Explore the vast range of titles available.

In recent advancements in spoken question answering (QA), end-to-end models have made significant strides. However, previous research has primarily focused on extractive span selection. While this extractive-based approach is effective when answers are present directly within the input, it falls short in addressing abstractive questions, where answers are not directly extracted but inferred from the given information. To bridge this gap, we introduce the first end-to-end Generative Spoken Question Answering (GSQA) model that empowers the system to engage in abstractive reasoning. The challenge in training our GSQA model lies in the absence of a spoken abstractive QA dataset. We propose using text models for initialization and leveraging the extractive QA dataset to transfer knowledge from the text generative model to the spoken generative model. Experimental results indicate that our model surpasses the previous extractive model by 3% on extractive QA datasets. Furthermore, the GSQA model has only been fine-tuned on the spoken extractive QA dataset. Despite not having seen any spoken abstractive QA data, it can still closely match the performance of the cascade model. In conclusion, our GSQA model shows the potential to generalize to a broad spectrum of questions, thus further expanding the spoken question answering capabilities of abstractive QA. Our code is available at https://voidful.github.io/GSQA

Instruction-based speech processing is becoming popular. Studies show that training with multiple tasks boosts performance, but collecting diverse, large-scale tasks and datasets is expensive. Thus, it is highly desirable to design a fundamental task that benefits other downstream tasks. This paper introduces a multi-talker speaking style captioning task to enhance the understanding of speaker and prosodic information. We used large language models to generate descriptions for multi-talker speech. Then, we trained our model with pre-training on this captioning task followed by instruction tuning. Evaluation on Dynamic-SUPERB shows our model outperforming the baseline pre-trained only on single-talker tasks, particularly in speaker and emotion recognition. Additionally, tests on a multi-talker QA task reveal that current models struggle with attributes such as gender, pitch, and speaking rate. The code and dataset are available at https://github.com/cyhuang-tw/speechcaps.

Parameter-efficient methods (like Prompt or Adapters) for adapting pre-trained language models to downstream tasks have been popular recently. However, hindrances still prevent these methods from reaching their full potential. For example, two significant challenges are few-shot adaptation and cross-task generalization ability. To tackle these issues, we propose a general framework to enhance the few-shot adaptation and cross-domain generalization ability of parameter-efficient methods. In our framework, we prime the self-supervised model for parameter-efficient methods to rapidly adapt to various downstream few-shot tasks. To evaluate the authentic generalization ability of these parameter-efficient methods, we conduct experiments on a few-shot cross-domain benchmark containing 160 diverse NLP tasks. The experiment result reveals that priming by tuning PLM only with extra training tasks leads to the best performance. Also, we perform a comprehensive analysis of various parameter-efficient methods under few-shot cross-domain scenarios.

This study proposes a gradient-boosting-based machine learning approach for predicting the PM 2.5 concentration in Taiwan. The proposed mechanism is evaluated on a large-scale database built by the Environmental Protection Administration, and Central Weather Bureau, Taiwan, which includes data from 77 air monitoring stations and 580 weather stations performing hourly measurements over 1 year. By learning from past records of PM 2.5 and neighboring weather stations’ climatic information, the forecasting model works well for 24-h prediction at most air stations. This study also investigates the geographical and meteorological divergence for the forecasting results of seven regional monitoring areas. We also compare the prediction performance between Taiwan, Taipei, and London; analyze the impact of industrial pollution; and propose an enhanced version of the prediction model to improve the prediction accuracy. The results indicate that Taipei and London have similar prediction results because these two cities have similar topography (basin) and are financial centers without domestic pollution sources. The results also suggest that after considering industrial impacts by incorporating additional features from the Taichung and Thong-Siau power plants, the proposed method achieves significant improvement in the coefficient of determination ( R 2 ) from 0.58 to 0.71. Moreover, for Taichung City the root-mean-square error decreases from 8.56 for the conventional approach to 7.06 for the proposed method.

Hepatitis E virus is a common cause of illness worldwide and is associated with severe complications, especially in pregnant women. In this report, the long-term efficacy, immunogenicity, and safety of a hepatitis E vaccine are described. Hepatitis E virus (HEV) is a common cause of acute hepatitis worldwide. 1 , 2 HEV infection occurs in two distinct epidemiologic patterns. 3 The most common pattern is waterborne infection, which is caused by HEV genotype 1 or 2 and occurs mainly in resource-limited countries, often in large, protracted outbreaks or in sporadic cases associated with high mortality among pregnant women. 4 – 6 The other pattern is transmission from animals and humans, which is caused by HEV genotype 3 or 4 and occurs widely in both resource-limited and developed countries. 1 , 7 – 9 Rein et al. 10 estimated the incidence of hepatitis E in areas . . .

Seroprevalence data suggest that a third of the world's population has been infected with the hepatitis E virus. Our aim was to assess efficacy and safety of a recombinant hepatitis E vaccine, HEV 239 (Hecolin; Xiamen Innovax Biotech, Xiamen, China) in a randomised, double-blind, placebo-controlled, phase 3 trial. Healthy adults aged 16–65 years in, Jiangsu Province, China were randomly assigned in a 1:1 ratio to receive three doses of HEV 239 (30 μg of purified recombinant hepatitis E antigen adsorbed to 0·8 mg aluminium hydroxide suspended in 0·5 mL buffered saline) or placebo (hepatitis B vaccine) given intramuscularly at 0, 1, and 6 months. Randomisation was done by computer-generated permuted blocks and stratified by age and sex. Participants were followed up for 19 months. The primary endpoint was prevention of hepatitis E during 12 months from the 31st day after the third dose. Analysis was based on participants who received all three doses per protocol. Study participants, care givers, and investigators were all masked to group and vaccine assignments. This trial is registered with ClinicalTrials.gov, number NCT01014845. 11 165 of the trial participants were tested for hepatitis E virus IgG, of which 5285 (47%) were seropositive for hepatitis E virus. Participants were randomly assigned to vaccine (n=56 302) or placebo (n=56 302). 48 693 (86%) participants in the vaccine group and 48 663 participants (86%) in the placebo group received three vaccine doses and were included in the primary efficacy analysis. During the 12 months after 30 days from receipt of the third dose 15 per-protocol participants in the placebo group developed hepatitis E compared with none in the vaccine group. Vaccine efficacy after three doses was 100·0% (95% CI 72·1–100·0). Adverse effects attributable to the vaccine were few and mild. No vaccination-related serious adverse event was noted. HEV 239 is well tolerated and effective in the prevention of hepatitis E in the general population in China, including both men and women age 16–65 years. Chinese National High-tech R&D Programme (863 programme), Chinese National Key Technologies R&D Programme, Chinese National Science Fund for Distinguished Young Scholars, Fujian Provincial Department of Sciences and Technology, Xiamen Science and Technology Bureau, and Fujian Provincial Science Fund for Distinguished Young Scholars.

Acute pharyngotonsillitis is a common reason to visit primary care providers. Group A Streptococcal (GAS) pharyngitis is the most common bacterial infection which needs antibiotic treatment. GAS accounts for only 10–15% of adult acute pharyngitis cases. The overuse of antibiotics for viral pharyngotonsillitis is common and may lead to inappropriate antimicrobial stewardship and the emergence of bacterial resistance. However, the etiology of acute pharyngotonsillitis for hospitalized adult patients is rarely studied. So, we reported the 10-year surveillance data of hospitalized adult patients with acute pharyngotonsillitis in a regional hospital in Taiwan. Every consecutive adult patient admitted with acute pharyngotonsillitis in 2011–2021 was recruited for a complete etiology study. The etiology of acute pharyngotonsillitis was identified in 117 patients. Overall, 42 herpes simplex virus cases, 26 adenovirus cases, 16 acute human immunodeficiency virus cases, 12 influenza cases, three parainfluenza cases, six Epstein–Barr virus cases, one cytomegalovirus case, four enterovirus cases, one varicella-zoster virus case, four Mycoplasma pneumoniae cases, one Chlamydophila pneumoniae case, and only one GAS case were identified. The average of the points for the Modified Centor Criteria was 1.38 (55% of patients with 0–1 points and 45% with 2–3 points). However, 88.9%of patients received antibiotics at the emergency department, and 76.9%also received antibiotics while hospitalized. Only a few patients required antibiotic treatment, while the majority of patients with viral illness needed only symptomatic treatment. However, distinguishing viral etiology from GAS pharyngitis is challenging even in the presence of tonsil exudates, high C-reactive protein, and leukocytosis. A diagnostic algorithm and the application of the Modified Centor Criteria should be considered for hospitalized adults with acute pharyngotonsillitis to improve antimicrobial stewardship.

Ischemia followed by blood supply reperfusion in cardiomyocytes leads to an overproduction of free radicals and a rapid decrease of adenosine triphosphate concentration. The cardioprotective effect of a potential drug, adenine, was evaluated using H9c2 rat cardiomyoblasts. After hypoxia–reoxygenation (HR) treatment consisting of hypoxia for 21 h followed by reoxygenation for 6 h, it was revealed that pretreatment with 200 µM adenine for 2 h effectively prevented HR-induced cell death. Adenine also significantly decreased the production of reactive oxygen species and reduced cell apoptosis after HR injury. The antioxidant effect of adenine was also revealed in this study. Adenine pretreatment significantly reduced the expression of activating transcription factor 4 (ATF4) and glucose-regulated protein 78 (GRP78) proteins, and protein disulfide isomerase induced a protective effect on mitochondria after HR stimulation. Intracellular adenosine monophosphate-activated protein kinase, peroxisome proliferator-activated receptor delta (PPARδ), and perilipin levels were increased by adenine after HR stimulation. Adenine had a protective effect in HR-damaged H9c2 cells. It may be used in multiple preventive medicines in the future.

Post-translational regulation plays an important role in cellular metabolism. Earlier studies showed that the activity of plastidial starch phosphorylase (Pho1) may be regulated by proteolytic modification. During the purification of Pho1 from sweet potato roots, we observed an unknown high molecular weight complex (HX) showing Pho1 activity. The two-dimensional gel electrophoresis, mass spectrometry, and reverse immunoprecipitation analyses showed that HX is composed of Pho1 and the 20S proteasome. Incubating sweet potato roots at 45°C triggers a stepwise degradation of Pho1; however, the degradation process can be partially inhibited by specific proteasome inhibitor MG132. The proteolytically modified Pho1 displays a lower binding affinity toward glucose 1-phosphate and a reduced starch-synthesizing activity. This study suggests that the 20S proteasome interacts with Pho1 and is involved in the regulation of the catalytic activity of Pho1 in sweet potato roots under heat stress conditions.

The shallow Taiwan Strait at the southern opening of the East China Sea (ECS) receives abundant sediments from turbid mountainous rivers in Taiwan. The volume of sediment is among the highest sediment yields on the global surface. This large amount of sediment discharged from modern Taiwan (range: 175 - 380 Mt y-1 based on 50-yr data) is comparable to that discharged from Changjaing (500 Mt y-1-decreasing in recent decades), underscoring the importance of sediment budget in the Taiwan Strait and sediment flux from Taiwan into the ECS.We documented fluvial mud and sand concentrations during flash flooding with our observations indicating that fluvial materials in Taiwan's rivers are chiefly composed of mud (> 70% and up to 98% ). By contrast, sand fraction dominates (> 85% for most stations) surface sediments in the Taiwan Strait. Super typhoon Herb alone delivered 130 Mt of sediments from Choshui, the largest river in Taiwan, yet only insignificant amounts of mud were found at the river mouth six months later. The actions of waves, tides, and currents apparently prevent the deposition of fine grained sediments. Assuming sand occupied 30% (the maximum) of the 60 Mt y-1 total sediment input from major western Taiwanese rivers, our annual budget estimate shows that the amount of sand input (18 ± 5 Mt y-1) is comparable to the burial output of sand (12 ± 10 Mt y-1). However, mud burial (6 ± 5 Mt y-1) in the strait is far below the estimated mud input (42 ± 11 Mt y-1), resulting in a significant shortfall. Hydrodynamic conditions were synthesized to explain the distribution pattern of limited mud patches in the strait and to reveal potential pathways by which fine-grain sediment transportation takes place in the seas surrounding Taiwan. A significant shortfall in the mud budget in the Taiwan Strait suggests that ~85% of the fluvial mud left the strait. Alternatively, the 50-year modern sediment flux data used in this study reflects exacerbated sediment flux due to human activity and is possibly too high to represent loads during pre-Anthropocene. Additional studies are needed to explore the flux and fate of mud in and surrounding Taiwan over a longer time scale.