Probability Distributions.- Linear Models for Regression.- Linear Models for Classification.- Neural Networks.- Kernel Methods.- Sparse Kernel Machines.- Graphical Models.- Mixture Models and EM.- Approximate Inference.- Sampling Methods.- Continuous Latent Variables.- Sequential Data.- Combining Models.

Pattern Recognition and Machine Learning

Objective To review and appraise the validity and usefulness of published and preprint reports of prediction models for diagnosing coronavirus disease 2019 (covid-19) in patients with suspected infection, for prognosis of patients with covid-19, and for detecting people in the general population at increased risk of covid-19 infection or being admitted to hospital with the disease. Design Living systematic review and critical appraisal by the COVID-PRECISE (Precise Risk Estimation to optimise covid-19 Care for Infected or Suspected patients in diverse sEttings) group. Data sources PubMed and Embase through Ovid, up to 1 July 2020, supplemented with arXiv, medRxiv, and bioRxiv up to 5 May 2020. Study selection Studies that developed or validated a multivariable covid-19 related prediction model. Data extraction At least two authors independently extracted data using the CHARMS (critical appraisal and data extraction for systematic reviews of prediction modelling studies) checklist; risk of bias was assessed using PROBAST (prediction model risk of bias assessment tool). Results 37 421 titles were screened, and 169 studies describing 232 prediction models were included. The review identified seven models for identifying people at risk in the general population; 118 diagnostic models for detecting covid-19 (75 were based on medical imaging, 10 to diagnose disease severity); and 107 prognostic models for predicting mortality risk, progression to severe disease, intensive care unit admission, ventilation, intubation, or length of hospital stay. The most frequent types of predictors included in the covid-19 prediction models are vital signs, age, comorbidities, and image features. Flu-like symptoms are frequently predictive in diagnostic models, while sex, C reactive protein, and lymphocyte counts are frequent prognostic factors. Reported C index estimates from the strongest form of validation available per model ranged from 0.71 to 0.99 in prediction models for the general population, from 0.65 to more than 0.99 in diagnostic models, and from 0.54 to 0.99 in prognostic models. All models were rated at high or unclear risk of bias, mostly because of non-representative selection of control patients, exclusion of patients who had not experienced the event of interest by the end of the study, high risk of model overfitting, and unclear reporting. Many models did not include a description of the target population (n=27, 12%) or care setting (n=75, 32%), and only 11 (5%) were externally validated by a calibration plot. The Jehi diagnostic model and the 4C mortality score were identified as promising models. Conclusion Prediction models for covid-19 are quickly entering the academic literature to support medical decision making at a time when they are urgently needed. This review indicates that almost all pubished prediction models are poorly reported, and at high risk of bias such that their reported predictive performance is probably optimistic. However, we have identified two (one diagnostic and one prognostic) promising models that should soon be validated in multiple cohorts, preferably through collaborative efforts and data sharing to also allow an investigation of the stability and heterogeneity in their performance across populations and settings. Details on all reviewed models are publicly available at https://www.covprecise.org/. Methodological guidance as provided in this paper should be followed because unreliable predictions could cause more harm than benefit in guiding clinical decisions. Finally, prediction model authors should adhere to the TRIPOD (transparent reporting of a multivariable prediction model for individual prognosis or diagnosis) reporting guideline. Systematic review registration Protocol https://osf.io/ehc47/, registration https://osf.io/wy245. Readers’ note This article is a living systematic review that will be updated to reflect emerging evidence. Updates may occur for up to two years from the date of original publication. This version is update 3 of the original article published on 7 April 2020 (BMJ 2020;369:m1328). Previous updates can be found as data supplements (https://www.bmj.com/content/369/bmj.m1328/related#datasupp). When citing this paper please consider adding the update number and date of access for clarity.

https://www.bmj.com/content/bmj/369/bmj.m1328.full.pdf

Prediction models for diagnosis and prognosis of covid-19: systematic review and critical appraisal

IEEE Transactions on Pattern Analysis and Machine Intelligence

Pattern recognition

Biophysics of Computation

O presente estudo aborda o desenvolvimento e a sistematizacao de uma interface cerebro-computador, baseada em potenciais evocados visualmente em regime estacionario, para funcionar acoplada a execucao de tarefas. Esse sistema de comunicacao, com o uso de sinais provenientes de eletroencefalografia, foi criado para vir a integrar uma versao preliminar de um controlador compartilhado destinado a navegacao de uma cadeira de rodas. Em principio, quando um individuo foca o olhar em estimulos visuais periodicos, certos potenciais evocados podem ser detectados entre a atividade natural do cerebro, especialmente na regiao occipital. Se identificados de forma suficientemente precisa, entao, esses potenciais poderiam ser associados aos comandos de acionamento da referida cadeira. Para atingir esse proposito, a interface foi estruturada em cinco modulos: o de estimulacao, responsavel por gerar e exibir os estimulos visuais repetitivos; o de aquisicao, dedicado a captura e ao armazenamento dos dados eletroencefalograficos; o de processamento, que realiza a remocao de artefatos, a extracao de caracteristicas e a formatacao das mesmas; o de treinamento, que executa a selecao de atributos e que calcula as projecoes de um classificador linear para identificar as manifestacoes cerebrais relacionados aos estimulos; e, por fim, o de comunicacao e controle, que (com o uso do mencionado classificador) detecta os potenciais evocados e, a partir disso, gera e envia um sinal de controle para a aplicacao. Alem desses modulos, entre estes tambem foi proposto um protocolo de operacao, dividido em etapa de treinamento e etapa de identificacao em tempo de execucao. Ao longo do desenvolvimento da interface cerebro-computador, o funcionamento dos modulos foi validado com sinais adquiridos de dez voluntarios saudaveis e uma versao funcional desse sistema foi testada com tres deles, como forma de verificar a factibilidade de integracao da interface com o controlador da cadeira de rodas. 
Abstract

http://repositorio.unicamp.br/jspui/bitstream/REPOSIP/259836/1/Costa_ThiagoBulhoesdaSilva_M.pdf

Desenvolvimento de interface cérebro-computador baseada em potenciais evocados visualmente em regime estacionário

Lenovo Innovation Center ³CPqD - Telecom and IT Solutions barbosal@dca.fee.unicamp.br, attux@dca.fee.unicamp.br, godoy@dca.fee.unicamp.br Keywords: Assortativity, Digital Library, Graph Databases, Knowledge Discovery 1. INTRODUCTION “Big Data” is a recent phenomenon characterized by the increasingly fast growth of data availability over many human spheres, including Science, Engineering, Biology and others. Academic publications are no exception to this trend and this rising number of scientific documents arouses the interest in the new kinds of scientific discoveries and envisions a distinct way to analyze and visualize such information. In this work in progress, we are developing a method to represent and analyze metadata from scientific papers using multiplex complex networks. The proposed method makes use of text mining techniques to extract suitable data as attributes and later insert them as elements in a complex network. The main goal is, given a set of scientific papers, to investigate how concepts, research institutions and individuals are related and to reflect about the collaboration in the field under analysis. How partnerships between authors and countries are established? Is there any correlation between topics researched in different countries? These are some of the questions we want begin to tackle in this study. 2. IMPLEMENTATION The proposed method is based on the technique defined in [1] to extract attributes (titles, authors, countries, keywords and publication year) and their relationships from scientific articles in PDF format, and lastly, use a graph database to process the data [2]. In this work we intend to analyze assortativity and similarities in the network formed by the extracted data. Previous studies have explored these approaches, however considering complex networks composed by a single node type. In contrast, we propose to apply bibliometrics measures on a multiplex network, illustrated in Figure 1. Thereby, it is possible to evaluate different views of collaboration in academia, in order to identify information and patterns which are hidden a priori. This approach, described in Table 1, allows, for instance, a study of the correlation between publication dates and keywords, showing possible research trends. Characteristic Description Directed Relationship direction Weighted Number of occurrences of each relationship Semantic Sub-networks induction by label filtering Multiplex Multiple node types and relationships Graph Database Supports the dynamic study of the network Table 1. Proposed network profile The analysis performed in scientific networks is based on:  Degree assortativity – How are authors, countries and articles connections associated in the network?  Discrete assortativity by year – How are keywords related according to the [year of publication?  Discrete assortativity by continent – Do countries tend to collaborate with countries from the same continent?  Similarities – Which is the research correlation between countries regarding the article keywords? Are the most cited authors (and their countries) also collaborative? Useful information about assortativeness and network science can be respectively found at [3, 4]. Figure 1. Scientific network partially demonstrated 3. DISCUSSION Preliminary results were obtained from an initial sample with 214 scientific articles related to the index term “collaborative” [1], gathered from the IEEE Xplore database. Positive degree assortativity for authors (r=0.788) and negative for countries (r=-0.116) might indicate that the scientific collaboration is more prone to be established among individuals than among countries. Negative assortativity for articles (r=-0.327) shows that the few most cited articles are associated with several less cited ones. Through the discrete assortativity method, it was possible to identify in the database a moderate propensity (r=0.307) of articles to cite papers written by authors from the same continent. Regarding the similarity analysis, citations are correlated to others aspects: authors and countries of the most cited articles do collaborate in their respective networks and are network hubs. Keywords from these articles are also widely adopted by authors. Also it was possible to highlight some patterns in the international collaboration, such as research syntony, based on article keywords and the information source, considering the origin country. In this initial phase of the work, we could observe that the approach using multiplex networks allows an analysis from different points of view of the same network, using edge labels to induce subnets, apply metrics and then correlate the values found. The results from this work corroborate the idea that collaboration in the scientific world has great impact on the dissemination of information and knowledge throughout the world, as well for the researchers themselves. Acknowledgements to the National Council for Scientific and Technological Development (CNPq) for the financial support. 4. FUTURE WORK  Expand the database in order to validate the results  Analyze the geographic influence on scientific collaboration 5. REFERENCES 1]Barbosa,L.M.,Attux,R.,GodoyA.2014.Umaanalisedeassortatividade e similaridade para Artigos Cientificos. XI Brazilian Symposium on Collaborative Systems, Curitiba, Brazil. [2] Neo4j Graph Database. http://www.neo4j.org. [3] Newman, M. 2002. Assortative Mixing Networks. University of Michigan. Santa Fe Institute. [4] Baronchelli, A., Ferrer-i-Cancho, R., Pastor-Satorras, R., Chater, N., Christiansen, M. 2013. Networks in Cognitive Science. Elsevier.

Multiplex Network Approach for Scientific Articles

In this paper, we address the problem of blind source separation for linear quadratic mixtures The proposed approach relies on the assumption that the input signals are band-limited As the nonlinearity of the mixing process tends to widen the spectra of the mixture signals, and taking into account the fact that there are more mixtures than sources in the overdetermined version of the problem, we propose a method that uses the additional mixtures to eliminate the nonlinearities of the observed signals This gives rise to a linear problem that can be solved with standard methods previously analyzed in the literature Numerical experiments show that the proposed algorithm successfully separates the sources under the proposed conditions

Blind source separation for overdetermined linear quadratic mixtures of bandlimited signals

Signal processing methods based on blind source separation (BSS) have been successfully applied to deal with the interference problem in ion-selective electrode (ISE) arrays. So far, the development of BSS methods to ISE arrays strongly relied on the Nicolsky-Eisenman (NE) model. Since this model provides a rough approximation in certain situations, the present paper aims at improving the existing BSS solutions by refining the mixing model based on the NE equation. More precisely, we place a set of polynomial monotonic functions after the standard logarithmic functions within the NE. The obtained results are preliminary but attest that the proposal lead to slight improvements and open the way for investigations on alternative monotonic compensating functions.

A novel blind source separation method based on monotonic functions and its application to ion-selective electrode arrays

Motor imagery (MI), the mental rehearsal of a movement task, is known to activate similar brain areas to the ones related to actual motor execution. Based on this, MI has been used in many brain-computer interface (BCI) applications, ranging from motor rehabilitation to the actual control of external hardware and other devices. Although great improvement has been made in the field, MI-BCIs still face several issues that limit their applicability to the clinical environment. The aim of this work was to improve on the understanding of one of these issues - namely, the inter and intra-subject variability regarding the electroencephalography (EEG) signals produced by MI tasks. EEG data from 10 healthy subjects who underwent 12 hands-MI sessions, without any feedback, were collected. Differently than most current studies in the field, we screened our analysis into small frequency intervals, from 5–26 Hz, at 4 Hz steps. We then computed how often a given (electrode, frequency interval) pair was optimal for BCI classification, across all sessions, attempting to identify features that would maximize reproducibility. Although similar electrodes are most recurrent across all subjects, each participant displayed their own particularities. Also, these individual patterns did not necessarily reflect the traditional locations over the primary sensorimotor cortex of most EEG-MI studies. Furthermore, our classification results suggested that identifying the best spectral intervals, and not just electrodes, is crucial for improving results. Given the reduced number of existing training protocols supporting individualization, we emphasize that considering subjects’ particularities is essential.

Romis Attux

Papers

Desenvolvimento de interface cérebro-computador baseada em potenciais evocados visualmente em regime estacionário

Multiplex Network Approach for Scientific Articles

Blind source separation for overdetermined linear quadratic mixtures of bandlimited signals

A novel blind source separation method based on monotonic functions and its application to ion-selective electrode arrays

Considerations on the Individualization of Motor Imagery Neurofeedback Training