Search

We study the relationship between the performance of firms and their technological portfolios using tools borrowed from complexity science. In particular, we ask whether the accumulation of knowledge and capabilities associated with a coherent set of technologies leads firms to experience advantages in terms of productive efficiency. To this end, we analyze both the balance sheets and the patenting activity of about 70 thousand firms that have filed at least one patent over the period 2004-2013. We define a measure of corporate coherent diversification, based on the bipartite network linking companies with the technological fields in which they patent, and relate it to firm performance in terms of labor productivity. Our measure favors technological portfolios that can be decomposed into large blocks of closely related fields over portfolios with the same breadth of scope, but a more scattered diversification structure. We find that the coherent diversification of firms is quantitatively related with their economic performance and captures relevant information about their productive structure. In particular, we prove on a statistical basis that a naive definition of technological diversification can explain labor productivity only as a proxy of size and coherent diversification. This approach can be used to investigate possible synergies within firms and to recommend viable partners for mergers and acquisitions. [ABSTRACT FROM AUTHOR]

Stock price prediction is a popular yet challenging task and deep learning provides the means to conduct the mining for the different patterns that trigger its dynamic movement. In this paper, the task is to predict the close price for 25 companies enlisted at the Bucharest Stock Exchange, from a novel data set introduced herein. Towards this scope, two traditional deep learning architectures are designed in comparison: a long short-memory network and a temporal convolutional neural model. Based on their predictions, a trading strategy, whose decision to buy or sell depends on two different thresholds, is proposed. A hill climbing approach selects the optimal values for these parameters. The prediction of the two deep learning representatives used in the subsequent trading strategy leads to distinct facets of gain. [ABSTRACT FROM AUTHOR]

Objectives: Hepatitis C virus (HCV) causes both acute and chronic infection, which can potentially develop into cirrhosis and liver cancer. Healthcare systems are struggling to finance costly direct-acting antiviral agents through public funding for uninsured patients, despite the unprecedented high cure rates of these agents. Vulnerable populations are at higher risk of HCV infection. The personal importation scheme is based on the legal right to import any unauthorized generics for personal use. This study was designed to assess the knowledge and perceptions of stakeholders on unauthorized generics. Methods: We conducted an anonymous online survey based on the fictitious situation of a patient diagnosed with HCV who lacked mandatory health insurance and personal financial resources. Results: We obtained a sample of 781 respondents: 445 physicians, 77 pharmacists, 51 patients and 207 non-healthcare professionals. We found that only 36% and 58% of respondents believe that the quality and efficacy, respectively, of unauthorized generics are equivalent to their corresponding brand. An overwhelming majority (98%) favoured quality control upon arrival, and 31% felt they could recognize fraudulent websites. A total of 79% expressed support for financial assistance for vulnerable patients, and support among physicians was 83%. Conclusions: Overall, the limited knowledge of the efficacy and quality of unauthorized generics, despite evidence in peer-reviewed literature, contrasts with the overwhelmingly positive attitudes toward financial assistance for personal import. This finding emphasizes the need for clearer information on imported generics and the potential safety provided by buyers' club schemes to complete the WHO agenda of eradicating viral hepatitis by 2030 within otherwise excluded vulnerable populations. [ABSTRACT FROM AUTHOR]

Many real systems can be described by a set of interacting entities forming a complex network. To some surprise, these have been shown to share a number of structural properties regardless of their type or origin. It is thus of vital importance to design simple and intuitive models that can explain their intrinsic structure and dynamics. These can, for instance, be used to study networks analytically or to construct networks not observed in real life. Most models proposed in the literature are of two types. A model can be either static, where edges are added between a fixed set of nodes according to some predefined rule, or evolving, where the number of nodes or edges increases over time. However, some real networks do not grow but rather shrink, meaning that the number of nodes or edges decreases over time. We here propose a simple model of shrinking networks called the war pact model. We show that networks generated in such a way exhibit common structural properties of real networks. Furthermore, compared to classical models, these resemble international trade, correlates of war, Bitcoin transactions and other networks more closely. Network shrinking may therefore represent a reasonable explanation of the evolution of some networks and greater emphasis should be put on such models in the future. [ABSTRACT FROM AUTHOR]

The diversity of bats worldwide includes large numbers of cryptic species, partly because divergence in acoustic traits such as echolocation calls are under stronger selection than differences in visual appearance in these nocturnal mammals. Island faunas often contain disproportionate numbers of endemic species, and hence we might expect cryptic, endemic species to be discovered relatively frequently in bats inhabiting islands. Species are best defined when multiple lines of evidence supports their diagnosis. Here we use morphometric, acoustic, and molecular phylogenetic data to show that a horseshoe bat in the Andaman Islands is distinct in all three aspects, supporting its status as a distinct species. We recommend investigation into possible new and endemic bat species on islands by using integrated approaches that provide independent lines of evidence for taxonomic distinctiveness. We provide a formal redescription of the taxon newly raised to species level, Rhinolophus andamanensis Dobson, 1872. [ABSTRACT FROM AUTHOR]

Background: Rare diseases (RD) result in a wide variety of clinical presentations, and this creates a significant diagnostic challenge for health care professionals. We hypothesized that there exist a set of consistent and shared phenomena among all individuals affected by (different) RD during the time before diagnosis is established. Objective: We aimed to identify commonalities between different RD and developed a machine learning diagnostic support tool for RD. Methods: 20 interviews with affected individuals with different RD, focusing on the time period before their diagnosis, were performed and qualitatively analyzed. Out of these pre-diagnostic experiences, we distilled key phenomena and created a questionnaire which was then distributed among individuals with the established diagnosis of i.) RD, ii.) other common non-rare diseases (NRO) iii.) common chronic diseases (CD), iv.), or psychosomatic/somatoform disorders (PSY). Finally, four combined single machine learning methods and a fusion algorithm were used to distinguish the different answer patterns of the questionnaires. Results: The questionnaire contained 53 questions. A total sum of 1763 questionnaires (758 RD, 149 CD, 48 PSY, 200 NRO, 34 healthy individuals and 574 not evaluable questionnaires) were collected. Based on 3 independent data sets the 10-fold stratified cross-validation method for the answer-pattern recognition resulted in sensitivity values of 88.9% to detect the answer pattern of a RD, 86.6% for NRO, 87.7% for CD and 84.2% for PSY. Conclusion: Despite the great diversity in presentation and pathogenesis of each RD, patients with RD share surprisingly similar pre-diagnosis experiences. Our questionnaire and data-mining based approach successfully detected unique patterns in groups of individuals affected by a broad range of different rare diseases. Therefore, these results indicate distinct patterns that may be used for diagnostic support in RD. [ABSTRACT FROM AUTHOR]

Introduction: Endometriosis has a significant cost of illness burden in Europe, UK and the USA, with the majority of costs coming from reductions in productivity. However, information is scarce on if there is a differing impact between endometriosis and other causes of chronic pelvic pain, and if there are modifiable factors, such as pain severity, that may be significant contributors to the overall burden. Methods: An online survey was hosted by SurveyMonkey and the link was active between February to April 2017. Women aged 18–45, currently living in Australia, who had either a confirmed diagnosis of endometriosis via laparoscopy or chronic pelvic pain without a diagnosis of endometriosis were included. The retrospective component of the WERF EndoCost tool was used to determine direct healthcare costs, direct non-healthcare costs (carers) and indirect costs due to productivity loss. Estimates were extrapolated to the Australian population using published prevalence estimates. Results: 407 valid responses were received. The cost of illness burden was significant in women with chronic pelvic pain (Int $16,970 to $ 20,898 per woman per year) irrespective of whether they had a diagnosis of endometriosis. The majority of costs (75–84%) were due to productivity loss. Both absolute and relative productivity costs in Australia were higher than previous estimates based on data from Europe, UK and USA. Pain scores showed the strongest relationship to productivity costs, a 12.5-fold increase in costs between minimal to severe pain. The total economic burden per year in Australia in the reproductive aged population (at 10% prevalence) was 6.50 billion Int $. Conclusion: Similar to studies in European, British and American populations, productivity costs are the greatest contributor to overall costs. Given pain is the most significant contributor, priority should be given to improving pain control in women with pelvic pain [ABSTRACT FROM AUTHOR]

The problem of handling non-ignorable non-response has been typically addressed under the design-based approach using the well-known sub-sampling technique introduced by Hansen and Hurwitz [1946, Journal of the American Statistical Association, Vol 41(236), Page 517- 529]. Alternatively, the model-based paradigm emphasizes on utilizing the underlying model relationship between the outcome variable and one or more covariate(s) whose population values are known prior to the survey. This article utilizes the model relationship between the study variable and covariate(s) for handling non-ignorable non-response and obtaining an unbiased estimator for the population total under the sub-sampling technique. The main idea is to combine the estimates obtained from the sample on first call and the sub-sample from second call using separate model relationships. The contribution of this paper helps us in providing unbiased estimates with an improved efficiency under model-based paradigm in presence of non-ignorable non-response. The provided method is more economical than the available estimators under callback methods as we are working sub-sampling and also increase response rate as a stronger mode of interview is employed for data collection. A numerical study using Monte Carlo is presented to illustrate the behavior of the proposed and the efficiency comparison. [ABSTRACT FROM AUTHOR]

Slope one is a popular recommendation algorithm due to its simplicity and high efficiency for sparse data. However, it often suffers from under-fitting since the global information of all relevant users/items are considered. In this paper, we propose a new scheme called enhanced slope one recommendation through local information embedding. First, we employ clustering algorithms to obtain the user clusters as well as item clusters to represent local information. Second, we predict ratings using the local information of users and items in the same cluster. The local information can detect strong localized associations shared within clusters. Third, we design different fusion approaches based on the local information embedding. In this way, both under-fitting and over-fitting problems are alleviated. Experiment results on the real datasets show that our approaches defeats slope one in terms of both mean absolute error and root mean square error. [ABSTRACT FROM AUTHOR]

In recent years, with the development of computer science, deep learning is held as competent enough to solve the problem of inference and learning in high dimensional space. Therefore, it has received unprecedented attention from both the academia and the business community. Compared with CPU/GPU, FPGA has attracted much attention for its high-energy efficiency, short development cycle and reconfigurability in the aspect of deep learning algorithm. However, because of the limited research on OpenCL optimization on FPGA of deep learning algorithms, OpenCL tools and models applied to CPU/GPU cannot be directly used on FPGA. This makes it difficult for software programmers to use FPGA when implementing deep learning algorithms for a rewarding performance. To solve this problem, this paper proposed an OpenCL computational model based on FPGA template architecture to optimize the time-consuming convolution layer in deep learning. The comparison between the program applying the computational model and the corresponding optimization program provided by Xilinx indicates that the former is 8-40 times higher than the latter in terms of performance. [ABSTRACT FROM AUTHOR]

Background: Many countries have initiated school-based human papillomavirus (HPV) vaccination programs. The real-life effectiveness of HPV vaccines has become increasingly evident, especially among girls vaccinated before HPV exposure in countries with high vaccine uptake. In 2009, Norway initiated a school-based HPV vaccination program for 12-year-old girls using the quadrivalent HPV vaccine (Gardasil®), which targets HPV6, 11, 16, and 18. Here, we aim to assess type-specific vaginal and oral HPV prevalence in vaccinated compared with unvaccinated girls in the first birth cohort eligible for school-based vaccination (born in 1997). Methods: This observational, cross-sectional study measured the HPV prevalence ratio (PR) between vaccinated and unvaccinated girls in Norway. Facebook advertisement was used to recruit participants and disseminate information about the study. Participants self-sampled vaginal and oral specimens using an Evalyn® Brush and a FLOQSwab™, respectively. Sexual behavior was ascertained through a short questionnaire. Results: Among the 312 participants, 239 (76.6%) had received at least one dose of HPV vaccine prior to sexual debut. 39.1% of vaginal samples were positive for any HPV type, with similar prevalence among vaccinated and unvaccinated girls (38.5% vs 41.1%, PR: 0.93, 95% confidence interval [CI]: 0.62–1.41). For vaccine-targeted types there was some evidence of lower prevalence in the vaccinated (0.4%) compared to the unvaccinated (6.8%) group (PR: 0.06, 95%CI: 0.01–0.52). This difference remained after adjusting for sexual behavior (PR: 0.04, 95%CI: 0.00–0.42). Only four oral samples were positive for any HPV type, and all of these participants had received at least one dose of HPV vaccine at least 1 year before oral sexual debut. Conclusion: There is evidence of a lower prevalence of vaccine-targeted HPV types in the vagina of vaccinated girls from the first birth cohort eligible for school-based HPV vaccination in Norway; this was not the case when considering all HPV types or types not included in the quadrivalent HPV vaccine. [ABSTRACT FROM AUTHOR]

We use agent-based computer simulation to test the effect of environmental conditions (available biomass/carrying capacity and environmental risk) on the development of wealth inequality and patron-client herding networks in nomadic pastoral economies. Our results show that 1) wealth inequality reaches very high levels when carrying capacity is high and risk is low, and 2) patron-client contract herding networks increase in size and duration when carrying capacity is high and risk is low. We compare empirical data from the Mongol (1206–1368 CE) and Xiongnu (209 BCE– 48 CE) empires with simulation results to develop an explanatory mechanism for the apparent correlation between nomadic empire creation and positive environmental conditions. We argue that the internal dynamics of nomadic pastoral societies are sufficient to produce high degrees of inequality and hierarchical herding networks. Nomadic empires are more likely to form during key periods of increased biomass and decreased environmental risk. [ABSTRACT FROM AUTHOR]

The question of how self-driving cars should behave in dilemma situations has recently attracted a lot of attention in science, media and society. A growing number of publications amass insight into the factors underlying the choices we make in such situations, often using forced-choice paradigms closely linked to the trolley dilemma. The methodology used to address these questions, however, varies widely between studies, ranging from fully immersive virtual reality settings to completely text-based surveys. In this paper we compare virtual reality and text-based assessments, analyzing the effect that different factors in the methodology have on decisions and emotional response of participants. We present two studies, comparing a total of six different conditions varying across three dimensions: The level of abstraction, the use of virtual reality, and time-constraints. Our results show that the moral decisions made in this context are not strongly influenced by the assessment, and the compared methods ultimately appear to measure very similar constructs. Furthermore, we add to the pool of evidence on the underlying factors of moral judgment in traffic dilemmas, both in terms of general preferences, i.e., features of the particular situation and potential victims, as well as in terms of individual differences between participants, such as their age and gender. [ABSTRACT FROM AUTHOR]

We aim to determine whether a game-theoretic model between an insurer and a healthcare practice yields a predictive equilibrium that incentivizes either player to deviate from a fee-for-service to capitation payment system. Using United States data from various primary care surveys, we find that non-extreme equilibria (i.e., shares of patients, or shares of patient visits, seen under a fee-for-service payment system) can be derived from a Stackelberg game if insurers award a non-linear bonus to practices based on performance. Overall, both insurers and practices can be incentivized to embrace capitation payments somewhat, but potentially at the expense of practice performance. [ABSTRACT FROM AUTHOR]

Existing algorithms of speech-based deception detection are severely restricted by the lack of sufficient number of labelled data. However, a large amount of easily available unlabelled data has not been utilized in reality. To solve this problem, this paper proposes a semi-supervised additive noise autoencoder model for deception detection. This model updates and optimizes the semi-supervised autoencoder and it consists of two layers of encoder and decoder, and a classifier. Firstly, it changes the activation function of the hidden layer in network according to the characteristics of the deception speech. Secondly, in order to prevent over-fitting during training, the specific ratio dropout is added at each layer cautiously. Finally, we directly connected the supervised classification task in the output of encoder to make the network more concise and efficient. Using the feature set specified by the INTERSPEECH 2009 Emotion Challenge, the experimental results on Columbia-SRI-Colorado (CSC) corpus and our own deception corpus show that the proposed model can achieve more advanced performance than other alternative methods with only a small amount of labelled data. [ABSTRACT FROM AUTHOR]

To understand the effect of attention on neuronal dynamics, we propose a multi-module network, with each module consisting of fully interconnected groups of excitatory and inhibitory neurons. This network shows transitive dynamics among quasi-attractors as its typical dynamics. When the release of acetylcholine onto the network is simulated by attention, the transitive dynamics change into stable dynamics in which the system converges to an attractor. We found that this network can reproduce three experimentally observed properties of attention-dependent response modulation, namely an increase in the firing rate, a decrease in the Fano factor of the firing rate, and a decrease in the correlation coefficients between the firing rates of pairs of neurons. Moreover, we also showed theoretically that the release of acetylcholine increases the sensitivity to bottom-up inputs by changing the response function. [ABSTRACT FROM AUTHOR]

Electroencephalographic (EEG) activity in the gamma (30–80 Hz) range is related to a variety of sensory and cognitive processes which are frequently impaired in schizophrenia. Auditory steady-state response at 40-Hz (40-Hz ASSR) is utilized as an index of gamma activity and is proposed as a biomarker of schizophrenia. Nevertheless, the link between ASSRs and cognitive functions is not clear. This study explores a possible relationship between the performance on cognitive tasks and the 40-Hz ASSRs in a controlled uniform sample of young healthy males, as age and sex may have complex influence on ASSRs. Twenty-eight young healthy male volunteers participated (mean age ± SD 25.8±3.3) in the study. The 40-Hz click trains (500 ms) were presented 150 times with an inter-stimulus interval set at 700–1000 ms. The phase-locking index (PLI) and event-related power perturbation (ERSP) of the ASSR were calculated in the 200–500 ms latency range, which corresponds to the steady part of the response. The Psychology Experiment Building Language (PEBL) task battery was used to assess five cognitive subdomains: the Choice response time task, the Stroop test, the Tower of London test, the Lexical decision task and the Semantic categorisation task. Pearson‘s correlation coefficients were calculated to access the relationships; no multiple-test correction was applied as the tests were explorative in nature. A significant positive correlation was observed for the late-latency gamma and the mean number of steps in the Tower of London task reflecting planning and problem-solving abilities. These findings support the concept that 40-Hz ASSR might highlight top-down mechanisms which are related to cognitive functioning. Therefore, 40-Hz ASSRs can be used to explore the relationship between cognitive functioning and neurophysiological indices of brain activity. [ABSTRACT FROM AUTHOR]

Objective: To explore, through face to face interviews with a selection of General Dental Practitioners (GDPs), their views and experiences of having used the International Caries Detection and Assessment System (ICDAS) within primary care research studies for recording caries. Methods: This qualitative study involved one on one interviews with eight GDP’s who had previously used ICDAS on patients in their dental practices as part of a research study. The participants were selected from among those who had taken part in two clinical studies in the UK using convenient, but purposive sampling. The interviews were tape-recorded and transcribed; the data analysis was conducted by thematic analysis. Results: GDP’s indicated their beliefs that ICDAS had an important role in caries prevention but reported four main barriers while using the full (6 caries stages) ICDAS coding system in their practices: lack of simplicity of coding, financial implications and time consumption (in both training and use of ICDAS) and inadequate undergraduate training. An overarching theme identified from the GDPS was the willingness to offer potential solutions to their barriers which might improve the utilisation of the system in primary care. Conclusion: The GDPs experienced common obstacles in using ICDAS in the primary care setting, many of which have relatively straight-forward solutions which they put forward themselves such as: incentivisation, undergraduate-level training in ICDAS for both dentists and nurses and computerized data entry. Further qualitative and quantitative research is needed on how to facilitate the utilisation of the system in dental practice. It is also recommended to explore the influences of wider agencies on influencing primary dental care professionals’ caries management, including appropriate recording of diagnosis and risk assessment. [ABSTRACT FROM AUTHOR]

When probed with complex stimuli that extend beyond their classical receptive field, neurons in primary visual cortex display complex and non-linear response characteristics. Sparse coding models reproduce some of the observed contextual effects, but still fail to provide a satisfactory explanation in terms of realistic neural structures and cortical mechanisms, since the connection scheme they propose consists only of interactions among neurons with overlapping input fields. Here we propose an extended generative model for visual scenes that includes spatial dependencies among different features. We derive a neurophysiologically realistic inference scheme under the constraint that neurons have direct access only to local image information. The scheme can be interpreted as a network in primary visual cortex where two neural populations are organized in different layers within orientation hypercolumns that are connected by local, short-range and long-range recurrent interactions. When trained with natural images, the model predicts a connectivity structure linking neurons with similar orientation preferences matching the typical patterns found for long-ranging horizontal axons and feedback projections in visual cortex. Subjected to contextual stimuli typically used in empirical studies, our model replicates several hallmark effects of contextual processing and predicts characteristic differences for surround modulation between the two model populations. In summary, our model provides a novel framework for contextual processing in the visual system proposing a well-defined functional role for horizontal axons and feedback projections. [ABSTRACT FROM AUTHOR]

Temporal hierarchies have been widely used during the past few years as they are capable to provide more accurate coherent forecasts at different planning horizons. However, they still display some limitations, being mainly subject to the forecasting methods used for generating the base forecasts and the particularities of the examined series. This paper deals with such limitations by considering three different strategies: (i) combining forecasts of multiple methods, (ii) applying bias adjustments and (iii) selectively implementing temporal hierarchies to avoid seasonal shrinkage. The proposed strategies can be applied either separately or simultaneously, being complements to the method considered for reconciling the base forecasts and completely independent from each other. Their effect is evaluated using the monthly series of the M and M3 competitions. The results are very promising, displaying lots of potential for improving the performance of temporal hierarchies, both in terms of accuracy and bias. [ABSTRACT FROM AUTHOR]

Background and objective: To take full advantage of decision support, machine learning, and patient-level prediction models, it is important that models are not only created, but also deployed in a clinical setting. The KETOS platform demonstrated in this work implements a tool for researchers allowing them to perform statistical analyses and deploy resulting models in a secure environment. Methods: The proposed system uses Docker virtualization to provide researchers with reproducible data analysis and development environments, accessible via Jupyter Notebook, to perform statistical analysis and develop, train and deploy models based on standardized input data. The platform is built in a modular fashion and interfaces with web services using the Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR) standard to access patient data. In our prototypical implementation we use an OMOP common data model (OMOP-CDM) database. The architecture supports the entire research lifecycle from creating a data analysis environment, retrieving data, and training to final deployment in a hospital setting. Results: We evaluated the platform by establishing and deploying an analysis and end user application for hemoglobin reference intervals within the University Hospital Erlangen. To demonstrate the potential of the system to deploy arbitrary models, we loaded a colorectal cancer dataset into an OMOP database and built machine learning models to predict patient outcomes and made them available via a web service. We demonstrated both the integration with FHIR as well as an example end user application. Finally, we integrated the platform with the open source DataSHIELD architecture to allow for distributed privacy preserving data analysis and training across networks of hospitals. Conclusion: The KETOS platform takes a novel approach to data analysis, training and deploying decision support models in a hospital or healthcare setting. It does so in a secure and privacy-preserving manner, combining the flexibility of Docker virtualization with the advantages of standardized vocabularies, a widely applied database schema (OMOP-CDM), and a standardized way to exchange medical data (FHIR). [ABSTRACT FROM AUTHOR]

Rationale: Multiple clinical trials support the effectiveness of cardiac resynchronization therapy (CRT); however, optimal patient selection remains challenging due to substantial treatment heterogeneity among patients who meet the clinical practice guidelines. Objective: To apply machine learning to create an algorithm that predicts CRT outcome using electronic health record (EHR) data avaible before the procedure. Methods and results: We applied machine learning and natural language processing to the EHR of 990 patients who received CRT at two academic hospitals between 2004–2015. The primary outcome was reduced CRT benefit, defined as <0% improvement in left ventricular ejection fraction (LVEF) 6–18 months post-procedure or death by 18 months. Data regarding demographics, laboratory values, medications, clinical characteristics, and past health services utilization were extracted from the EHR available before the CRT procedure. Bigrams (i.e., two-word sequences) were also extracted from the clinical notes using natural language processing. Patients accrued on average 75 clinical notes (SD, 29) before the procedure including data not captured anywhere else in the EHR. A machine learning model was built using 80% of the patient sample (training and validation dataset), and tested on a held-out 20% patient sample (test dataset). Among 990 patients receiving CRT the mean age was 71.6 (SD, 11.8), 78.1% were male, 87.2% non-Hispanic white, and the mean baseline LVEF was 24.8% (SD, 7.69). Out of 990 patients, 403 (40.7%) were identified as having a reduced benefit from the CRT device (<0% LVEF improvement in 25.2%, death by 18 months in 15.6%). The final model identified 26% of these patients at a positive predictive value of 79% (model performance: Fβ (β = 0.1): 77%; recall 0.26; precision 0.79; accuracy 0.65). Conclusions: A machine learning model that leveraged readily available EHR data and clinical notes identified a subset of CRT patients who may not benefit from CRT before the procedure. [ABSTRACT FROM AUTHOR]

Introduction: The earliest changes in the brain due to Alzheimer’s disease are associated with the neural networks related to memory function. We investigated changes in functional and structural connectivity among regions that support memory function in prodromal Alzheimer’s disease, i.e., during the mild cognitive impairment (MCI) stage. Methods: Twenty-three older healthy controls and 25 adults with MCI underwent multimodal MRI scanning. Limbic white matter tracts–the fornix, parahippocampal cingulum, retrosplenial cingulum, subgenual cingulum and uncinate fasciculus–were reconstructed in ExploreDTI using constrained spherical deconvolution-based tractography. Using a network-of-interest approach, resting-state functional connectivity time-series correlations among sub-parcellations of the default mode and limbic networks, the hippocampus and the thalamus were calculated in Conn. Analysis: Controlling for age, education, and gender between group linear regressions of five diffusion-weighted measures and of resting state connectivity measures were performed per hemisphere. FDR-corrections were performed within each class of measures. Correlations of within-network Fisher Z-transformed correlation coefficients and the mean diffusivity per tract were performed. Whole-brain graph theory measures of cluster coefficient and average path length were inspecting using the resting state data. Results & conclusion: MCI-related changes in white matter structure were found in the fornix, left parahippocampal cingulum, left retrosplenial cingulum and left subgenual cingulum. Functional connectivity decreases were observed in the MCI group within the DMN-a sub-network, between the hippocampus and sub-areas -a and -c of the DMN, between DMN-c and DMN-a, and, in the right hemisphere only between DMN-c and both the thalamus and limbic-a. No relationships between white matter tract ‘integrity’ (mean diffusivity) and within sub-network functional connectivity were found. Graph theory revealed that changes in the MCI group was mostly restricted to diminished between-neighbour connections of the hippocampi and of nodes within DMN-a and DMN-b. [ABSTRACT FROM AUTHOR]

Predictive processing (e.g., predictive coding) is a predominant paradigm in cognitive neuroscience. This Primer considers the various levels of commitment neuroscientists have to the neuronal process theories that accompany the principles of predictive processing. Specifically, it reviews and contextualises a recent PLOS Biology study of alpha oscillations and travelling waves. We will see that alpha oscillations emerge naturally under the computational architectures implied by predictive coding-and may tell us something profound about recurrent message passing in brain hierarchies. Specifically, the bidirectional nature of forward and backward waves speaks to opportunities to understand attention and how it nuances bottom-up and top-down influences. [ABSTRACT FROM AUTHOR]

Seasonal influenza is a sometimes surprisingly impactful disease, causing thousands of deaths per year along with much additional morbidity. Timely knowledge of the outbreak state is valuable for managing an effective response. The current state of the art is to gather this knowledge using in-person patient contact. While accurate, this is time-consuming and expensive. This has motivated inquiry into new approaches using internet activity traces, based on the theory that lay observations of health status lead to informative features in internet data. These approaches risk being deceived by activity traces having a coincidental, rather than informative, relationship to disease incidence; to our knowledge, this risk has not yet been quantitatively explored. We evaluated both simulated and real activity traces of varying deceptiveness for influenza incidence estimation using linear regression. We found that deceptiveness knowledge does reduce error in such estimates, that it may help automatically-selected features perform as well or better than features that require human curation, and that a semantic distance measure derived from the Wikipedia article category tree serves as a useful proxy for deceptiveness. This suggests that disease incidence estimation models should incorporate not only data about how internet features map to incidence but also additional data to estimate feature deceptiveness. By doing so, we may gain one more step along the path to accurate, reliable disease incidence estimation using internet data. This capability would improve public health by decreasing the cost and increasing the timeliness of such estimates. [ABSTRACT FROM AUTHOR]

Objective: We examined the incremental cost-effectiveness between two mHealth programs, implemented from 2011 to 2015 in rural Bangladesh: (1) Comprehensive mCARE package as an intervention group and (2) Basic mCARE package as a control group. Methods: Both programs included a core package of census enumeration and pregnancy surveillance provided by an established cadre of digitally enabled community health workers (CHWs). In the comprehensive mCARE package, short message service (SMS) and home visit reminders were additionally sent to pregnant women (n = 610) and CHWs (n = 70) to promote the pregnant women’s care-seeking of essential maternal and newborn care services. Economic costs were assessed from a program perspective inclusive of development, start-up, and implementation phases. Effects were calculated as disability adjusted life years (DALYs) and the number of newborn deaths averted. For comparative purposes, we normalized our evaluation to estimate total costs and total newborn deaths averted per 1 million people in a community for both groups. Uncertainty was assessed using probabilistic sensitivity analyses with Monte Carlo simulation. Results: The addition of SMS and home visit reminders based on a mobile phone-facilitated pregnancy surveillance system was highly cost effective at a cost per DALY averted of $31 (95% uncertainty range: $19–81). The comprehensive mCARE program had at least 88% probability of being highly cost-effective as compared to the basic mCARE program based on the threshold of Bangladesh’s GDP per capita. Conclusion: mHealth strategies such as SMS and home visit reminders on a well-established pregnancy surveillance system may improve service utilization and program cost-effectiveness in low-resource settings. [ABSTRACT FROM AUTHOR]

The biology and ecology of Africa’s largest fruit bat remains largely understudied and enigmatic despite at least two highly unusual attributes. The acoustic lek mating behavior of the hammer-headed bat (Hypsignathus monstrosus) in the Congo basin was first described in the 1970s. More recently molecular testing implicated this species and other African bats as potential reservoir hosts for Ebola virus and it was one of only two fruit bat species epidemiologically linked to the 2008 Luebo, Democratic Republic of Congo, Ebola outbreak. Here we share findings from the first pilot study of hammer-headed bat movement using GPS tracking and accelerometry units and a small preceding radio-tracking trial at an apparent lekking site. The radio-tracking revealed adult males had high rates of nightly visitation to the site compared to females (only one visit) and that two of six females day-roosted ~100 m west of Libonga, the nearest village that is ~1.6 km southwest. Four months later, in mid-April 2018, five individual bats, comprised of four males and one female, were tracked from two to 306 days, collecting from 67 to 1022 GPS locations. As measured by mean distance to the site and proportion of nightly GPS locations within 1 km of the site (percent visitation), the males were much more closely associated with the site (mean distance 1.4 km; 51% visitation), than the female (mean 5.5 km; 2.2% visitation). Despite the small sample size, our tracking evidence supports our original characterization of the site as a lek, and the lek itself is much more central to male than female movement. Moreover, our pilot demonstrates the technical feasibility of executing future studies on hammer-headed bats that will help fill problematic knowledge gaps about zoonotic spillover risks and the conservation needs of fruit bats across the continent. [ABSTRACT FROM AUTHOR]

Given school enrollments but in the absence of a student residence census, we present a gravity-like model to infer the residential distribution of enrolled students across various administrative units. Multi-scale analysis of the effects of aggregation across different administrative levels allows for the identification of administrative units with sub-optimally located schools and highlights the challenges in allocating resources. Using this method, we verify that the current scheme of free cross-enrollment across administrative boundaries is needed in achieving universal education in the Philippines. [ABSTRACT FROM AUTHOR]

This paper presents the Hybrid Scalable-Minimized-Butterfly-Fat-Tree (H-SMBFT) topology for on-chip communication. Main aspects of this work are the description of the architectural design and the characteristics as well as a comparative analysis against two established indirect topologies namely Butterfly-Fat-Tree (BFT) and Scalable-Minimized-Butterfly-Fat-Tree (SMBFT). Simulation results demonstrate that the proposed topology outperforms its predecessors in terms of performance, area and power dissipation. Specifically, it improves the link interconnectivity between routing levels, such that the number of required links isreduced. This results into reduced router complexity and shortened routing paths between any pair of communicating nodes in the network. Moreover, simulation results under synthetic as well as real-world embedded applications workloads reveal that H-SMBFT can reduce the average latency by up-to35.63% and 17.36% compared to BFT and SMBFT, respectively. In addition, the power dissipation of the network can be reduced by up-to33.82% and 19.45%, while energy consumption can be improved byup-to32.91% and 16.83% compared to BFT and SMBFT, respectively. [ABSTRACT FROM AUTHOR]

In light of the problem of environmental pollution caused by fossil fuel combustion, and its association with rapid urbanization, China is grappling with the question of how to reduce carbon emissions through more efficient energy consumption while simultaneously advancing its economic development. We applied a directional distance function to estimate the carbon emission performance of 30 provinces in China during the period 2000–2016. We selected an index system to assess urbanization processes in these provinces and conducted a spatial analysis to investigate the relationship between urbanization and carbon emission performance. We obtained the following results. First, the carbon emission performance of the eastern region, valued at 0.853, was relatively higher than the corresponding values of 0.810, 0.804, and 0.843 in the central, western, and northeastern regions, respectively. However, during this period, disparities among provinces increased. Second, the average urbanization value for each province showed an upward trend during the study period, and urbanization assumed a “striped” spatial agglomeration pattern. A third finding was that carbon emission performance and urbanization demonstrated a relationship of positive spatial dependence. The average value of their coordinated coupling indicated an upward trend, with an annual increase of 0.85%. Last, we found that efforts to reduce carbon emissions that are solely based on carbon emission performance do not yield reliable results. Accordingly, measurements of urbanization values can enable more detailed differentiation. In conclusion, reasonable measures should be implemented to improve carbon emission performance and urbanization that are in alignment with the actual situation within a given region. [ABSTRACT FROM AUTHOR]

Collective behaviour is of fundamental importance in the life sciences, where it appears at levels of biological complexity from single cells to superorganisms, in demography and the social sciences, where it describes the behaviour of populations, and in the physical and engineering sciences, where it describes physical phenomena and can be used to design distributed systems. Reasoning about collective behaviour is inherently difficult, as the non-linear interactions between individuals give rise to complex emergent dynamics. Mathematical techniques have been developed to analyse systematically collective behaviour in such systems, yet these frequently require extensive formal training and technical ability to apply. Even for those with the requisite training and ability, analysis using these techniques can be laborious, time-consuming and error-prone. Together these difficulties raise a barrier-to-entry for practitioners wishing to analyse models of collective behaviour. However, rigorous modelling of collective behaviour is required to make progress in understanding and applying it. Here we present an accessible tool which aims to automate the process of modelling and analysing collective behaviour, as far as possible. We focus our attention on the general class of systems described by reaction kinetics, involving interactions between components that change state as a result, as these are easily understood and extracted from data by natural, physical and social scientists, and correspond to algorithms for component-level controllers in engineering applications. By providing simple automated access to advanced mathematical techniques from statistical physics, nonlinear dynamical systems analysis, and computational simulation, we hope to advance standards in modelling collective behaviour. At the same time, by providing expert users with access to the results of automated analyses, sophisticated investigations that could take significant effort are substantially facilitated. Our tool can be accessed online without installing software, uses a simple programmatic interface, and provides interactive graphical plots for users to develop understanding of their models. [ABSTRACT FROM AUTHOR]

People with schizophrenia report positive emotion during social interactions in ecological momentary assessment (EMA) studies; however, few of these studies examine the qualities of social interactions (e.g., intimacy) that may affect emotion experience. In the current EMA study, people with (n = 20) and without schizophrenia (n = 15) answered questions about the quality of their social interactions, including their emotion experiences. We also explored the relationship between EMA-reported social experiences and trait loneliness, negative symptoms, and social functioning. People with and without schizophrenia did not differ in EMA-reported proportion of time spent with others, extent of involvement during social interactions, intimacy of interactions, or average number of social interactions. Both people with and without schizophrenia reported more positive than negative emotion during social experiences. However, people with schizophrenia reported more loneliness, more severe negative symptoms, and impaired social functioning compared to people without schizophrenia. Further, specific qualities of social interactions (intimacy of interaction, involvement during interaction) were related to happiness during interactions only in people without schizophrenia. These results suggest that while people with and without schizophrenia report similar rates of in-the-moment social emotion experiences, the impact of social interaction quality on emotion may differ between groups. [ABSTRACT FROM AUTHOR]

This study aims to analyze the level of anger developed by drivers in Mexico City and also understand the behavior that those drivers use to express that anger, using four different survey methods. The first focuses on personal information, the second Driving Anger Expression Inventory (DAX), the third refers to a shorten version of Driving Anger Scale (DAS) and the fourth being the Dula Dangerous Driving Index (DDDI). These have previously been applied and validated in several different countries. The questionnaires were filled out online by 626 drivers. Using the data collected through the online platform, it was possible to identify the kind of reactions volunteers displayed while driving. Also, it was possible to identify that people in Mexico City developed anger depending on their driving area. Our analyses shows that in the Adaptive/Constructive Expression subscale, males and females show a significant difference in their mean score, with women express their anger in a more constructive way than males. [ABSTRACT FROM AUTHOR]

Background: Alcohol use disorder (AUD) is known to co-occur with other addictions, as well as with mental health problems. However, the effects of other addictions co-occurring with AUD on mental health problems were rarely studied and not considering them may bias estimates of the association between AUD and mental health problems. This study investigated which role co-occurring addictions play for the cross-sectional associations between self-reported AUD and mental health problems. Method: Participants were 5516 young Swiss men (73.0% of those that gave written informed consent) who completed a self-report questionnaire. Using short screening questionnaires, we assessed three substance use disorders (alcohol, cannabis and tobacco), seven behavioural addictions (internet, gaming, smartphone, internet sex, gambling, work, exercise) and four mental health problems (major depression, bipolar disorder, attention deficit hyperactivity disorder (ADHD) and social anxiety disorder). Differences in the proportions of mental health problems were tested using logistic regressions between (1) participants with no AUD and AUD, (2) participants with no AUD and AUD alone and (3) participants with no AUD and AUD plus at least one co-occurring addiction. Results: Overall, (1) participants with AUD had higher proportions of major depression (Odds ratio (OR [95% confidence interval]) = 3.51 [2.73, 4.52]; ADHD (OR = 3.12 [2.41, 4.03]); bipolar disorder (OR = 4.94 [3.38, 7.21]) and social anxiety (OR = 2.21 [1.79, 2.73])) compared to participants with no AUD. Considering only participants with AUD alone compared to participants with no AUD (2), differences in proportions were no longer significant for major depression (OR = 0.83 [0.42, 1.64]), bipolar disorder (OR = 1.69 [0.67, 4.22]), social anxiety (OR = 1.15 [0.77, 1.73]) and ADHD (OR = 1.65 [1.00, 2.72]) compared to participants with no AUD. In contrast, (3) proportions of mental health problems were considerably higher for participants with AUD plus at least one other addiction when compared to participants with no AUD, with OR’s ranging from 2.90 [2.27, 3.70] for social anxiety, 4.03 [3.02, 5.38] for ADHD, 5.29 [4.02, 6.97] for major depression to 6.64 [4.44, 9.94] for bipolar disorder. Conclusions: AUD was associated with all four measured mental health problems. However, these associations were mainly due to the high proportions of these mental health problems in participants with AUD plus at least one co-occurring addiction and only to a lesser degree due to participants with AUD alone (i.e. without any other co-occurring addictions). Hence, estimates of the association between AUD and mental health problems that do not consider other addictions may be biased (i.e. overestimated). These findings imply that considering addictions co-occurring with AUD, including behavioural addictions, is important when investigating associations between AUD and mental health problems, and for the treatment of AUD and co-morbid disorders. [ABSTRACT FROM AUTHOR]

Biological network figures are ubiquitous in the biology and medical literature. On the one hand, a good network figure can quickly provide information about the nature and degree of interactions between items and enable inferences about the reason for those interactions. On the other hand, good network figures are difficult to create. In this paper, we outline 10 simple rules for creating biological network figures for communication, from choosing layouts, to applying color or other channels to show attributes, to the use of layering and separation. These rules are accompanied by illustrative examples. We also provide a concise set of references and additional resources for each rule. [ABSTRACT FROM AUTHOR]

Although information theoretic approaches have been used extensively in the analysis of the neural code, they have yet to be used to describe how information is accumulated in time while sensory systems are categorizing dynamic sensory stimuli such as speech sounds or visual objects. Here, we present a novel method to estimate the cumulative information for stimuli or categories. We further define a time-varying categorical information index that, by comparing the information obtained for stimuli versus categories of these same stimuli, quantifies invariant neural representations. We use these methods to investigate the dynamic properties of avian cortical auditory neurons recorded in zebra finches that were listening to a large set of call stimuli sampled from the complete vocal repertoire of this species. We found that the time-varying rates carry 5 times more information than the mean firing rates even in the first 100 ms. We also found that cumulative information has slow time constants (100–600 ms) relative to the typical integration time of single neurons, reflecting the fact that the behaviorally informative features of auditory objects are time-varying sound patterns. When we correlated firing rates and information values, we found that average information correlates with average firing rate but that higher-rates found at the onset response yielded similar information values as the lower-rates found in the sustained response: the onset and sustained response of avian cortical auditory neurons provide similar levels of independent information about call identity and call-type. Finally, our information measures allowed us to rigorously define categorical neurons; these categorical neurons show a high degree of invariance for vocalizations within a call-type. Peak invariance is found around 150 ms after stimulus onset. Surprisingly, call-type invariant neurons were found in both primary and secondary avian auditory areas. [ABSTRACT FROM AUTHOR]

Internet and social media participation open doors to a plethora of positive opportunities for the general public. However, in addition to these positive aspects, digital technology also provides an effective medium for spreading hateful content in the form of cyberbullying, bigotry, hateful ideologies, and harassment of individuals and groups. This research aims to investigate the growing body of online hate research (OHR) by mapping general research indices, prevalent themes of research, research hotspots, and influential stakeholders such as organizations and contributing regions. For this, we use scientometric techniques and collect research papers from the Web of Science core database published through March 2019. We apply a predefined search strategy to retrieve peer-reviewed OHR and analyze the data using CiteSpace software by identifying influential papers, themes of research, and collaborating institutions. Our results show that higher-income countries contribute most to OHR, with Western countries accounting for most of the publications, funded by North American and European funding agencies. We also observed increased research activity post-2005, starting from more than 50 publications to more than 550 in 2018. This applies to a number of publications as well as citations. The hotbeds of OHR focus on cyberbullying, social media platforms, co-morbid mental disorders, and profiling of aggressors and victims. Moreover, we identified four main clusters of OHR: (1) Cyberbullying, (2) Sexual solicitation and intimate partner violence, (3) Deep learning and automation, and (4) Extremist and online hate groups, which highlight the cross-disciplinary and multifaceted nature of OHR as a field of research. The research has implications for researchers and policymakers engaged in OHR and its associated problems for individuals and society. [ABSTRACT FROM AUTHOR]

Humans evolved to be endurance animals. Our ancestors were persistence hunters; they would chase animals, including gazelles, until they ran them into exhaustion. Put simply, people evolved in an ecological niche that selected for endurance and efficiency of locomotion. To locomote to any destination, one could take countless different paths, each requiring different amounts of energy. Because the ground is typically not flat or homogeneous, the straight direct path is often not the most energetically efficient. For hills below 14°, the direct straight path up the hill is the most energetically efficient. However, for hills above 14°, walkers would minimize their absolute energy expenditure by taking a zigzagged path so that their gradient of ascension is 14° []. In three experiments, we assessed the degree to which people make bioenergetically efficient decisions about locomotion through path selection. In Experiment 1, people were immersed into a virtual environment and adjusted the angle of ascension of a virtual path up hills of various gradients so that when taking the path, they would expend the least amount of energy when they reached the top. The second experiment was of a similar design, but was conducted in the real word. In the last experiment, in a virtual environment, participants choose between two paths up hills of various gradient, where these paths varied in the energy required for ascent. Participants made these judgements both before and after motor experience with gradient climbing on an incline trainer. For steep hills, we found that people choose much straighter paths over the bioenergetically optimal zigzagged paths. Motor experience did lead to higher probability for choosing optimal paths for steep hills, but lead to less optimal paths for shallower ones. These results show clearly that individuals show a straight path bias when deciding how to ascend hills. [ABSTRACT FROM AUTHOR]

With the rapid development of the internet, social media has become an essential tool for getting information, and attracted a large number of people join the social media platforms because of its low cost, accessibility and amazing content. It greatly enriches our life. However, its rapid development and widespread also have provided an excellent convenience for the range of fake news, people are constantly exposed to fake news and suffer from it all the time. Fake news usually uses hyperbole to catch people’s eyes with dishonest intention. More importantly, it often misleads the reader and causes people to have wrong perceptions of society. It has the potential for negative impacts on society and individuals. Therefore, it is significative research on detecting fake news. In the paper, we built a model named SMHA-CNN (Self Multi-Head Attention-based Convolutional Neural Networks) that can judge the authenticity of news with high accuracy based only on content by using convolutional neural networks and self multi-head attention mechanism. In order to prove its validity, we conducted experiments on a public dataset and achieved a precision rate of 95.5% with a recall rate of 95.6% under the 5-fold cross-validation. Our experimental result indicates that the model is more effective at detecting fake news. [ABSTRACT FROM AUTHOR]

Alpha event-related desynchronization (ERD) has been widely applied to understand the psychophysiological role of this band in cognition. In particular, a considerable number of publications have described spectral alterations in several pathologies using this time-frequency approach. However, ERD is not capable of specifically showing nonphase (induced) activity related to the presentation of stimuli. Recent studies have described an evoked and induced activity in the early phases (first 200 ms) of stimulus processing. However, scarce studies have analyzed induced and evoked modulations in longer latencies (>200 ms) and their potential roles in cognitive processing. The main goal of the present study was to analyze diverse evoked and induced modulations in response to visual stimuli. Thus, 58-channel electroencephalogram (EEG) was recorded in 21 healthy subjects during the performance of a visual attention task, and analyses were performed for both target and standard stimuli. The initial result showed that phase-locked and nonphase locked activities coexist in the early processing of target and standard stimuli as has been reported by previous studies. However, more modulations were evident in longer latencies in both evoked and induced activities. Correlation analyses suggest that similar maps were present for evoked and induced activities at different timepoints. In the discussion section, diverse proposals will be stated to define the potential roles of these modulations in the information processing for this cognitive task. As a general conclusion, induced activity enables the observation of cognitive mechanisms that are not visible by ERD or ERP modulations. [ABSTRACT FROM AUTHOR]

Industrial robots are a strategic future technology and an important part of the development of artificial intelligence, and they are a necessary means for the intelligent transformation of manufacturing industry. Based on global industrial robot trade data from 1998 to 2017, this paper applies the dynamic complex network analysis method to reveal the spatial and temporal variation characteristics and trade status evolution of the global industrial robot trade network. The results show that the global industrial robot network density has steadily increased, and the industrial robot trade has been characterized by ‘diversification’. The number of major industrial robot exporters in the world is increasing, and the import market is increasingly diversified. The export market structure is relatively tight, the centrality of the global industrial robot trade network shows a downward trend, and the dissimilarity of the ‘core-edge’ clusters decreases year by year. The trade status of ‘catch-up’ countries represented by China has rapidly increased. However, Japan, Germany, and Italy are still in the central position of the industrial robot trade. Moreover, trade of the ‘catch-up’ countries’ is dominated by imports, and exports of industrial robot products are insufficient. Finally, policy suggestions are provided according to the results. [ABSTRACT FROM AUTHOR]

Objective: To conduct a time-cost analysis of formatting in scientific publishing. Design: International, cross-sectional study (one-time survey). Setting: Internet-based self-report survey, live between September 2018 and January 2019. Participants: Anyone working in research, science, or academia and who submitted at least one peer-reviewed manuscript for consideration for publication in 2017. Completed surveys were available for 372 participants from 41 countries (60% of respondents were from Canada). Main outcome measure: Time (hours) and cost (wage per hour x time) associated with formatting a research paper for publication in a peer-reviewed academic journal. Results: The median annual income category was US$61,000–80,999, and the median number of publications formatted per year was four. Manuscripts required a median of two attempts before they were accepted for publication. The median formatting time was 14 hours per manuscript, or 52 hours per person, per year. This resulted in a median calculated cost of US$477 per manuscript or US$1,908 per person, per year. Conclusions: To our knowledge, this is the first study to analyze the cost of manuscript formatting in scientific publishing. Our results suggest that scientific formatting represents a loss of 52 hours, costing the equivalent of US$1,908 per researcher per year. These results identify the hidden and pernicious price associated with scientific publishing and provide evidence to advocate for the elimination of strict formatting guidelines, at least prior to acceptance. [ABSTRACT FROM AUTHOR]

A recent work (Hernández, et al., 2018) introduced a networked voting rule supported by a trust-based social network, where indications of possible representatives were based on individuals opinions. Individual contributions went beyond a simple vote-counting and were based on proxy voting. This mechanism selects committees with high levels of representativeness, weakening the possibility of patronage relations. By incorporating the integrity of individuals and its perception, we here address the question of the resulting committee’s trustability. Our results show that this voting rule provides sufficiently small committees with high levels of representativeness and integrity. Furthermore, the voting system displays robustness to strategic and untruthful application of the voting algorithm. [ABSTRACT FROM AUTHOR]

We show that Cohen’s Kappa and Matthews Correlation Coefficient (MCC), both extended and contrasted measures of performance in multi-class classification, are correlated in most situations, albeit can differ in others. Indeed, although in the symmetric case both match, we consider different unbalanced situations in which Kappa exhibits an undesired behaviour, i.e. a worse classifier gets higher Kappa score, differing qualitatively from that of MCC. The debate about the incoherence in the behaviour of Kappa revolves around the convenience, or not, of using a relative metric, which makes the interpretation of its values difficult. We extend these concerns by showing that its pitfalls can go even further. Through experimentation, we present a novel approach to this topic. We carry on a comprehensive study that identifies an scenario in which the contradictory behaviour among MCC and Kappa emerges. Specifically, we find out that when there is a decrease to zero of the entropy of the elements out of the diagonal of the confusion matrix associated to a classifier, the discrepancy between Kappa and MCC rise, pointing to an anomalous performance of the former. We believe that this finding disables Kappa to be used in general as a performance measure to compare classifiers. [ABSTRACT FROM AUTHOR]

Data quality control is important for any data collection program, especially in citizen science projects, where it is more likely that errors occur due to the human factor. Ideally, data quality control in citizen science projects is also crowdsourced so that it can handle large amounts of data. Here we present the CrowdWater game as a gamified method to check crowdsourced water level class data that are submitted by citizen scientists through the CrowdWater app. The app uses a virtual staff gauge approach, which means that a digital scale is added to the first picture taken at a site and this scale is used for water level class observations at different times. In the game, participants classify water levels based on the comparison of the new picture with the picture containing the virtual staff gauge. By March 2019, 153 people had played the CrowdWater game and 841 pictures were classified. The average water level for the game votes for the classified pictures was compared to the water level class submitted through the app to determine whether the game can improve the quality of the data submitted through the app. For about 70% of the classified pictures, the water level class was the same for the CrowdWater app and game. For a quarter of the classified pictures, there was disagreement between the value submitted through the app and the average game vote. Expert judgement suggests that for three quarters of these cases, the game based average value was correct. The initial results indicate that the CrowdWater game helps to identify erroneous water level class observations from the CrowdWater app and provides a useful approach for crowdsourced data quality control. This study thus demonstrates the potential of gamified approaches for data quality control in citizen science projects. [ABSTRACT FROM AUTHOR]

Twitter was an integral part of Donald Trump’s communication platform during his 2016 campaign. Although its topical content has been examined by researchers and the media, we know relatively little about the style of the language used on the account or how this style changed over time. In this study, we present the first detailed description of stylistic variation on the Trump Twitter account based on a multivariate analysis of grammatical co-occurrence patterns in tweets posted between 2009 and 2018. We identify four general patterns of stylistic variation, which we interpret as representing the degree of conversational, campaigning, engaged, and advisory discourse. We then track how the use of these four styles changed over time, focusing on the period around the campaign, showing that the style of tweets shifts systematically depending on the communicative goals of Trump and his team. Based on these results, we propose a series of hypotheses about how the Trump campaign used social media during the 2016 elections. [ABSTRACT FROM AUTHOR]

Objective: The Centers for Disease Control and Prevention (CDC) coordinates a labor-intensive process to measure the prevalence of autism spectrum disorder (ASD) among children in the United States. Random forests methods have shown promise in speeding up this process, but they lag behind human classification accuracy by about 5%. We explore whether more recently available document classification algorithms can close this gap. Materials and methods: Using data gathered from a single surveillance site, we applied 8 supervised learning algorithms to predict whether children meet the case definition for ASD based solely on the words in their evaluations. We compared the algorithms’ performance across 10 random train-test splits of the data, using classification accuracy, F1 score, and number of positive calls to evaluate their potential use for surveillance. Results: Across the 10 train-test cycles, the random forest and support vector machine with Naive Bayes features (NB-SVM) each achieved slightly more than 87% mean accuracy. The NB-SVM produced significantly more false negatives than false positives (P = 0.027), but the random forest did not, making its prevalence estimates very close to the true prevalence in the data. The best-performing neural network performed similarly to the random forest on both measures. Discussion: The random forest performed as well as more recently available models like the NB-SVM and the neural network, and it also produced good prevalence estimates. NB-SVM may not be a good candidate for use in a fully-automated surveillance workflow due to increased false negatives. More sophisticated algorithms, like hierarchical convolutional neural networks, may not be feasible to train due to characteristics of the data. Current algorithms might perform better if the data are abstracted and processed differently and if they take into account information about the children in addition to their evaluations. Conclusion: Deep learning models performed similarly to traditional machine learning methods at predicting the clinician-assigned case status for CDC’s autism surveillance system. While deep learning methods had limited benefit in this task, they may have applications in other surveillance systems. [ABSTRACT FROM AUTHOR]