Severe Imbalance Research Articles

Orismilast in moderate-to-severe psoriasis: Efficacy and safety from a 16-week, randomized, double-blinded, placebo-controlled, dose-finding, and phase 2b trial (IASOS)

BackgroundOrismilast is a novel oral phosphodiesterase 4 (PDE4) B/D inhibitor being investigated as a potential treatment for moderate-to-severe psoriasis. ObjectiveTo evaluate efficacy and safety of orismilast modified-release formulation in moderate-to-severe psoriasis. MethodsThis multicenter, randomized (1:1:1:1 to 20, 30, 40 mg orismilast or placebo, twice daily), double-blinded, placebo-controlled, parallel-group, phase 2b, 16-week, dose-ranging study evaluated orismilast in adults with moderate-to-severe plaque psoriasis (NCT05190419). Efficacy endpoints were analyzed using multiple imputation. ResultsOf 202 randomized patients, baseline characteristics were balanced across arms, except greater severe disease proportions for orismilast versus placebo. Orismilast showed significant improvements in the primary endpoint, percentage change in Psoriasis Area and Severity Index (PASI), from baseline to week 16 (orismilast –52.6% to –63.7%; placebo, –17.3%; all P<0.001). Greater proportions receiving orismilast achieved PASI75 (39.5% to 49.0%; P<0.05) and PASI90 (22.0% to 28.3%; P<0.05 for 20 and 40 mg) versus placebo (PASI75, 16.5%; PASI90, 8.3%) at week 16. Safety findings were as expected with PDE4 inhibition; dose-dependent tolerability effects observed. LimitationsSmall sample size, disease severity imbalance between groups, limited duration and diversity in study population. ConclusionOrismilast demonstrated greater efficacy versus placebo and a safety profile in line with PDE4 inhibition.

An adaptive load balancing strategy for stateful join operator in skewed data stream environments

As one of the most computationally intensive operations in stream processing applications, join operation can cause severe load imbalance problem when dealing with skewed data. Most of the popular solutions focused on monitoring-based dynamic balancing strategies, making it difficult to quickly adapt to the changing frequency of data stream, and sometimes failing the balancing strategies that try to address the skewed load in the cluster. To address these issues, we propose to use the prediction results of a deep reinforcement learning model and adjust the grouping strategy in advance before the frequency change of data stream. It will enable the system to quickly adapt to data stream fluctuation, while managing the resources for effective resource utilization. The following contributions are made in this paper: 1) Explore the main factors that trigger the load skewness problem in distributed stream join systems and carefully model the load balancing problem at the application level. 2) Develop a Gated Recurrent Unit Sequence to Sequence model to predict key frequency distribution of streams, and propose a dynamic grouping algorithm and a feedback-based resource elasticity scaling algorithm to solve the load imbalance problem caused by hot keys in real time. 3) Design and implement an adaptive stream join system Aj-Stream based on the prediction model and the proposed algorithm on Apache Storm. 4) Evaluate the system performance through extensive experiments on a large scale real-world dataset and multiple synthetic datasets. The experimental results demonstrate that the Aj-Stream proposed in this paper exhibits stable throughput and latency performance with both static data streams of varying skewnesses and dynamic data streams. In comparison to existing stream-connected systems, Aj-Stream demonstrated a 22.1% increase in system throughput and a 45.5% decrease in system latency when dealing with frequently fluctuating data streams.

Enhancing the Classification of Imbalanced Datasets through the Utilization of the Provisional Clarifier-Alternator (cPAen)

Warmcomfort data is vital for enhancing heating and cooling system efficiency through ML (ML) models. Yet, these datasets often suffer from severe class imbalance due to subjective feedback. To tackle this issue, we introduce a data augmentation algorithm, Provisional Clarifier-Alternator (cPAen). In our research, we evaluated cPAen using real Warmcomfort data and found it outperformed SMOTE, ADASYN, and cWGAN-GP in terms of F1 scores. Notably, cPAen is over 5 times faster than cWGAN-GP while maintaining high test accuracy.

Imbalanced learning for wind turbine blade icing detection via spatio-temporal attention model with a self-adaptive weight loss function

Accurate blade icing detection of wind turbines is of great significance to avoid secondary damages or accidents. Recently, data-driven methods have attracted increasing attention due to the availability of supervisory control and data acquisition (SCADA) data with their low-cost and high volume. However, SCADA data usually present complex characteristics of high dimensionality and strong spatio-temporal correlations among different sensor variables and severe class imbalance. It is challenging to extract effective features for accurate and reliable icing detection from imbalanced SCADA data. To address these challenges, we propose an imbalanced learning method for blade icing detection. First, we propose a spatio-temporal attention model to extract and assign weights to temporal-spatial features based on their contributions to blade icing detection. Then, to address the class imbalance issue, we design a self-adaptive weight (SAW) loss function to divide SCADA data into multiple batches and then classify them during model training. Unlike conventional loss functions, our SAW loss can adaptively assign weights to each data category according to their numbers in the divided batches. The effectiveness of the proposed method is verified on real SCADA datasets. The results demonstrate our designed spatio-temporal attention and SAW loss achieved better classification results and superior self-adaptability.

A CNN-LSTM Ensemble Model for Predicting Protein-Protein Interaction Binding Sites.

Proteins commonly perform biological functions through protein-protein interactions (PPIs). The knowledge of PPI sites is imperative for the understanding of protein functions, disease mechanisms, and drug design. Traditional biological experimental methods for studying PPI sites still incur considerable drawbacks, including long experimental time and high labor costs. Therefore, many computational methods have been proposed for predicting PPI sites. However, achieving high prediction performance and overcoming severe data imbalance remain challenging issues. In this paper, we propose a new sequence-based deep learning model called CLPPIS (standing for CNN-LSTM ensemble based PPI Sites prediction). CLPPIS consists of CNN and LSTM components, which can capture spatial features and sequential features simultaneously. Further, it utilizes a novel feature group as input, which has 7 physicochemical, biophysical, and statistical properties. Besides, it adopts a batch-weighted loss function to reduce the interference of imbalance data. Our work suggests that the integration of protein spatial features and sequential features provides important information for PPI sites prediction. Evaluation on three public benchmark datasets shows that our CLPPIS model significantly outperforms existing state-of-the-art methods.

Nereus: A Distributed Stream Band Join System With Adaptive Range Partitioning

Many real-time applications in consumer electronics rely on stream band join as a fundamental operation. With two streams, the band join operation targets at obtaining the pairs of tuples which are separately included in the two streams and have close values within a user specified range. Range partitioning keeps tuples with close values in a partition. For band join, the cost of employing range partitioning is less join cost than that of employing other partitioning strategies. However, the distribution of real-world data from consumer electronics over the range is skewed, causing severe load imbalance among instances in the distributed system that employs existing static range partitioning. Load migration can alleviate the load imbalance. For range partitioning, the migration has two kinds of objectives. Migrating load to the instance with adjacent partition controls the number of partitions. However, it causes an unacceptable high cost for migrating span multiple instances. While directly migrating a split partition to the lightest instance is low cost. However, it leads to an uncontrollable number of partitions. The system for consumer electronic applications cannot tolerate high migration cost and an uncontrollable number of partitions, which result in high latency and low throughput. In this work, we propose an adaptive range partitioning strategy to ensure a controllable number of partitions and load balancing with low cost. We implement Nereus, a distributed stream band join system. Nereus designs a migration benefit model using queuing theory measure, which integrates the benefits of partition’s change and load balancing. Such a design can obtain the most beneficial migration, which achieves low migration cost and an appropriate number of partitions. We conduct comprehensive experiments using large-scale datasets from real-world applications to evaluate this design. The results show that Nereus improves the throughput by 51% and reduces the processing latency by 99%, compared to existing designs.

Pedicle Subtraction Osteotomy on a Patient with Severe Sagittal Imbalance Progression But Proportionate Gap Score Case Report and Lliterature Review

To demonstrate the successful surgical management of a patient with Degenerative Disc Disease (DDD) who underwent multiple surgeries but continued to suffer from severe sagittal imbalance despite a proportioned GAP Score.

Survivors' Experiences of Hyperemesis Gravidarum.

Hyperemesis gravidarum can result in life-threatening physical and psychological maternal morbidity, including severe dehydration, weight loss, electrolyte imbalance, depression, and suicidal ideation. The reported prevalence of hyperemesis gravidarum ranges from 0.3% to 3.6%. The purpose of this qualitative study was to investigate what blogs can tell us about women's experiences of hyperemesis gravidarum. Thirty-three blogs written by hyperemesis gravidarum survivors posted on Hyperemesis Australia's website were analyzed using Krippendorff's qualitative content analysis method. Clustering was used and yielded 6 themes: (1) debilitating physical and mental health problems: digging deep to persevere, (2) heartbreaking choices, (3) lack of understanding and dismissed, (4) so much guilt surrounding their unborn infant, (5) it takes a village to support women with hyperemesis gravidarum, and (6) warriors and survivors: giving back. Infusion nurses are in a perfect position to provide support and compassionate care for women who are repeatedly coming to the hospital for rehydration treatment. Infusion nurses can validate women's physical and emotional struggles with hyperemesis gravidarum and help to no longer make women feel stigmatized.

α-Hederin induces human colorectal cancer cells apoptosis through disturbing protein homeostasis

Protein homeostasis and quality control are crucial for normal cellular activities, and a severe imbalance in protein homeostasis can lead to cell death. α-Hederin, a pentacyclic triterpenoid saponin isolated from Fructus Akebia, has a clear role in promoting colorectal cancer (CRC) cell apoptosis and has been recently used for CRC therapy. However, whether the pro-apoptotic activity of α-hederin in CRC cells involves disturbing protein homeostasis remains unknown. Here, we aimed to uncover the underlying molecular mechanism of α-hederin-induced apoptosis in CRC cells. Cell viability and proliferation were examined by 3-(4,5-dimethylthiazol-2-yl)-2,5 diphenyl tetrazolium bromide (MTT) and 5-ethynyl-2′-deoxyuridine (EdU) assays, respectively. Apoptosis was detected using flow cytometry and western blotting. Autophagic flux was examined by western blotting and AdPlus-mCherry-GFP-LC3B adenovirus infection assays, and western blotting and immunofluorescence staining were performed to detect the expression of proteins in related pathways. The results showed that α-hederin significantly inhibited the growth and promoted the apoptosis of human CRC cells. Furthermore, α-Hederin induced endoplasmic reticulum (ER) stress, but inhibited proteasomal degradation. Also, the autophagic flux was blocked by α-hederin although this drug promoted autophagosome formation, and the lysosomal degradation was inhibited. Expression of p-JNK and p-p38 were increased by α-hederin. So, our findings provide strong evidence that α-hederin disrupts protein homeostasis by blocking ER-associated degradation (ERAD) and autophagic flux, thereby contributing to apoptosis. PERK-eIF2α-ATF4-CHOP and IRE1-ASK1-JNK/p38 signal pathway may be involved in those regulation. Our results make it a promising alternative or adjunct therapeutic candidate for CRC.

Improving the Performance of Object Detection by Preserving Balanced Class Distribution

Object detection is a task that performs position identification and label classification of objects in images or videos. The information obtained through this process plays an essential role in various tasks in the field of computer vision. In object detection, the data utilized for training and validation typically originate from public datasets that are well-balanced in terms of the number of objects ascribed to each class in an image. However, in real-world scenarios, handling datasets with much greater class imbalance, i.e., very different numbers of objects for each class, is much more common, and this imbalance may reduce the performance of object detection when predicting unseen test images. In our study, thus, we propose a method that evenly distributes the classes in an image for training and validation, solving the class imbalance problem in object detection. Our proposed method aims to maintain a uniform class distribution through multi-label stratification. We tested our proposed method not only on public datasets that typically exhibit balanced class distribution but also on private datasets that may have imbalanced class distribution. We found that our proposed method was more effective on datasets containing severe imbalance and less data. Our findings indicate that the proposed method can be effectively used on datasets with substantially imbalanced class distribution.

MF2POSE: Multi-task Feature Fusion Pseudo-Siamese Network for intrusion detection using Category-distance Promotion Loss

Intrusion detection is a crucial aspect of modern cybersecurity, aimed at identifying and responding to potential security threats within computer systems, networks, and applications. One of the major challenges faced by intrusion detection systems is the accurate detection and response to attack traffic, which is typically much lower in volume compared to normal traffic. This challenge becomes even more pronounced when the attack traffic is further classified into different attack categories, which may suffer from more severe data imbalance. To address these challenges, we present a novel approach to intrusion detection using a Multi-task Feature Fusion Pseudo-Siamese Network (MF2POSE) and a Category-distance Promotion Loss (CP Loss). The proposed MF2POSE leverages the Pseudo-Siamese Network (POSE-Net), which incorporates both a main network and a siamese network, to simultaneously perform binary-class and multi-class classification of the traffic. Furthermore, the Multi-task Feature Fusion (MF2) module enhances the multi-class classification performance of the main network by incorporating multi-scale internal features from the siamese network. Additionally, the CP Loss is introduced to aggregate the internal feature from the main network belonging to the same category, and separate the internal feature belonging to the different categories. We have conducted comprehensive experiments across two popular intrusion detection datasets, and the experimental results demonstrate the superior performance of our MF2POSE compared to existing state-of-the-art techniques, particularly in multi-class classification scenarios.

Toxicity assessment of microcystin-leucine arginine in planarian Dugesia japonica.

Microcystin-leucine arginine (MC-LR), a representative cyanobacterial toxin, poses an increasing and serious threat to aquatic ecosystems. Despite investigating its toxic effects in various organisms and cells, the toxicity to tissue regeneration and stem cells in vivo still needs to be explored. Planarians are ideal regeneration and toxicology research models and have profound implications in ecotoxicology evaluation. This study conducted a systemic toxicity evaluation of MC-LR, including morphological changes, growth, regeneration, and the underlying cellular and molecular changes after MC-LR exposure, which were investigated in planarians. The results showed that exposure to MC-LR led to time- and dose-dependent lethal morphological changes, tissue damage, degrowth, and delayed regeneration in planarians. Furthermore, MC-LR exposure disturbed the activities of antioxidants, including total superoxide dismutase, catalase, glutathione peroxidase, glutathione S-transferase, and total antioxidant capacity, leading to oxidative stress and DNA damage, and then reduced the number of dividing neoblasts and promoted apoptosis. The results demonstrated that oxidative stress and DNA damage induced by MC-LR exposure caused apoptosis. Excessive apoptosis and suppressed neoblast activity led to severe homeostasis imbalance. This study explores the underlying mechanism of MC-LR toxicity in planarians and provides a basis for the toxicity assessment of MC-LR to aquatic organisms and ecological risk evaluation.

Lending to Priority Sector in India: Some issues of Credit and NPAs of Public and Private Sector Banks

Priority sector lending has been an essentialfeature of India’s credit policy.Priority sector lending has been in use for long, as an instrument to channelize credit to strategic sectors of the economy that have been marginalized by institutionalised credit delivery system. This programme ensures that the necessary bank credit be allocated forstrategically importantsectors of the economy.Priority sector lending was imbibed in the lending policy for the commercial banks in response to shortages of output in agricultural and industrieswhich caused severe imbalances in the economy after the independence and partisan of the country. Hence, the priority sector lending was used as a lively device for directing the credit flow to the neglected entities with a purpose to re-establish sectoral equilibrium. The program requires commercial banks to extend loans to strategic sectors: agriculture, exports, small business, housing, and economically weak sections of the society. Priority sector lending is still relevantfor economies like India as small scale entities have been backbone of our economy being employing the large section of population of the country. The present study is an attempt to analyse the status of credit allocation and NPAs of priority sector lending of public sector banks and private sector banks considering a period of 15 years from 2004-2005 to 2019-2020.

Mapping and assessing supply and demand of crop pollination services in Shandong Province, China

Urban expansion and large-scale expansion of intensive agriculture have resulted in loss and fragmentation of pollinator habitats, posing a significant threat to the pollination services they provide. Owing to rapid changes in land use and agricultural planting structure, it is critical to understand the spatio-temporal dynamics of supply, demand, and supply-demand balances of pollination services to ensure that pollination capacity meets the requirements of crop production. This study was conducted in an agroecosystem situated in the Shandong province, eastern China, the largest agriculture output value province in China, and the spatio-temporal dynamics of the pollination services that can support the agricultural demand from 1990 to 2018 were assessed. We quantified and mapped pollination demand, pollination capacity, and supply-demand (mis)matches of pollination during 1990–2018. The pollination results indicated that supply, demand, and supply-demand matches of pollination services decreased during the last three decades, and that the entire province of Shandong was on the verge of pollination imbalances in 2018. Moreover, in 28.99% of the province, there is a severe imbalance between pollination supply and demand, resulting in unmet needs and posing a threat to crop production. Regarding spatial patterns, supply, demand, and supply-demand matches of pollination services were heterogeneous in the province, and individually targeted conservation in different regions should be devised by the local government. We provided a spatially-explicit framework to identify spatial patterns linking pollination capacity and crop demand and optimize pollination-related management to ensure agricultural sustainability and crop production.

Dynamic Evolution and Trend Analysis of the Coupling Coordination Between Sci-Tech Popularization and Sci-Tech Innovation in China

Sci-Tech Popularization (STP) and Sci-Tech Innovation (STI) are two significant components to realize China’s innovation development. STI provides direction and content for STP, while STP creates an intense atmosphere for STI. By establishing two evaluation index systems of STP and STI, the STP index and the STI index of China and Chinese 31 provinces from 2010 to 2019 was calculated, and the STP index and the STI index from 2020 to 2024 were predicted. Finally, the relative development degree (RDD), the coupling coordination degree (CCD) and the dynamic degree of STP and STI were figured out. The results showed that: (1) From 2010 to 2024, the STP index and the STI index of China and Chinese 31 provinces showed a growing trend. The growth rate of STI was higher than that of STP, but the STP and the STI were low. (2) The RDD results showed that the development of STP lagged behind that of STI after 2018, and the gap between them showed a growing trend. (3) The CCD of STP and STI in China spanned three grades: moderate imbalance, slight imbalance, and approaching imbalance. The CCD in Chinese each province spanned five grades: severe imbalance, moderate imbalance, slight imbalance, approaching imbalance, reluctant coordination. The CCD was developing in a good direction, China and most Chinese provinces were still in an unbalanced level instead. Finally, suggestions on infrastructure, technical means, personnel, and social responsibility were put forward to promote the coordinated development between STP and STI in China.

Curriculum label distribution learning for imbalanced medical image segmentation.

Label distribution learning (LDL) has the potential to resolve boundary ambiguity in semantic segmentation tasks. However, existing LDL-based segmentation methods suffer from severe label distribution imbalance: the ambiguous label distributions contain a small fraction of the data, while the unambiguous label distributions occupy the majority of the data. The imbalanced label distributions induce model-biased distribution learning and make it challenging to accurately predict ambiguous pixels. In this paper, we propose a curriculum label distribution learning (CLDL) framework to address the above data imbalance problem by performing a novel task-oriented curriculum learning strategy. Firstly, the region label distribution learning (R-LDL) is proposed to construct more balanced label distributions and improves the imbalanced model learning. Secondly, a novel learning curriculum (TCL) is proposed to enable easy-to-hard learning in LDL-based segmentation by decomposing the segmentation task into multiple label distribution estimation tasks. Thirdly, the prior perceiving module (PPM) is proposed to effectively connect easy and hard learning stages based on the priors generated from easier stages. Benefiting from the balanced label distribution construction and prior perception, the proposed CLDL effectively conducts a curriculum learning-based LDL and significantly improves the imbalanced learning. We evaluated the proposed CLDL using the publicly available BRATS2018 and MM-WHS2017 datasets. The experimental results demonstrate that our method significantly improves different segmentation metrics compared to many state-of-the-art methods. The code will be available.1.

Machine Learning-Based Intrusion Detection for Rare-Class Network Attacks

Due to the severe imbalance in the quantities of normal samples and attack samples, as well as among different types of attack samples, intrusion detection systems suffer from low detection rates for rare-class attack data. In this paper, we propose a geometric synthetic minority oversampling technique based on the optimized kernel density estimation algorithm. This method can generate diverse rare-class attack data by learning the distribution of rare-class attack data while maintaining similarity with the original sample features. Meanwhile, the balanced data is input to a feature extraction module built upon multiple denoising autoencoders, reducing information redundancy in high-dimensional data and improving the detection performance for unknown attacks. Subsequently, a soft-voting ensemble learning technique is utilized for multi-class anomaly detection on the balanced and dimensionally reduced data. Finally, an intrusion detection system is constructed based on data preprocessing, imbalance handling, feature extraction, and anomaly detection modules. The performance of the system was evaluated using two datasets, NSL-KDD and N-BaIoT, achieving 86.39% and 99.94% multiclassification accuracy, respectively. Through ablation experiments and comparison with the baseline model, it is found that the inherent limitations of a single machine-learning model directly affect the accuracy of the intrusion detection system, while the superiority of the proposed multi-module model in detecting unknown attacks and rare classes of attack traffic is demonstrated.

A visual measurement algorithm for vibration displacement of rotating body using semantic segmentation network

Compared with the traditional vibration displacement measurement methods, visual vibration measurement offers several advantages such as long-distance capability, non-contact operation and easy installation. However, the phenomenon of low fitting accuracy of the bounding box often occurs when detecting rotating objects, resulting in a slight deviation in the relative offset of the center vibration point of the target between frames, which will cause a serious deviation in the regression of vibration displacement offset. In this paper, a high-speed industrial camera is employed as the image acquisition medium, and a deep learning-based semantic segmentation method is introduced to address visual vibration measurement challenges in rotating body. Specifically, the CSP module integrates different depth semantic information which is introduced into the Mobiledets backbone network in a targeted manner. This is not only strengthens the performance of the network for segmenting vibration objects, but also dramatically improves the practical performance of the algorithm. The conventional Relu activation function is substituted with Mish activation function, making the network more adept at segmenting rotating body in challenging backgrounds with varying illumination, blur, and similarity. The CSP+Mobiledets backbone network constructed in this study outperforms the U-Net network in terms of feature extraction effectiveness. Adding Dice-loss to the original loss function can more effectively solve the severe imbalance problem of samples caused by long-distance image acquisition. We take the most representative rotating body-rotor as the experimental subject. The displacement curve obtained by the existing algorithm has the best degree of fit with the signal curve collected by the eddy sensor. The results of different segmentation algorithms and detection algorithms on time domain curve plot, frequency domain plot and axis orbit plot are collectively compared. Furthermore, the results also provide valuable guidance for visual measurement of the vibration displacement of the rotating body in specific industrial scenarios.

Empirical Study and Reflection on the Impact of COVID-19 Pandemic on Chinese A-share Market--Based on Fama French Five-factor Model

The COVID-19 pandemic has caused severe imbalances in the global economy and intensified stock market turmoil worldwide. Based on Fama-French five-factor model, this paper studies Chinas A-share market changes before and after the epidemic. The stocks treated by ST (special treatment) and * ST (delisting warning) are screened, excluding companies and banking stocks with shortlisting time. The equities are finally divided into groups using the two-by-three grouping approach. According to the median of the total market value of stocks, the entire sample is separated into two categories: modest market value (S) and large market value (B). The samples are classified into three categories based on the book-to-market ratio's 30% and 70% quantiles: high (H), medium (N), and low (L). Comparing the different times of the impact of the five factors on stock earnings is conducive to explaining the changes in the factors affecting the stock price of listed companies before and after the epidemic. It is possible to analyze which factors have declined and fluctuated after the epidemic, resulting in a decline in stock prices, a slowdown in the growth of listed companies, or even a recession. The analysis of the five-factor model can clarify which factors have led to the decline in stock prices after the beginning of the epidemic, which is conducive to the formulation of related policies and plans by listed companies and makes up for the decline in stock prices and the slowdown in company growth caused by the corresponding decline, which has apparent practical value.

Geoscience academic hiring networks reinforce historic patterns of inequity

Abstract An analysis of the academic hiring networks in geoscience reveals a severe imbalance that favors graduates from a small handful of institutions. In this study, social network analysis was conducted on a database consisting of every individual with a Ph.D. working in a geoscience degree-granting program in the United States (n = 6694) between 2015 and 2021. Individuals were mapped from the institution where they earned their Ph.D. to the institution where they currently work. Of the 895 geoscience degree-granting institutions included in the database, 10 alone produced nearly a quarter (24.6%) of the entire academic geoscience workforce. Network analysis also identified a small, closed network consisting of five of the top-10 institutions, which suggests that these networks hire more frequently from one another than from other institutions in the network. When academic rank was used to analyze the network for change over time, no significant shift in the hiring patterns was found. These imbalances in faculty production disadvantage scientists who are educated at programs other than the top-placing institutions and ultimately reinforces longstanding inequities in the field, such as the underrepresentation of people who are Black, Indigenous, People of Color (BIPOC), and first-generation college students in geoscience faculty. These patterns of inequity have also been shown to limit the spread of new scientific ideas throughout research communities.