299645 research outputs found
Sort by
Generative AI for Deep Reinforcement Learning: Framework, Analysis, and Use Cases
No full textAs a form of artificial intelligence (AI) technology based on interactive learning, deep reinforcement learning (DRL) has been widely applied across various fields and has achieved remarkable accomplishments. However, DRL faces certain limitations, including low sample efficiency and poor generalization. Therefore, in this article, we show how to leverage generative AI (GAI) to address these issues and enhance the performance of DRL algorithms. We first introduce several classic GAI and DRL algorithms and demonstrate the applications of GAI-enhanced DRL algorithms. Then, we discuss how to use GAI to improve DRL algorithms from the data and policy perspectives. Subsequently, we introduce a framework that demonstrates an actual and novel integration of GAI with DRL, that is, GAI-enhanced DRL. Additionally, we provide a case study of the framework for UAV-assisted integrated near-field/far-field communication to validate the performance of the proposed framework. Moreover, we present several future directions. Finally, the related code is available at: https:// xiewenwen22.github.io/GAI-enhanced-DRL
Fully-Decoupled RAN for Feedback-Free Multi-Base Station Transmission in MIMO-OFDM System
No full textCoordinated multi-base station (BS) transmission has emerged as a fundamental access technology to augment network capability and improve spectrum efficiency. However, the computation-intensive feedback of channel state information (CSI) poses significant challenges in determining physical-layer parameters for coordinated BSs. In this paper, we investigate a feedback-free mechanism that leverages fixed precoding matrix indicator (PMI), rank indicator (RI), and channel quality indicator (CQI) for coordinated BS transmission over a fully-decoupled radio access network (FD-RAN). Aiming to maximize user equipment (UE) throughput without CSI feedback, we calculate an optimal feedback-free parameter across spatial, frequency, and time domains only through UE geolocations. First, to determine MIMO transmission layer and precoding strategy in the spatial domain, we introduce a hierarchical reinforcement learning (HRL) framework to jointly select PMI and RI for coordinated BSs. Subsequently, for designing a more fine-grained subband transmission, transformer module is employed to capture the subcarrier correlations within OFDM symbols. Finally, given the unpredictable channel variations, we leverage a diffusion model to generate representative channel for fixed PMI, RI, and CQI over time-varied networks. Simulations demonstrate that 2 BSs feedback-free transmission can enhance 13% throughput compared with 1 BS CLSM transmission, which provides a design principle for next-generation transceiver technologies
Impact of Hyponatraemia at Clinical Stable-State on Survival in Patients with Chronic Obstructive Pulmonary Disease
No full textIntroduction: Hyponatraemia has been suggested to be associated with morbidity and mortality among various medical disorders. Evidence on the association between stable-state hyponatraemia and prognosis in patients with chronic obstructive pulmonary disease (COPD) is lacking. Methods: All COPD patients followed up in a regional hospital in year 2015 were included, with their clinical outcomes reviewed in the subsequent eight years. Association between stable-state hyponatraemia and mortality was evaluated. Stable-state hyponatraemia is defined as baseline serum sodium levels, at least 90 days away from the last AECOPD <135 mmol/L. Results: There were 271 COPD patients included. Hyponatraemia was associated with shorter overall survival with adjusted hazard ratio (aHR) 1.74 (95% CI = 1.07–2.65, p = 0.026). The median overall survival was 3.05 years (95% CI = 2.65–3.46) for patients in the hyponatraemia group, in contrast to 3.35 years (95% CI = 2.86–3.83) for those without hyponatraemia. The highest baseline serum sodium levels were significantly negatively associated with annual acute exacerbation of COPD (AECOPD) and annual hospitalized AECOPD frequency in the follow-up period, with Pearson correlation coefficient of −0.16 (p = 0.011) and – 0.14 (p = 0.027), respectively. Conclusion: Stable-state hyponatraemia was associated with increased mortality and probably AECOPD frequency among patients with COPD.</p
Uplifting ethnic minority voices in the Western Pacific region: a priority for regional suicide research
No full textNeutrophil serine proteases NE and PR3 controlled by the miR-223/STAT3 axis potentiate MASH and liver fibrosis
No full textBackground & Aims: Metabolic dysfunction-associated steatohepatitis (MASH) and its related liver fibrosis represent a substantial public health burden with limited treatment options. Although MASH is associated with enhanced neutrophil infiltration in the liver, the mediators and mechanisms underlying neutrophil-driven progression of MASH and fibrosis remain largely unknown. This study aimed to investigate the role of neutrophil serine proteases neutrophil elastase (NE) and proteinase 3 (PR3) in the development of MASH and fibrosis.Approach: Liver biopsies from 121 morbidly obese patients were recruited for analysis. NE-/-, PR3-/-, microRNA-223 (miR-223)-/- mice and their wild type controls were fed a choline-deficient, L-amino acid-defined, high-fat diet to induce MASH fibrosis. Bone marrow transplantation was performed to generate mice with miR-223 chimerism in bone marrow-derived cells.Results: NE and PR3 content in human liver with MASH and fibrosis was markedly increased in close association with histological features. Genetic ablation or AAV-mediated inhibition of NE and PR3 substantially alleviated MASH and liver fibrosis in mice. Mechanistic study revealed that miR-223 suppressed neutrophilic NE and PR3 by targeting STAT3. MiR-223 deficiency augmented inflammation and fibrosis in mouse liver. Bone marrow transplantation-induced miR-223 chimerism significantly affected hepatic NE/PR3 content and the progression of MASH fibrosis in mice.Conclusions: Our findings reveal that NE and PR3 are key factors triggering liver inflammation to potentiate the development of MASH and liver fibrosis, offering insight into the development of new therapeutic approaches that target NE and PR3.</p
Cyborg microbe biohybrids with metal–organic coating layers: Strategies, functionalisation and potential applications
No full textThe integration of living microbes, specifically bacteria and fungi, with metal–organic nanocoatings has led to the recent development of cyborg microbe biohybrids, which show excellent adaptability and functionality for a wide range of potential applications in biotechnology and medicine. This review discusses the strategies, functionalisation, and applications of these biohybrids, which are categorised into two types of coatings: metal–organic frameworks (MOFs) and metal–phenolic networks (MPNs). Key advances in their synthetic approaches via in-situ and pre-synthesised coatings are crucially addressed, and yet the methodology details and specific advantages are highlighted. Despite the notable advancements, there are various limitations and challenges, such as determination of the long-term viability and stability of the biohybrids, insufficient work on their theranostic applications and essentially scaling-up difficulties for industrial and clinical translation. The latest advancements in the biohybrids and related technology have established a critical foundation for enhancing innovative studies through the strong interdisciplinary teamwork
Monitoring early development and learning of young children in China : findings from adult report and direct assessment
No full textThis thesis evaluates progress towards Sustainable Development Goal (SDG) Target 4.2 in China, which aims to ensure access to quality early childhood development, care, and pre-primary education for all children by 2030. Two indicators are used to monitor progress: Indicator 4.2.1 measures the "proportion of children aged 24-59 months who are developmentally on track in health, learning, and psychosocial well-being," and Indicator 4.2.2 focuses on the "participation rate in organised learning (one year before the official primary entry age)."
Study One analyses China's current status and challenges in meeting SDG Target 4.2. Although significant progress has been made in providing one year of pre-primary education, as evidenced by SDG Target Indicator 4.2.2 data, less attention has been given to measuring SDG Target Indicator 4.2.1. Consequently, there is limited information available on children's developmental status nationwide.
UNICEF developed a parent report measure, the Early Childhood Development Index 2030 (ECDI2030), to assess SDG Target Indicator 4.2.1. However, concerns remain regarding the tool's cross-cultural comparability and neutrality, with no consensus on the optimal method for assessing children's development at the population level. Both direct assessment and adult report have strengths and limitations in generating population-based data on early childhood development. Comparisons between methods can help determine the most effective way of measuring SDG Target Indicator 4.2.1. Previous studies documenting young children's development in China have not systematically compared the reliability and validity of different approaches to measuring child development. Hence, Studies Two, Three, and Four address this research gap.
Study Two conducted a systematic review and meta-analysis of international literature to investigate the concordance between different methods in assessing young children's learning and developmental competencies across contexts. An analysis of 43 studies revealed moderate relations among direct assessment, parent report, and teacher report, bolstering the case for using multiple methods when collecting population-level data on child development.
Study Three measured the developmental status of 309 children (Mage = 49.66, SD = 5.50) using direct assessment and parent report longitudinally. It also compared the results of parent and teacher reports, showing that both overestimated children's direct assessment scores. Parent report was not as sensitive as direct assessment in describing and predicting children's development at disaggregated levels, and teacher report may not reflect the differences in development across socioeconomic status and urbanicity.
Study Four explored the appropriateness of the ECDI2030 in assessing SDG Target Indicator 4.2.1 of 45 children aged 2 years (Mage = 30.06 months, SD = 3.49). The ECDI2030 results were associated with children's urbanicity and showed different developmental patterns across domains. However, not all items have effectively captured the variations in the development of 2-year-olds.
In conclusion, this thesis underscores potential issues in measuring SDG Target Indicator 4.2.1. There are disadvantages associated with relying on a single measure to assess development. Further research is needed to develop psychometrically robust and contextually relevant, population-based measures of early childhood development in China. These measures will continue to support the trend of evidence-based early childhood education policymaking.published_or_final_versionEducationDoctoralDoctor of Philosoph
Leveraging genetic data to explore the effects of antihypertensive drugs on health outcomes
No full textHypertension is a leading risk factor for chronic diseases, affecting over one-third of the population worldwide. Antihypertensive drugs play a crucial role in reducing cardiovascular morbidity and mortality, although different drug classes have varied mechanisms. Growing evidence from experiments, observational studies, and randomized controlled trials has highlighted the effects of antihypertensive drugs on various health outcomes, such as type 2 diabetes, cancer, and mental and neurodegenerative disorders. Although they are recommended for general use in polypills for older people, whether antihypertensive drugs have repurposing potential or have harmful off-target effects remains unclear.
Utilizing large summary genetic datasets, I applied Mendelian randomization (MR)—which is less likely to be affected by confounding than conventional observational studies—to investigate the potential off-target effects of antihypertensive drugs on cancer and mental disorders, as well as their impact on longevity. Genetic proxies for antihypertensive drug classes were obtained from published studies, and their genetic associations with systolic blood pressure were extracted from published genome-wide association studies (GWAS). Publicly available summary statistics for cancer, mental disorders, and longevity were obtained from the Pan-cancer study and cancer consortia, Psychiatric Genomics Consortium, and a GWAS meta-analysis of the UK Biobank and LifeGen, respectively. Finally, I conducted a systematic review of published MR studies to summarize the current state of knowledge on the effects of antihypertensive drugs on different diseases.
This thesis revealed a null association of genetically proxied calcium channel blockers (CCBs) with various cancers, specifically non-Hodgkin’s lymphoma, melanoma, leukemia, thyroid, rectal, pancreatic, oral cavity/pharyngeal, kidney, esophagus/stomach, colon, bladder, endometrial, cervical, breast, prostate, lung, and ovarian cancer. Genetically proxied angiotensin-converting enzyme (ACE) inhibitors were associated with an increased risk of schizophrenia in Europeans and East Asians. The thesis showed robust evidence of positive associations of genetically proxied CCBs, beta-blockers (BBs), and vasodilators with longevity. The protective effects of CCBs and BBs were particularly evident in men. Antihypertensive drug target genes CPT2, CPT1A, CACNB3, ADRB1, CACNA2D2 and EDNRA were related to increased lifespan. In proteome-wide mediation analysis among 2291 proteins, 86 were linked to increased lifespan. CDH1 protein possibly plays a mediating role between CCBs and longevity. Finally, the systematic review included 37 studies, with most studies of high quality. Strong evidence supports the protective effects of antihypertensive drugs on cardiovascular diseases. Genetically proxied ACE inhibitors showed strong protective effects on diabetes, whereas BBs showed adverse effects. Genetically proxied ACE inhibitors could increase the risk of psoriasis, schizophrenia, and Alzheimer’s disease, but they did not affect COVID-19 risk. Genetically proxied ACE inhibitors and CCBs were beneficial for kidney function and immune function, and CCBs also showed a safe profile for disorders of pregnancy.
This thesis offers novel insights into the effects of major antihypertensive drugs on health. Specifically, use of BBs, CCBs and vasodilators can be considered in health policies (i.e., polypills) to reduce the burden of chronic diseases and improve lifespan for the general population. The thesis provides credible evidence for the use of antihypertensive drugs that informs clinical decision-making, pharmacovigilance, and drug repurposing for public health.published_or_final_versionPublic HealthDoctoralDoctor of Philosoph
Achievement and photodynamic of the high-level intersystem crossing in thermally activated delayed fluorescence materials
No full textThermally activated delayed fluorescence (TADF) is widely applied in the luminescence materials and exhibits promising potential in organic light-emitting diode (OLED), time-resolved bioimaging, fluorescent sensors, and other areas. Among the TADF, the one resulting from the high-level intersystem crossing (hRISC) is appealing to the increasing interest recently. Traditionally, the emissive singlet state is generated from the lowest triplet state by reversed intersystem crossing (RISC), while the singlet exciton of hot exciton emission arises from the higher level of triplet state. The luminescence material facilitated with hot exciton emission feature will further enhance the utilization of the triplet exciton and can reach the maximum internal quantum efficiency (IQE) in theory. Although numerous cases of hot exciton emission material have been reported, the excited-state behaviors and photodynamic of them is still insufficient. Here, in this thesis, we investigated the photophysical properties of several luminescence molecules, and directly observed their hRISC process after excitation by femtosecond transient absorption (fs-TA) spectroscopy
In Chapter 2, the difference in luminescence properties for three compounds comprised of benzophenone and triphenylamine was investigated, and the structural-property correlation was clarified. Among three molecules, o-TPA-FBP, m-TPA-FBP, p-TPA-FBP, only the para-substituted one presents the TADF feature. The results suggested that the p-TPA-FBP has a satisfactory ΔEST value and spin-orbital coupling (SOC) effect to realize hRISC, and a clear simultaneous decay of the signal for singlet state and high-level triplet state can be observed. The fs-TA spectra of o-TPA-FBP only presents singlet state signal, as it shows a too large ΔEST and the ISC is forbidden. The SOC effect between T2 and S1 state of m-TPA-FBP is too weak so it will quickly decay to T1 state by radiationless transition, and in the fs-TA spectra, the feature the lowest triplet state will immediately formed in few picoseconds. These results well unveil the mechanism of substituent-position-dependent hot exciton emission.
In Chapter 3, the mechanism of aggregated-induced hot exciton emission was revealed. In this part, the solid-state luminescence properties and photodynamic of DPA-FBP and TPA-FBP were studied in the PMMA doped film with different weight fraction. For these two molecules, only the 50wt% doped film (aggregated state) presents TADF, whereas the 1wt% doped film (single-molecular state) only displays prompt fluorescence. Fs-TA spectroscopy reveals that the 50wt% doped film tends to generate charge transfer (CT) state, whereas as 1wt% doped film shows a relatively long-lived local excited (LE) state. The film of aggregated state shows a much narrower ΔEST than single-molecular state, illustrated from the comparison of fluorescence and phosphorescence peak in photoluminescence spectra, so the efficiency of ISC and RISC will enormously improve. Hence, the key factor for the aggregated-induced hot exciton emission is uncovered to be the CT process in aggregated state.
In Chapter 4, the F-AQ comprised of fluorene and anthraquinone is reported, and it exhibits a polymorphism with muti-color emission and TADF from (hRISC). In the fs-TA result, an unambiguous signal of through space charge transfer (TSCT) was observed for the crystal with the π-π interaction between the fluorene and anthraquinone groups, whereas the other amorphous solids and crystal only show a conventional deactivation pathway of hot exciton emission. This is the first study that directly observed the morphism-dependent TSCT in a crystal to our knowledge,
These studies provide a new insight into the deactivation mechanism and photodynamic of the TADF from hRISC and supplement the deficiency in the ultrafast spectroscopy investigation of solid-state materials.published_or_final_versionChemistryDoctoralDoctor of Philosoph
How do font size and word segmentation influence reading behavior : perspective from eye movement analysis using hidden Markov Models
No full textTo better understand how font size and word segmentation influenced reading behavior, we used the Eye Movement analysis with hidden Markov models (EMHMM) approach, which takes both spatial and temporal dimensions of eye movement behavior into account to quantify participants’ eye movement patterns and consistency.
Study 1 examined how enlarging the font size from a regular size that young children were typically exposed to affect their reading performance and eye movement behavior. We showed that this manipulation impaired word pronunciation accuracy. This impairment was associated with non-verbal IQ but not changes in eye movement behavior, suggesting that it may be more related to the development of font-size dependent perceptual representations of words than information extraction strategies. The font size change also decreased children’s gaze transition consistency among words, suggesting that it interfered with eye movement planning for implementing and developing visual routines for reading. These results suggested that young children may have initially developed font-size dependent perceptual representations for words and for eye movement planning during sentence reading before forming font-size independent reading behavior as observed in adults through increasing reading experiences with different font sizes.
Study 2 examined whether word boundary information, including using space or alternating colors as word boundary markers, could facilitate Chinese sentence reading performance in children. EMHMM revealed two representative gaze transition patterns through clustering: a more sequential and a less sequential pattern. As compared with regular sentences, when using alternating colors as word boundary markers, a larger gaze transition pattern change towards the more sequential pattern was correlated with a larger increase in comprehension accuracy. This effect was not observed when using spaces as word boundary markers. Thus, an instructional design should consider its potential costs on the cognitive processes established during prior learning, as well as the potential individual differences in the ability to adjust the existing processing strategy according to the new instructions.
Study 3 compared large language models (LLMs) and humans in resolving semantic ambiguities by examining the case of segmenting overlapping ambiguous strings in Chinese sentence reading, where three characters “ABC” could be segmented in either “AB/C” or “A/BC” depending on the context. We showed that although LLMs performed worse than humans, they demonstrated a similar interaction effect between segmentation structure and word frequency order, suggesting that this effect could be accounted for by statistical learning of word/token occurrence regularities without assuming an explicit semantic word representation. Nevertheless, across stimuli LLMs’ responses were not correlated with any human performance or eye movement measures, suggesting differences in the underlying processing mechanisms.
In conclusion, here we used EMHMM to understand how font size and word segmentation influenced reading behavior. First, we examined whether manipulating features in reading materials including font size and word boundary information would facilitate children’s reading. The results showed that children have an initial established visual routine for reading, and only individuals with more flexible visual routine would facilitate from the instruction. Then, we examined the performance and processing of word segmentation during Chinese sentence reading in LLMs and adults. The results suggested that it is essential to understand these differences through XAI methods to facilitate LLM adoption.published_or_final_versionPsychologyDoctoralDoctor of Philosoph