Neural Machine Translation is one of the key research directions in Natural Language Processing.However,limited by the scale and quality of parallel corpus,the translation quality of low-resource Neural Machine Transl...Neural Machine Translation is one of the key research directions in Natural Language Processing.However,limited by the scale and quality of parallel corpus,the translation quality of low-resource Neural Machine Translation has always been unsatisfactory.When Reinforcement Learning from Human Feedback(RLHF)is applied to lowresource machine translation,commonly encountered issues of substandard preference data quality and the higher cost associated with manual feedback data.Therefore,a more cost-effective method for obtaining feedback data is proposed.At first,optimizing the quality of preference data through the prompt engineering of the Large Language Model(LLM),then combining human feedback to complete the evaluation.In this way,the reward model could acquire more semantic information and human preferences during the training phase,thereby enhancing feedback efficiency and the result’s quality.Experimental results demonstrate that compared with the traditional RLHF method,our method has been proven effective on multiple datasets and exhibits a notable improvement of 1.07 in BLUE.Meanwhile,it is also more favorably received in the assessments conducted by human evaluators and GPT-4o.展开更多
The alarming state of global insulin access in low-resource settings presents a major barrier to diabetes care.A comprehensive review of these challenges is lacking at the global level.To address this weakness,enhance...The alarming state of global insulin access in low-resource settings presents a major barrier to diabetes care.A comprehensive review of these challenges is lacking at the global level.To address this weakness,enhance affordability and build capacity for a more sustainable approach to scaling up access.This review analyzes the specific issue of inconsistent access to insulin in low-and middle-income countries.Using this analysis,we mapped the scope and intensity of issues such as the unaffordability and unavailability of insulin.We also identified six innovative and integrative strategies for increasing and securing accessibility in the areas of policy making,marketing,clinical practice,health education,domestication,and multisectoral approaches.展开更多
Africa faces significant challenges in terms of material and personnel resources for oncology interventions. This is particularly evident in South Africa, where resources are divided into high- and low-resource settin...Africa faces significant challenges in terms of material and personnel resources for oncology interventions. This is particularly evident in South Africa, where resources are divided into high- and low-resource settings. High-resource settings cater to those with financial means to access private oncology facilities. However, many breast cancer patients receive care in South Africa’s low-resource settings, such as public hospital oncology clinics. Unfortunately, these settings have limited service providers and fail to offer comprehensive interventions, resulting in poor outcomes. However, recent research has highlighted the significance of socially supportive relationships in promoting healing and overall individual well-being, and spirituality has been identified as a source of positive outcomes in cancer patients. This systematic review paper explores the feasibility of implementing support group cancer care and interventions that incorporate social support networks available in community settings, and spiritual practices facilitated by traditional healers, and religious/spiritual leaders. These interventions can be provided within low-resource settings to women diagnosed with breast cancer. Inclusive participation of spouses, children, and extended family in the support group cancer care can facilitate healing for the entire system. Focusing on the strengths and resources within communities and incorporating these complementary services, can enhance the well-being and quality of life for Black African women diagnosed with breast cancer, despite low-resource settings. This approach acknowledges the potential of community-based support networks and encourages collaboration between various stakeholders, including community health educators, nurses, lay counselors, and community volunteers, to address the complex needs of these patients.展开更多
Background: Bearing in mind the recent advances in obstetric anesthesia, the safety of both mother and child is of paramount importance, especially in a setting where resources are limited. We set out to find the patt...Background: Bearing in mind the recent advances in obstetric anesthesia, the safety of both mother and child is of paramount importance, especially in a setting where resources are limited. We set out to find the pattern of cases presenting for cesarean delivery and the types of anesthesias provided for the management of these patients. Methods: A retrospective survey was conducted involving all anesthetics provided for cesarean delivery from January 2006 to December 2009 in Ahmadu Bello University Teaching Hospital, Zaria, Nigeria. Information such as age, indications and anesthetic technique, including drugs used, were extracted from patients’ records. Data were subjected to statistical analysis using Statistical Package for Social Sciences (SPSS) version 17.0. Results: There were a total of 577 anaesthetics conducted for cesarean delivery during the period under review out of 4277 live births, giving a cesarean delivery rate of 13.5%. General anesthesia (GA) was administered on 266 (46%) of these patients, while 279 (48%) cases were done under subarachnoid block(SAB). 16 (3%) patients had combined GA and SAB, while 16 (3%) patients received epidural anesthesia. There were 302 emergency cesarean deliveries out of 577 cases, giving an emergency cesarean delivery rate of 52%. The commonest indication for cesarean delivery was two previous cesarean deliveries. Conclusion: A large percentage of our obstetric cases are being done under general anesthesia. Though majority of the conducted regional anesthesia were spinals (SAB), only a few cases were done under epidural block. Subspecialty training of anesthetists will go a long way to improve the current trends.展开更多
Objective: Intravenous labetalol and hydralazine are both considered first-line medications for the management of acute-onset, severe hypertension in pregnant and postpartum women. The study compared the efficacy and ...Objective: Intravenous labetalol and hydralazine are both considered first-line medications for the management of acute-onset, severe hypertension in pregnant and postpartum women. The study compared the efficacy and safety profile of intravenous labetalol and hydralazine in the control hypertension in severe pre-eclampsia. Materials and Methods: One hundred patients who presented with severe pre-eclampsia were randomized into two study groups. The fifty patients in each arm of the study received either intravenous labetalol or intravenous hydralazine for the control of blood pressure. Results: The mean age of the labetalol subjects was 28.6 ± 5.47 years while that of their hydralazine counterparts was 29.12 ± 5.77 years. The majority of respondents in both groups were primigravidae (76% vs. 78%) (P = 0.813). The number of doses of drug needed to significantly lower the mean systolic blood pressure was slightly lower in the labetalol group (2 doses) compared to the hydralazine group (5 doses) (t = 0.803<sup>Y</sup>, P = 0.977). The incidence of headaches which were the commonest complaints was comparable in both groups 8% and 10% of respondents respectively (P > 0.05). Conclusion: Although both intravenous labetalol and hydralazine are useful in patients with severe pre-eclampsia, the response to labetalol was better with comparable side effects.展开更多
Cross-lingual summarization(CLS)is the task of generating a summary in a target language from a document in a source language.Recently,end-to-end CLS models have achieved impressive results using large-scale,high-qual...Cross-lingual summarization(CLS)is the task of generating a summary in a target language from a document in a source language.Recently,end-to-end CLS models have achieved impressive results using large-scale,high-quality datasets typically constructed by translating monolingual summary corpora into CLS corpora.However,due to the limited performance of low-resource language translation models,translation noise can seriously degrade the performance of these models.In this paper,we propose a fine-grained reinforcement learning approach to address low-resource CLS based on noisy data.We introduce the source language summary as a gold signal to alleviate the impact of the translated noisy target summary.Specifically,we design a reinforcement reward by calculating the word correlation and word missing degree between the source language summary and the generated target language summary,and combine it with cross-entropy loss to optimize the CLS model.To validate the performance of our proposed model,we construct Chinese-Vietnamese and Vietnamese-Chinese CLS datasets.Experimental results show that our proposed model outperforms the baselines in terms of both the ROUGE score and BERTScore.展开更多
Event Extraction(EE)is a key task in information extraction,which requires high-quality annotated data that are often costly to obtain.Traditional classification-based methods suffer from low-resource scenarios due to...Event Extraction(EE)is a key task in information extraction,which requires high-quality annotated data that are often costly to obtain.Traditional classification-based methods suffer from low-resource scenarios due to the lack of label semantics and fine-grained annotations.While recent approaches have endeavored to address EE through a more data-efficient generative process,they often overlook event keywords,which are vital for EE.To tackle these challenges,we introduce KeyEE,a multi-prompt learning strategy that improves low-resource event extraction by Event Keywords Extraction(EKE).We suggest employing an auxiliary EKE sub-prompt and concurrently training both EE and EKE with a shared pre-trained language model.With the auxiliary sub-prompt,KeyEE learns event keywords knowledge implicitly,thereby reducing the dependence on annotated data.Furthermore,we investigate and analyze various EKE sub-prompt strategies to encourage further research in this area.Our experiments on benchmark datasets ACE2005 and ERE show that KeyEE achieves significant improvement in low-resource settings and sets new state-of-the-art results.展开更多
Stance detection is the view towards a specific target by a given context(e.g.tweets,commercial reviews).Target-related knowledge is often needed to assist stance detection models in understanding the target well and ...Stance detection is the view towards a specific target by a given context(e.g.tweets,commercial reviews).Target-related knowledge is often needed to assist stance detection models in understanding the target well and making detection correctly.However,prevailing works for knowledge-infused stance detection predominantly incorporate target knowledge from a singular source that lacks knowledge verification in limited domain knowledge.The low-resource training data further increase the challenge for the data-driven large models in this task.To address those challenges,we propose a collaborative knowledge infusion approach for low-resource stance detection tasks,employing a combination of aligned knowledge enhancement and efficient parameter learning techniques.Specifically,our stance detection approach leverages target background knowledge collaboratively from different knowledge sources with the help of knowledge alignment.Additionally,we also introduce the parameter-efficient collaborative adaptor with a staged optimization algorithm,which collaboratively addresses the challenges associated with low-resource stance detection tasks from both network structure and learning perspectives.To assess the effectiveness of our method,we conduct extensive experiments on three public stance detection datasets,including low-resource and cross-target settings.The results demonstrate significant performance improvements compared to the existing stance detection approaches.展开更多
Most State-Of-The-Art(SOTA) Neural Machine Translation(NMT) systems today achieve outstanding results based only on large parallel corpora.The large-scale parallel corpora for high-resource languages is easily obtaina...Most State-Of-The-Art(SOTA) Neural Machine Translation(NMT) systems today achieve outstanding results based only on large parallel corpora.The large-scale parallel corpora for high-resource languages is easily obtainable.However,the translation quality of NMT for morphologically rich languages is still unsatisfactory,mainly because of the data sparsity problem encountered in Low-Resource Languages(LRLs).In the low-resource NMT paradigm,Transfer Learning(TL) has been developed into one of the most efficient methods.It is difficult to train the model on high-resource languages to include the information in both parent and child models,as well as the initially trained model that only contains the lexicon features and word embeddings of the parent model instead of the child languages feature.In this work,we aim to address this issue by proposing the language-independent Hybrid Transfer Learning(HTL) method for LRLs by sharing lexicon embedding between parent and child languages without leveraging back translation or manually injecting noises.First,we train the High-Resource Languages(HRLs) as the parent model with its vocabularies.Then,we combine the parent and child language pairs using the oversampling method to train the hybrid model initialized by the previously parent model.Finally,we fine-tune the morphologically rich child model using a hybrid model.Besides,we explore some exciting discoveries on the original TL approach.Experimental results show that our model consistently outperforms five SOTA methods in two languages Azerbaijani(Az) and Uzbek(Uz).Meanwhile,our approach is practical and significantly better,achieving improvements of up to 4:94 and 4:84 BLEU points for low-resource child languages Az ! Zh and Uz ! Zh,respectively.展开更多
This study proposes a novel AC vector magnetometer developed using a low-resource magneto-impedance sensor for China’s Feng-Yun meteorological satellite(FY-3E).It was calibrated and characterized to determine its per...This study proposes a novel AC vector magnetometer developed using a low-resource magneto-impedance sensor for China’s Feng-Yun meteorological satellite(FY-3E).It was calibrated and characterized to determine its performance parameters.The total weight of the AC vector magnetometer is 51 g(the aluminum box excluded),while the total power consumption is 310 m W.The proposed AC vector magnetometer can detect magnetic field variations in the range of±1000 nT and noise power spectral density of≤50 pT/Hz^(1/2)@1 Hz.Furthermore,the proposed device has a maximum nonlinearity of≤0.71‰over the entire range and a nonorthogonality error of 3.07 nT or 0.15%(root mean square).The total dose hardness of the sensor is≥30 krad(Si).Furthermore,we propose the first survey results of a magnetometer equipped aboard a Chinese FY-3E satellite in a Sunsynchronous orbit.The data revealed that the AC vector magnetometer can detect transient physical signals such as quasistatic field-aligned currents(~50 nT)and waves at the auroral latitudes.These features render the proposed AC vector magnetometer suitable for space-based applications,particularly those involving the study of geomagnetic activity.展开更多
AIM: To evaluate the image quality of a telemedicine screening program for retinal disease using a nonmydriatic camera among rural island communities in Bocas del Toro, Panama.METHODS: In June 2018, a group of three m...AIM: To evaluate the image quality of a telemedicine screening program for retinal disease using a nonmydriatic camera among rural island communities in Bocas del Toro, Panama.METHODS: In June 2018, a group of three medical students volunteered at clinics operated by the Floating Doctors in the province of Bocas del Toro, Panama. Nonmydriatic images of the retina were obtained using the Pictor Plus(Volk Optical, Mentor OH), randomized, and sent to two board-certified ophthalmologists at the University of California, Irvine for analysis using a modified version of the FOTO-ED scale. Inter-rater reliability was calculated using the kappa statistic.RESULTS: Seventy patients provided a total of 127 images. Average image quality was 3.31, and most frequent image quality was 4/5 on the FOTO-ED scale. Thirty patients had at least one eye image with ideal quality(42.86%), while only one patient had no adequate photos taken(1.43%). However, high quality images were obtained in both eyes in only 12 patients(17.14%). The inter-rater reliability between the two ophthalmologists was 0.614.CONCLUSION: Further improvements are necessary to acquire higher quality images more reliably. This may include further training and experience or mydriasis.展开更多
As one of Chinese minority languages,Tibetan speech recognition technology was not researched upon as extensively as Chinese and English were until recently.This,along with the relatively small Tibetan corpus,has resu...As one of Chinese minority languages,Tibetan speech recognition technology was not researched upon as extensively as Chinese and English were until recently.This,along with the relatively small Tibetan corpus,has resulted in an unsatisfying performance of Tibetan speech recognition based on an end-to-end model.This paper aims to achieve an accurate Tibetan speech recognition using a small amount of Tibetan training data.We demonstrate effective methods of Tibetan end-to-end speech recognition via cross-language transfer learning from three aspects:modeling unit selection,transfer learning method,and source language selection.Experimental results show that the Chinese-Tibetan multi-language learning method using multilanguage character set as the modeling unit yields the best performance on Tibetan Character Error Rate(CER)at 27.3%,which is reduced by 26.1%compared to the language-specific model.And our method also achieves the 2.2%higher accuracy using less amount of data compared with the method using Tibetan multi-dialect transfer learning under the same model structure and data set.展开更多
Background: A recent survey of in-hospital reprocessing in Tanzanian hospitals identified bag-valve masks (BVM) as a commonly reused single-use device. In low- and middle-income countries (LMIC), in-hospital reprocess...Background: A recent survey of in-hospital reprocessing in Tanzanian hospitals identified bag-valve masks (BVM) as a commonly reused single-use device. In low- and middle-income countries (LMIC), in-hospital reprocessing supports neonatal resuscitation strategies by helping to maintain adequate supplies of BVM. However, there is a need for device-specific protocols defining reprocessing procedures and inspection criteria to overcome variations in reprocessing practices between hospitals. The purposes of this study were: 1) to complete a comprehensive design review and identify challenges to reprocessing BVMs;and 2) to investigate three different residual bioburden analysis methods for assessing the efficacy of decontaminating a disposable BVM. Methods: New, unused bag-valve-masks were contaminated with Staphylococcus epidermidis and Artificial Mucus Soil to simulate the worst case soiling conditions. Devices underwent one of five disinfection protocols, including one currently used in a LMIC hospital. Three analytical (two quantitative and one qualitative) methods were selected to evaluate residual bioburden on the device following decontamination. Results: Of all protocols tested, only the positive control and the Soap and Bleach protocols met disinfection targets. Most cleaning outcomes were consistent from trial to trial for each protocol. However, cleaning outcomes varied greatly for the Alcohol Wipe protocol. For the residual bioburden analyses, the two quantitative methods produced similar results, but the qualitative measurement exhibited increased variability. Conclusion: While this study revealed positive disinfection outcomes for the Tanzanian hospital decontamination protocol, more studies are required to support these findings. Design features of the BVM mask presented challenges to cleaning and drying during different decontamination protocols, as seen in the variability in the Alcohol Wipe protocol performance. These findings support the case for a device-specific protocol for the BVM. Given proper hospital personnel training and available resources, in-hospital reprocessing could support neonatal resuscitation strategies and other demands for manual resuscitation by helping to maintain adequate supplies of BVM.展开更多
This paper presents a case study of implementing a trauma registry in Mozambique, a low-income country with limited current trauma surveillance. An outline of the importance of trauma registries is presented followed ...This paper presents a case study of implementing a trauma registry in Mozambique, a low-income country with limited current trauma surveillance. An outline of the importance of trauma registries is presented followed by an evidence-based approach to building a sustainable and ethical partnership with local stakeholders.展开更多
Objectives:?Investigating the relation between perinatal outcomes and?hospital working shifts.?Methods:?We conducted a cross-sectional study at Philippe Maguilen Senghor health center (PMSHC) in Dakar, Senegal from Ja...Objectives:?Investigating the relation between perinatal outcomes and?hospital working shifts.?Methods:?We conducted a cross-sectional study at Philippe Maguilen Senghor health center (PMSHC) in Dakar, Senegal from January, 1st?2011 to December, 31th 2018. The study population was comprised of all mothers who had delivered at PMSHC and their newborns after completing 22 weeks of gestation. Time of delivery was?divided into three periods of working hours: morning shift (deliveries occurred between 7 am and 4:59 pm);evening shift from 5 pm to 10:59 pm and night shift from 11?pm to 6:59 am.?Maternal outcomes were assessed by mode of delivery, epsisotomy and perineal injuries.?The Apgar scoring system was used to assess newborns at first minute after they were born. Other adverse perinatal outcomes included fresh stillbirth, neonatal referral and early neonatal death. Data were analyzed using Statistical Package for Social Science software (SPSS 24, Mac version).?Results:?A total of 48,270 mothers and their newborns met eligibility criteria. Caesarean section deliveries were less likely to occur during evening (OR 0.84, 95% CI;0.79?-?0.89, p = 0.001) and night shifts (OR 0.45, CI;0.47?-?0.53, p = 0.001).?Evening shift deliveries had 1.1 the odds of poor perinatal outcome (Apgar score ?- 1.18, p = 0.012). No significant difference was found in the odds of neonate referrals and deaths across the three shifts.?Night shift deliveries had 1.1 the odds of perineal injuries compared to morning shift deliveries (OR 1.11, 95% CI;1.04?- 1.18, p = 0.001, for episiotomy and OR, 1.14;95% CI, 1.04?- 1.26, p = 0.008, for perineal tears). Conclusion:?Off-hours deliveries, particularly during the night shift, were significantly associated with higher proportions of perineal injuries compared to morning shift.?However, no significant difference was found in the odds of neonate referrals and deaths across the three shifts.?Our findings suggest to set up a Neonatology unit at the CSPMS as well as a perinatal network across the country.展开更多
This paper focuses on term-status pair extraction from medical dialogues(MD-TSPE),which is essential in diagnosis dia-logue systems and the automatic scribe of electronic medical records(EMRs).In the past few years,wo...This paper focuses on term-status pair extraction from medical dialogues(MD-TSPE),which is essential in diagnosis dia-logue systems and the automatic scribe of electronic medical records(EMRs).In the past few years,works on MD-TSPE have attracted increasing research attention,especially after the remarkable progress made by generative methods.However,these generative methods output a whole sequence consisting of term-status pairs in one stage and ignore integrating prior knowledge,which demands a deeper un-derstanding to model the relationship between terms and infer the status of each term.This paper presents a knowledge-enhanced two-stage generative framework(KTGF)to address the above challenges.Using task-specific prompts,we employ a single model to com-plete the MD-TSPE through two phases in a unified generative form:We generate all terms the first and then generate the status of each generated term.In this way,the relationship between terms can be learned more effectively from the sequence containing only terms in the first phase,and our designed knowledge-enhanced prompt in the second phase can leverage the category and status candidates of the generated term for status generation.Furthermore,our proposed special status"not mentioned"makes more terms available and en-riches the training data in the second phase,which is critical in the low-resource setting.The experiments on the Chunyu and CMDD datasets show that the proposed method achieves superior results compared to the state-of-the-art models in the full training and low-re-sourcesettings.展开更多
Entity alignment(EA)is an important technique aiming to find the same real entity between two different source knowledge graphs(KGs).Current methods typically learn the embedding of entities for EA from the structure ...Entity alignment(EA)is an important technique aiming to find the same real entity between two different source knowledge graphs(KGs).Current methods typically learn the embedding of entities for EA from the structure of KGs for EA.Most EA models are designed for rich-resource languages,requiring sufficient resources such as a parallel corpus and pre-trained language models.However,low-resource language KGs have received less attention,and current models demonstrate poor performance on those low-resource KGs.Recently,researchers have fused relation information and attributes for entity representations to enhance the entity alignment performance,but the relation semantics are often ignored.To address these issues,we propose a novel Semantic-aware Graph Neural Network(SGNN)for entity alignment.First,we generate pseudo sentences according to the relation triples and produce representations using pre-trained models.Second,our approach explores semantic information from the connected relations by a graph neural network.Our model captures expanded feature information from KGs.Experimental results using three low-resource languages demonstrate that our proposed SGNN approach out performs better than state-of-the-art alignment methods on three proposed datasets and three public datasets.展开更多
Machine translation(MT)is a technique that leverages computers to translate human languages automatically.Nowadays,neural machine translation(NMT)which models direct mapping between source and target languages with de...Machine translation(MT)is a technique that leverages computers to translate human languages automatically.Nowadays,neural machine translation(NMT)which models direct mapping between source and target languages with deep neural networks has achieved a big breakthrough in translation performance and become the de facto paradigm of MT.This article makes a review of NMT framework,discusses the challenges in NMT,introduces some exciting recent progresses and finally looks forward to some potential future research trends.展开更多
Background In 2003, China＇s National Free Antiretroviral Treatment Program （NFATP） was initiated as a pilot, which covered only 100 HIV/AIDS patients. By 2011, the pilot had evolved into a nationwide program and ha...Background In 2003, China＇s National Free Antiretroviral Treatment Program （NFATP） was initiated as a pilot, which covered only 100 HIV/AIDS patients. By 2011, the pilot had evolved into a nationwide program and had provided free treatment for over 150 000 patients. The objective of this study was to report and evaluate the progress of China＇s free antiretroviral treatment program. Methods The NFATP Database was systematically reviewed and a total of 150 692 HIV/AIDS patients were included in this study. Program progress indicators including the number of treated HIV/AIDS patients, follow-up visit rate, CD4 test rate, and viral load test rate were summarized and examined over a calendar year to evaluate the progress of NFATP quantitatively and qualitatively. Results By the end of 2011, a total of 150 692 HIV/AIDS patients had been treated through the NFATP and 122 613 of them were still on treatment. Of all patients, about 72% were enrolled during the past four years. The dominant transmission route was blood related in the early phase of the NFATP, but gradually changed to sexual contact. Besides quantitative improvements, progress indicators also demonstrated significant qualitative improvements that the program had made during the past 9 years. Conclusions Great achievement has been made by China＇s NFATP. China＇s experience indicates the importance of a comprehensive response to the success of its treatment program. However, to ensure the quality and sustainability of treatment in the long term, more attention and resources should be paid towards program management.展开更多
Statistical machine translation for low-resource language suffers from the lack of abundant training corpora. Several methods, such as the use of a pivot language, have been proposed as a bridge to translate from one ...Statistical machine translation for low-resource language suffers from the lack of abundant training corpora. Several methods, such as the use of a pivot language, have been proposed as a bridge to translate from one language to another. However, errors will accumulate during the extensive translation pipelines. In this paper, we propose an approach to low-resource language translation by exploiting the pronunciation correlations between languages. We find that the pronunciation features can improve both Chinese-Vietnamese and Vietnamese- Chinese translation qualities. Experimental results show that our proposed model yields effective improvements, and the translation performance （bilingual evaluation understudy score） is improved by a maximum value of 1.03.展开更多
基金supported by the National Natural Science Foundation of China under Grant No.61862064.
文摘Neural Machine Translation is one of the key research directions in Natural Language Processing.However,limited by the scale and quality of parallel corpus,the translation quality of low-resource Neural Machine Translation has always been unsatisfactory.When Reinforcement Learning from Human Feedback(RLHF)is applied to lowresource machine translation,commonly encountered issues of substandard preference data quality and the higher cost associated with manual feedback data.Therefore,a more cost-effective method for obtaining feedback data is proposed.At first,optimizing the quality of preference data through the prompt engineering of the Large Language Model(LLM),then combining human feedback to complete the evaluation.In this way,the reward model could acquire more semantic information and human preferences during the training phase,thereby enhancing feedback efficiency and the result’s quality.Experimental results demonstrate that compared with the traditional RLHF method,our method has been proven effective on multiple datasets and exhibits a notable improvement of 1.07 in BLUE.Meanwhile,it is also more favorably received in the assessments conducted by human evaluators and GPT-4o.
文摘The alarming state of global insulin access in low-resource settings presents a major barrier to diabetes care.A comprehensive review of these challenges is lacking at the global level.To address this weakness,enhance affordability and build capacity for a more sustainable approach to scaling up access.This review analyzes the specific issue of inconsistent access to insulin in low-and middle-income countries.Using this analysis,we mapped the scope and intensity of issues such as the unaffordability and unavailability of insulin.We also identified six innovative and integrative strategies for increasing and securing accessibility in the areas of policy making,marketing,clinical practice,health education,domestication,and multisectoral approaches.
文摘Africa faces significant challenges in terms of material and personnel resources for oncology interventions. This is particularly evident in South Africa, where resources are divided into high- and low-resource settings. High-resource settings cater to those with financial means to access private oncology facilities. However, many breast cancer patients receive care in South Africa’s low-resource settings, such as public hospital oncology clinics. Unfortunately, these settings have limited service providers and fail to offer comprehensive interventions, resulting in poor outcomes. However, recent research has highlighted the significance of socially supportive relationships in promoting healing and overall individual well-being, and spirituality has been identified as a source of positive outcomes in cancer patients. This systematic review paper explores the feasibility of implementing support group cancer care and interventions that incorporate social support networks available in community settings, and spiritual practices facilitated by traditional healers, and religious/spiritual leaders. These interventions can be provided within low-resource settings to women diagnosed with breast cancer. Inclusive participation of spouses, children, and extended family in the support group cancer care can facilitate healing for the entire system. Focusing on the strengths and resources within communities and incorporating these complementary services, can enhance the well-being and quality of life for Black African women diagnosed with breast cancer, despite low-resource settings. This approach acknowledges the potential of community-based support networks and encourages collaboration between various stakeholders, including community health educators, nurses, lay counselors, and community volunteers, to address the complex needs of these patients.
文摘Background: Bearing in mind the recent advances in obstetric anesthesia, the safety of both mother and child is of paramount importance, especially in a setting where resources are limited. We set out to find the pattern of cases presenting for cesarean delivery and the types of anesthesias provided for the management of these patients. Methods: A retrospective survey was conducted involving all anesthetics provided for cesarean delivery from January 2006 to December 2009 in Ahmadu Bello University Teaching Hospital, Zaria, Nigeria. Information such as age, indications and anesthetic technique, including drugs used, were extracted from patients’ records. Data were subjected to statistical analysis using Statistical Package for Social Sciences (SPSS) version 17.0. Results: There were a total of 577 anaesthetics conducted for cesarean delivery during the period under review out of 4277 live births, giving a cesarean delivery rate of 13.5%. General anesthesia (GA) was administered on 266 (46%) of these patients, while 279 (48%) cases were done under subarachnoid block(SAB). 16 (3%) patients had combined GA and SAB, while 16 (3%) patients received epidural anesthesia. There were 302 emergency cesarean deliveries out of 577 cases, giving an emergency cesarean delivery rate of 52%. The commonest indication for cesarean delivery was two previous cesarean deliveries. Conclusion: A large percentage of our obstetric cases are being done under general anesthesia. Though majority of the conducted regional anesthesia were spinals (SAB), only a few cases were done under epidural block. Subspecialty training of anesthetists will go a long way to improve the current trends.
文摘Objective: Intravenous labetalol and hydralazine are both considered first-line medications for the management of acute-onset, severe hypertension in pregnant and postpartum women. The study compared the efficacy and safety profile of intravenous labetalol and hydralazine in the control hypertension in severe pre-eclampsia. Materials and Methods: One hundred patients who presented with severe pre-eclampsia were randomized into two study groups. The fifty patients in each arm of the study received either intravenous labetalol or intravenous hydralazine for the control of blood pressure. Results: The mean age of the labetalol subjects was 28.6 ± 5.47 years while that of their hydralazine counterparts was 29.12 ± 5.77 years. The majority of respondents in both groups were primigravidae (76% vs. 78%) (P = 0.813). The number of doses of drug needed to significantly lower the mean systolic blood pressure was slightly lower in the labetalol group (2 doses) compared to the hydralazine group (5 doses) (t = 0.803<sup>Y</sup>, P = 0.977). The incidence of headaches which were the commonest complaints was comparable in both groups 8% and 10% of respondents respectively (P > 0.05). Conclusion: Although both intravenous labetalol and hydralazine are useful in patients with severe pre-eclampsia, the response to labetalol was better with comparable side effects.
基金Project supported by the National Natural Science Foundation of China(Nos.U21B2027,62266027,61972186,62241604)the Yunnan Provincial Major Science and Technology Special Plan Projects,China(Nos.202302AD080003,202103AA080015,and 202202AD080003)+1 种基金the General Projects of Basic Research in Yunnan Province,China(Nos.202301AT070471 and 202301AT070393)the Kunming University of Science and Technology“Double First-Class”Joint Project,China(No.202201BE070001-021)。
文摘Cross-lingual summarization(CLS)is the task of generating a summary in a target language from a document in a source language.Recently,end-to-end CLS models have achieved impressive results using large-scale,high-quality datasets typically constructed by translating monolingual summary corpora into CLS corpora.However,due to the limited performance of low-resource language translation models,translation noise can seriously degrade the performance of these models.In this paper,we propose a fine-grained reinforcement learning approach to address low-resource CLS based on noisy data.We introduce the source language summary as a gold signal to alleviate the impact of the translated noisy target summary.Specifically,we design a reinforcement reward by calculating the word correlation and word missing degree between the source language summary and the generated target language summary,and combine it with cross-entropy loss to optimize the CLS model.To validate the performance of our proposed model,we construct Chinese-Vietnamese and Vietnamese-Chinese CLS datasets.Experimental results show that our proposed model outperforms the baselines in terms of both the ROUGE score and BERTScore.
基金supported by the National Key Research and Development Program of China(No.2021YFF1201200)the Science and Technology Major Project of Changsha(No.kh2202004)the Natural Science Foundation of China(No.62006251)。
文摘Event Extraction(EE)is a key task in information extraction,which requires high-quality annotated data that are often costly to obtain.Traditional classification-based methods suffer from low-resource scenarios due to the lack of label semantics and fine-grained annotations.While recent approaches have endeavored to address EE through a more data-efficient generative process,they often overlook event keywords,which are vital for EE.To tackle these challenges,we introduce KeyEE,a multi-prompt learning strategy that improves low-resource event extraction by Event Keywords Extraction(EKE).We suggest employing an auxiliary EKE sub-prompt and concurrently training both EE and EKE with a shared pre-trained language model.With the auxiliary sub-prompt,KeyEE learns event keywords knowledge implicitly,thereby reducing the dependence on annotated data.Furthermore,we investigate and analyze various EKE sub-prompt strategies to encourage further research in this area.Our experiments on benchmark datasets ACE2005 and ERE show that KeyEE achieves significant improvement in low-resource settings and sets new state-of-the-art results.
基金supported by the RCA founding of A*STAR and DSO National Laboratory(Nos.2208-526-RCA-CFAR and SC23/22-3204FA)。
文摘Stance detection is the view towards a specific target by a given context(e.g.tweets,commercial reviews).Target-related knowledge is often needed to assist stance detection models in understanding the target well and making detection correctly.However,prevailing works for knowledge-infused stance detection predominantly incorporate target knowledge from a singular source that lacks knowledge verification in limited domain knowledge.The low-resource training data further increase the challenge for the data-driven large models in this task.To address those challenges,we propose a collaborative knowledge infusion approach for low-resource stance detection tasks,employing a combination of aligned knowledge enhancement and efficient parameter learning techniques.Specifically,our stance detection approach leverages target background knowledge collaboratively from different knowledge sources with the help of knowledge alignment.Additionally,we also introduce the parameter-efficient collaborative adaptor with a staged optimization algorithm,which collaboratively addresses the challenges associated with low-resource stance detection tasks from both network structure and learning perspectives.To assess the effectiveness of our method,we conduct extensive experiments on three public stance detection datasets,including low-resource and cross-target settings.The results demonstrate significant performance improvements compared to the existing stance detection approaches.
基金supported by the National Key R&D Program of China (No. 2017YFB0202204)the National Natural Science Foundation of China (Nos. 61925601, 61761166008, and 61772302)+1 种基金Beijing Advanced Innovation Center for Language Resources (No. TYR17002)the NExT ++ project which supported by the National Research Foundation, Prime Ministers Office, Singapore under its IRC@Singapore Funding Initiative。
文摘Most State-Of-The-Art(SOTA) Neural Machine Translation(NMT) systems today achieve outstanding results based only on large parallel corpora.The large-scale parallel corpora for high-resource languages is easily obtainable.However,the translation quality of NMT for morphologically rich languages is still unsatisfactory,mainly because of the data sparsity problem encountered in Low-Resource Languages(LRLs).In the low-resource NMT paradigm,Transfer Learning(TL) has been developed into one of the most efficient methods.It is difficult to train the model on high-resource languages to include the information in both parent and child models,as well as the initially trained model that only contains the lexicon features and word embeddings of the parent model instead of the child languages feature.In this work,we aim to address this issue by proposing the language-independent Hybrid Transfer Learning(HTL) method for LRLs by sharing lexicon embedding between parent and child languages without leveraging back translation or manually injecting noises.First,we train the High-Resource Languages(HRLs) as the parent model with its vocabularies.Then,we combine the parent and child language pairs using the oversampling method to train the hybrid model initialized by the previously parent model.Finally,we fine-tune the morphologically rich child model using a hybrid model.Besides,we explore some exciting discoveries on the original TL approach.Experimental results show that our model consistently outperforms five SOTA methods in two languages Azerbaijani(Az) and Uzbek(Uz).Meanwhile,our approach is practical and significantly better,achieving improvements of up to 4:94 and 4:84 BLEU points for low-resource child languages Az ! Zh and Uz ! Zh,respectively.
基金supported by the National Natural Science Foundation of China (Grant No.42074223)。
文摘This study proposes a novel AC vector magnetometer developed using a low-resource magneto-impedance sensor for China’s Feng-Yun meteorological satellite(FY-3E).It was calibrated and characterized to determine its performance parameters.The total weight of the AC vector magnetometer is 51 g(the aluminum box excluded),while the total power consumption is 310 m W.The proposed AC vector magnetometer can detect magnetic field variations in the range of±1000 nT and noise power spectral density of≤50 pT/Hz^(1/2)@1 Hz.Furthermore,the proposed device has a maximum nonlinearity of≤0.71‰over the entire range and a nonorthogonality error of 3.07 nT or 0.15%(root mean square).The total dose hardness of the sensor is≥30 krad(Si).Furthermore,we propose the first survey results of a magnetometer equipped aboard a Chinese FY-3E satellite in a Sunsynchronous orbit.The data revealed that the AC vector magnetometer can detect transient physical signals such as quasistatic field-aligned currents(~50 nT)and waves at the auroral latitudes.These features render the proposed AC vector magnetometer suitable for space-based applications,particularly those involving the study of geomagnetic activity.
文摘AIM: To evaluate the image quality of a telemedicine screening program for retinal disease using a nonmydriatic camera among rural island communities in Bocas del Toro, Panama.METHODS: In June 2018, a group of three medical students volunteered at clinics operated by the Floating Doctors in the province of Bocas del Toro, Panama. Nonmydriatic images of the retina were obtained using the Pictor Plus(Volk Optical, Mentor OH), randomized, and sent to two board-certified ophthalmologists at the University of California, Irvine for analysis using a modified version of the FOTO-ED scale. Inter-rater reliability was calculated using the kappa statistic.RESULTS: Seventy patients provided a total of 127 images. Average image quality was 3.31, and most frequent image quality was 4/5 on the FOTO-ED scale. Thirty patients had at least one eye image with ideal quality(42.86%), while only one patient had no adequate photos taken(1.43%). However, high quality images were obtained in both eyes in only 12 patients(17.14%). The inter-rater reliability between the two ophthalmologists was 0.614.CONCLUSION: Further improvements are necessary to acquire higher quality images more reliably. This may include further training and experience or mydriasis.
基金This work was supported by three projects.Zhao Y received the Grant with Nos.61976236 and 2020MDJC06Bi X J received the Grant with No.20&ZD279.
文摘As one of Chinese minority languages,Tibetan speech recognition technology was not researched upon as extensively as Chinese and English were until recently.This,along with the relatively small Tibetan corpus,has resulted in an unsatisfying performance of Tibetan speech recognition based on an end-to-end model.This paper aims to achieve an accurate Tibetan speech recognition using a small amount of Tibetan training data.We demonstrate effective methods of Tibetan end-to-end speech recognition via cross-language transfer learning from three aspects:modeling unit selection,transfer learning method,and source language selection.Experimental results show that the Chinese-Tibetan multi-language learning method using multilanguage character set as the modeling unit yields the best performance on Tibetan Character Error Rate(CER)at 27.3%,which is reduced by 26.1%compared to the language-specific model.And our method also achieves the 2.2%higher accuracy using less amount of data compared with the method using Tibetan multi-dialect transfer learning under the same model structure and data set.
文摘Background: A recent survey of in-hospital reprocessing in Tanzanian hospitals identified bag-valve masks (BVM) as a commonly reused single-use device. In low- and middle-income countries (LMIC), in-hospital reprocessing supports neonatal resuscitation strategies by helping to maintain adequate supplies of BVM. However, there is a need for device-specific protocols defining reprocessing procedures and inspection criteria to overcome variations in reprocessing practices between hospitals. The purposes of this study were: 1) to complete a comprehensive design review and identify challenges to reprocessing BVMs;and 2) to investigate three different residual bioburden analysis methods for assessing the efficacy of decontaminating a disposable BVM. Methods: New, unused bag-valve-masks were contaminated with Staphylococcus epidermidis and Artificial Mucus Soil to simulate the worst case soiling conditions. Devices underwent one of five disinfection protocols, including one currently used in a LMIC hospital. Three analytical (two quantitative and one qualitative) methods were selected to evaluate residual bioburden on the device following decontamination. Results: Of all protocols tested, only the positive control and the Soap and Bleach protocols met disinfection targets. Most cleaning outcomes were consistent from trial to trial for each protocol. However, cleaning outcomes varied greatly for the Alcohol Wipe protocol. For the residual bioburden analyses, the two quantitative methods produced similar results, but the qualitative measurement exhibited increased variability. Conclusion: While this study revealed positive disinfection outcomes for the Tanzanian hospital decontamination protocol, more studies are required to support these findings. Design features of the BVM mask presented challenges to cleaning and drying during different decontamination protocols, as seen in the variability in the Alcohol Wipe protocol performance. These findings support the case for a device-specific protocol for the BVM. Given proper hospital personnel training and available resources, in-hospital reprocessing could support neonatal resuscitation strategies and other demands for manual resuscitation by helping to maintain adequate supplies of BVM.
文摘This paper presents a case study of implementing a trauma registry in Mozambique, a low-income country with limited current trauma surveillance. An outline of the importance of trauma registries is presented followed by an evidence-based approach to building a sustainable and ethical partnership with local stakeholders.
文摘Objectives:?Investigating the relation between perinatal outcomes and?hospital working shifts.?Methods:?We conducted a cross-sectional study at Philippe Maguilen Senghor health center (PMSHC) in Dakar, Senegal from January, 1st?2011 to December, 31th 2018. The study population was comprised of all mothers who had delivered at PMSHC and their newborns after completing 22 weeks of gestation. Time of delivery was?divided into three periods of working hours: morning shift (deliveries occurred between 7 am and 4:59 pm);evening shift from 5 pm to 10:59 pm and night shift from 11?pm to 6:59 am.?Maternal outcomes were assessed by mode of delivery, epsisotomy and perineal injuries.?The Apgar scoring system was used to assess newborns at first minute after they were born. Other adverse perinatal outcomes included fresh stillbirth, neonatal referral and early neonatal death. Data were analyzed using Statistical Package for Social Science software (SPSS 24, Mac version).?Results:?A total of 48,270 mothers and their newborns met eligibility criteria. Caesarean section deliveries were less likely to occur during evening (OR 0.84, 95% CI;0.79?-?0.89, p = 0.001) and night shifts (OR 0.45, CI;0.47?-?0.53, p = 0.001).?Evening shift deliveries had 1.1 the odds of poor perinatal outcome (Apgar score ?- 1.18, p = 0.012). No significant difference was found in the odds of neonate referrals and deaths across the three shifts.?Night shift deliveries had 1.1 the odds of perineal injuries compared to morning shift deliveries (OR 1.11, 95% CI;1.04?- 1.18, p = 0.001, for episiotomy and OR, 1.14;95% CI, 1.04?- 1.26, p = 0.008, for perineal tears). Conclusion:?Off-hours deliveries, particularly during the night shift, were significantly associated with higher proportions of perineal injuries compared to morning shift.?However, no significant difference was found in the odds of neonate referrals and deaths across the three shifts.?Our findings suggest to set up a Neonatology unit at the CSPMS as well as a perinatal network across the country.
基金This work was supported by the Key Research Program of the Chinese Academy of Sciences(No.ZDBSSSW-JSC006)the National Natural Science Foundation of China(No.62206294).
文摘This paper focuses on term-status pair extraction from medical dialogues(MD-TSPE),which is essential in diagnosis dia-logue systems and the automatic scribe of electronic medical records(EMRs).In the past few years,works on MD-TSPE have attracted increasing research attention,especially after the remarkable progress made by generative methods.However,these generative methods output a whole sequence consisting of term-status pairs in one stage and ignore integrating prior knowledge,which demands a deeper un-derstanding to model the relationship between terms and infer the status of each term.This paper presents a knowledge-enhanced two-stage generative framework(KTGF)to address the above challenges.Using task-specific prompts,we employ a single model to com-plete the MD-TSPE through two phases in a unified generative form:We generate all terms the first and then generate the status of each generated term.In this way,the relationship between terms can be learned more effectively from the sequence containing only terms in the first phase,and our designed knowledge-enhanced prompt in the second phase can leverage the category and status candidates of the generated term for status generation.Furthermore,our proposed special status"not mentioned"makes more terms available and en-riches the training data in the second phase,which is critical in the low-resource setting.The experiments on the Chunyu and CMDD datasets show that the proposed method achieves superior results compared to the state-of-the-art models in the full training and low-re-sourcesettings.
基金National Natural Science Foundation of China(Nos.U21B2027,61972186,61732005)Major Science and Technology Projects of Yunnan Province(Nos.202202AD080003,202203AA080004).
文摘Entity alignment(EA)is an important technique aiming to find the same real entity between two different source knowledge graphs(KGs).Current methods typically learn the embedding of entities for EA from the structure of KGs for EA.Most EA models are designed for rich-resource languages,requiring sufficient resources such as a parallel corpus and pre-trained language models.However,low-resource language KGs have received less attention,and current models demonstrate poor performance on those low-resource KGs.Recently,researchers have fused relation information and attributes for entity representations to enhance the entity alignment performance,but the relation semantics are often ignored.To address these issues,we propose a novel Semantic-aware Graph Neural Network(SGNN)for entity alignment.First,we generate pseudo sentences according to the relation triples and produce representations using pre-trained models.Second,our approach explores semantic information from the connected relations by a graph neural network.Our model captures expanded feature information from KGs.Experimental results using three low-resource languages demonstrate that our proposed SGNN approach out performs better than state-of-the-art alignment methods on three proposed datasets and three public datasets.
基金the National Natural Science Foundation of China(Grant Nos.U1836221 and 61673380)the Beijing Municipal Science and Technology Project(Grant No.Z181100008918017)。
文摘Machine translation(MT)is a technique that leverages computers to translate human languages automatically.Nowadays,neural machine translation(NMT)which models direct mapping between source and target languages with deep neural networks has achieved a big breakthrough in translation performance and become the de facto paradigm of MT.This article makes a review of NMT framework,discusses the challenges in NMT,introduces some exciting recent progresses and finally looks forward to some potential future research trends.
文摘Background In 2003, China＇s National Free Antiretroviral Treatment Program （NFATP） was initiated as a pilot, which covered only 100 HIV/AIDS patients. By 2011, the pilot had evolved into a nationwide program and had provided free treatment for over 150 000 patients. The objective of this study was to report and evaluate the progress of China＇s free antiretroviral treatment program. Methods The NFATP Database was systematically reviewed and a total of 150 692 HIV/AIDS patients were included in this study. Program progress indicators including the number of treated HIV/AIDS patients, follow-up visit rate, CD4 test rate, and viral load test rate were summarized and examined over a calendar year to evaluate the progress of NFATP quantitatively and qualitatively. Results By the end of 2011, a total of 150 692 HIV/AIDS patients had been treated through the NFATP and 122 613 of them were still on treatment. Of all patients, about 72% were enrolled during the past four years. The dominant transmission route was blood related in the early phase of the NFATP, but gradually changed to sexual contact. Besides quantitative improvements, progress indicators also demonstrated significant qualitative improvements that the program had made during the past 9 years. Conclusions Great achievement has been made by China＇s NFATP. China＇s experience indicates the importance of a comprehensive response to the success of its treatment program. However, to ensure the quality and sustainability of treatment in the long term, more attention and resources should be paid towards program management.
基金supported by the National key Basic Research and Development(973)Program of China(No.2013CB329303)the National Natural Science Foundation of China(Nos.61502035,61132009,and 61671064)Beijing Advanced Innovation Center for Imaging Technology(No.BAICIT-2016007)
文摘Statistical machine translation for low-resource language suffers from the lack of abundant training corpora. Several methods, such as the use of a pivot language, have been proposed as a bridge to translate from one language to another. However, errors will accumulate during the extensive translation pipelines. In this paper, we propose an approach to low-resource language translation by exploiting the pronunciation correlations between languages. We find that the pronunciation features can improve both Chinese-Vietnamese and Vietnamese- Chinese translation qualities. Experimental results show that our proposed model yields effective improvements, and the translation performance （bilingual evaluation understudy score） is improved by a maximum value of 1.03.
