Agency for Palestine Refugees in the Near East (UNRWA) decided to replace antenatal folic acid (FA) and IFA use with MMS, starting in Jordan, guided by implementation research.
Following an enabling period, including a 6-month pilot, we conducted a systems trial comparing antenatal MMS to FA/IFA delivery. Thirteen clinics (two pilot plus 11 randomized) provided 15-nutrient, UNIMMAP-formulated MMS in 180-count bottles at registration. Twelve clinics continued providing daily FA tablets in the 1st trimester and IFA 2-7-times weekly throughout the 2nd-3rd trimesters, in 10-count blister packs. Outcomes included coverage, adherence, side effects, acceptability expressed at anonymous exit interviews, mid-pregnancy hemoglobin (Hb) and anemia, and costs to recipients and the Agency. Deidentified data were abstracted from UNRWA’s eHealth system. The study was approved by the UNRWA Research Review Board and Johns Hopkins IRB.
Over 10 months, 9754 and 5574 women registered in MMS and FA/IFA clinics. Initial coverage was 94.4% for MMS and 70.4/80.6% FA/IFA. Throughout follow-up visits, adherence ≥90% was 81.7-85.9% and 74.5-82.5%, while adherence <70% was 5.1-6.4% and 10.1-17% among MMS and IFA recipients, respectively (all p<0.04). MMS recipients reported fewer side effects and less constipation, although side effects comparably reduced adherence in both groups. More MMS recipients reported improved appetite, feeling healthier and having more energy (all p<0.001). Running out of tablets was less common with MMS than IFA recipients (3.2% versus 27.2%), as was purchasing vitamins (3.3-5.7% versus 13.5-29.6%).
Among 4212 and 2349 registrants with Hb>10 g/dL, the adjusted relative risk (aRR, 95% CI) of mid-pregnancy moderate-to-severe anemia (Hb<10 g/dL) among MMS versus FA/IFA recipients was 0.66 (0.50-0.86). Among registrants with Hb<10 g/dL, adding MMS to IFA anemia treatment improved mid-pregnancy recovery (Hb≥11 g/dL), per aRR=1.29 (1.04-1.59), and required fewer follow-up Hb tests.
These findings showed MMS outperformed FA/IFA, prompting UNRWA to scale up MMS in Agency clinics across the Middle East