Navegando por Palavras-chave "Pattern recognition"
Agora exibindo 1 - 5 de 5
Resultados por página
Opções de Ordenação
- ItemAcesso aberto (Open Access)Aplicação de redes neurais artificiais em transplantes renais: classificação de nefrotoxicidade e rejeição celular aguda(Universidade Federal de São Paulo (UNIFESP), 2010-10-27) Maciel, Rafael Fabio [UNIFESP]; Pisa, Ivan Torres [UNIFESP]; Universidade Federal de São Paulo (UNIFESP)BACKGROUND: Complications associated with kidney transplantation and immunosuppression can be prevented or treated effectively if diagnosed in early stages with monitoring post-transplant. OBJECTIVE: To present the results of comparison of automatic pattern classifiers using different techniques of artificial intelligence to predict events of nephrotoxicity and acute cellular rejection (RCA), with up to one year of renal transplantation METHODS: Statistical tests were performed on the prevalence and linear regression in variables regarding nephrotoxicity and RCA. We used different classifiers (neural networks, support vector machines (SVM), decision trees, Bayesian inference, and closest neighbors) in order to provide RCA and nephrotoxicity. The classifiers were evaluated according to the value of sensitivity, specificity and area under ROC curve (AUC). RESULTS: The prevalence of acute cellular rejection was 31.0% and 26.9% of nephrotoxicity. The technique had the highest sensitivity value prediction for the submission to the transplanted kidney biopsy was SVM (LIBSVM algorithm) with sensitivity rates of 0.87 (accuracy rate 79.86; specificity 0.70; AUC 0.79). The technique had the highest AUC for predicting nephrotoxicity and RCA was bayesian inference (NaiveBayes), with AUC rates of 0.8 (accuracy rate 75.92). CONCLUSION: The results are encouraging, with rates of trial and error consistent with the determination of acute cellular rejection and nephrotoxicity.
- ItemAcesso aberto (Open Access)Comparison of Brazilian and American norms for the International Affective Picture System (IAPS)(Associação Brasileira de Psiquiatria - ABP, 2005-09-01) Ribeiro, Rafaela Larsen [UNIFESP]; Pompéia, Sabine [UNIFESP]; Bueno, Orlando Francisco Amodeo [UNIFESP]; Universidade Federal de São Paulo (UNIFESP)OBJECTIVE: The present article compares Brazilian and American norms for the International Affective Picture System (IAPS), a set of normative emotional photographic slides for experimental investigations. METHODS: Subjects were 1,062 Brazilian university students (364 men and 698 women) who rated 707 pictures from the IAPS in terms of pleasure, arousal, and dominance following the methodology of the original normative study in the US, enabling direct comparison of data from the two samples through Pearson product moment correlation and Student t test. RESULTS: All correlations were highly significant with the highest level for the pleasure dimension, followed by dominance and arousal. However, contrary to the American normative values, our data showed that Brazilian subjects generally assigned higher arousal ratings overall. CONCLUSION: Our findings confirm that this set of stimuli can be used in Brazil as an affective rating tool due to the high correlations found across the two populations, despite differences on the arousal dimension, which are discussed in detail.
- ItemSomente MetadadadosFeature selection before EEG classification supports the diagnosis of Alzheimer's disease(Elsevier Ireland Ltd, 2017) Trambaiolli, L. R.; Spolaor, N.; Lorena, A. C. [UNIFESP]; Anghinah, R.; Sato, J. R.Objective: In many decision support systems, some input features can be marginal or irrelevant to the diagnosis, while others can be redundant among each other. Thus, feature selection (FS) algorithms are often considered to find relevant/non-redundant features.& para;& para;Objective: This study aimed to evaluate the relevance of FS approaches applied to Alzheimer's Disease (AD) EEG-based diagnosis and compare the selected features with previous clinical findings.& para;& para;Methods: Eight different FS algorithms were applied to EEG spectral measures from 22 AD patients and 12 healthy age-matched controls. The FS contribution was evaluated by considering the leave-one-subject-out accuracy of Support Vector Machine classifiers built in the datasets described by the selected features.& para;& para;Results: The Filtered Subset Evaluator technique achieved the best performance improvement both on a per-patient basis (91.18% of accuracy) and on a per-epoch basis (85.29 +/- 21.62%), after removing 88.76 +/- 1.12% of the original features. All algorithms found out that alpha and beta bands are relevant features, which is in agreement with previous findings from the literature.& para;& para;Conclusion: Biologically plausible EEG datasets could achieve improved accuracies with pre-processing FS steps.& para;& para;Significance: The results suggest that the FS and classification techniques are an attractive complementary tool in order to reveal potential biomarkers aiding the AD clinical diagnosis. (C) 2017 Published by Elsevier Ireland Ltd on behalf of International Federation of Clinical Neurophysiology.
- ItemSomente MetadadadosPattern-reversal visual evoked potentials as a diagnostic tool for ocular malingering(Consel Brasil Oftalmologia, 2016) Soares, Tarciana de Souza [UNIFESP]; Sacai, Paula Yuri [UNIFESP]; Berezovsky, Adriana [UNIFESP]; Rocha, Daniel Martins [UNIFESP]; Watanabe, Sung Eun Song [UNIFESP]; Salomao, Solange Rios [UNIFESP]Purpose: To investigate the contributions of transient pattern-reversal visual evoked potentials in the diagnosis of ocular malingering at a Brazilian university hospital. Methods: Adult patients with suspected malingering in one or both eyes were referred for visual evoked potential testing. Data from patients' medical records were reviewed and analyzed retrospectively. Data analysis included the distance optotype visual acuity based on a ETDRS retro-illuminated chart and the transient pattern-reversal visual evoked potential parameters of latency (milliseconds) and amplitude (microvolts) for the P100 component, using checkerboards with visual subtenses of 15' and 60'. Motivations for malingering were noted. Results: The 20 subjects included 11 (55%) women. Patient ages ranged from 21 to 61 years (mean = 45.05 +/- 11.76 years
- ItemSomente MetadadadosStacking machine learning classifiers to identify Higgs bosons at the LHC(Iop Publishing Ltd, 2017) Alves, A. [UNIFESP]Machine learning (ML) algorithms have been employed in the problem of classifying signal and background events with high accuracy in particle physics. In this paper, we compare the performance of a widespread ML technique, namely, stacked generalization, against the results of two state-of-art algorithms: (1) a deep neural network (DNN) in the task of discovering a new neutral Higgs boson and (2) a scalable machine learning system for tree boosting, in the Standard Model Higgs to tau leptons channel, both at the 8 TeV LHC. In a cut-and-count analysis, stacking three algorithms performed around 16% worse than DNN but demanding far less computation efforts, however, the same stacking outperforms boosted decision trees. Using the stacked classifiers in a multivariate statistical analysis (MVA), on the other hand, significantly enhances the statistical significance compared to cut-and-count in both Higgs processes, suggesting that combining an ensemble of simpler and faster ML algorithms with MVA tools is a better approach than building a complex state-of-art algorithm for cut-and-count.