RESUMO

This study aimed to determine the feasibility of applying machine-learning methods to assess the progression of chronic kidney disease (CKD) in patients with coronavirus disease (COVID-19) and acute renal injury (AKI). The study was conducted on patients aged 18 years or older who were diagnosed with COVID-19 and AKI between April 2020 and March 2021, and admitted to a second-level hospital in Mérida, Yucatán, México. Of the admitted patients, 47.92% died and 52.06% were discharged. Among the discharged patients, 176 developed AKI during hospitalization, and 131 agreed to participate in the study. The study's results indicated that the area under the receiver operating characteristic curve (AUC-ROC) for the four models was 0.826 for the support vector machine (SVM), 0.828 for the random forest, 0.840 for the logistic regression, and 0.841 for the boosting model. Variable selection methods were utilized to enhance the performance of the classifier, with the SVM model demonstrating the best overall performance, achieving a classification rate of 99.8% ± 0.1 in the training set and 98.43% ± 1.79 in the validation set in AUC-ROC values. These findings have the potential to aid in the early detection and management of CKD, a complication of AKI resulting from COVID-19. Further research is required to confirm these results.

RESUMO

Objective: To develop a natural language processing application capable of automatically identifying benign gallbladder diseases that require surgery, from radiology reports. Materials and Methods: We developed a text classifier to classify reports as describing benign diseases of the gallbladder that do or do not require surgery. We randomly selected 1,200 reports describing the gallbladder from our database, including different modalities. Four radiologists classified the reports as describing benign disease that should or should not be treated surgically. Two deep learning architectures were trained for classification: a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) network. In order to represent words in vector form, the models included a Word2Vec representation, with dimensions of 300 or 1,000. The models were trained and evaluated by dividing the dataset into training, validation, and subsets (80/10/10). Results: The CNN and BiLSTM performed well in both dimensional spaces. For the 300- and 1,000-dimensional spaces, respectively, the F1-scores were 0.95945 and 0.95302 for the CNN model, compared with 0.96732 and 0.96732 for the BiLSTM model. Conclusion: Our models achieved high performance, regardless of the architecture and dimensional space employed.

Objetivo: Desenvolver uma aplicação de processamento de linguagem natural capaz de identificar automaticamente doenças cirúrgicas benignas da vesícula biliar a partir de laudos radiológicos. Materiais e Métodos: Desenvolvemos um classificador de texto para classificar laudos como contendo ou não doenças cirúrgicas benignas da vesícula biliar. Selecionamos aleatoriamente 1.200 laudos com descrição da vesícula biliar de nosso banco de dados, incluindo diferentes modalidades. Quatro radiologistas classificaram os laudos como doença benigna cirúrgica ou não. Duas arquiteturas de aprendizagem profunda foram treinadas para a classificação: a rede neural convolucional (convolutional neural network - CNN) e a memória longa de curto prazo bidirecional (bidirectional long short-term memory - BiLSTM). Para representar palavras de forma vetorial, os modelos incluíram uma representação Word2Vec, com dimensões variando de 300 a 1000. Os modelos foram treinados e avaliados por meio da divisão do conjunto de dados entre treinamento, validação e teste (80/10/10). Resultados: CNN e BiLSTM tiveram bom desempenho em ambos os espaços dimensionais. Relatamos para 300 e 1000 dimensões, respectivamente, as pontuações F1 de 0,95945 e 0,95302 para o modelo CNN e de 0,96732 e 0,96732 para a BiLSTM. Conclusão: Nossos modelos alcançaram alto desempenho, independentemente de diferentes arquiteturas e espaços dimensionais.

RESUMO

Traditional monitoring of asian soybean rust severity is a time- and labor-intensive task, as it requires visual assessments by skilled professionals in the field. Thus, the use of remote sensing and machine learning (ML) techniques in data processing has emerged as an approach that can increase efficiency in disease monitoring, enabling faster, more accurate and time- and labor-saving evaluations. The aims of the study were: (i) to identify the spectral signature of different levels of Asian soybean rust severity; (ii) to identify the most accurate machine learning algorithm for classifying disease severity levels; (iii) which spectral input provides the highest classification accuracy for the algorithms; (iv) to determine a sample size of leaves that guarantees the best accuracy for the algorithms. A field experiment was carried out in the 2022/2023 harvest in a randomized block design with a 6x3 factorial scheme (ML algorithms x severity levels) and four replications. Disease severity levels assessed were: healthy leaves, 25 % severity, and 50 % severity. Leaf hyperspectral analysis was carried out over a wide range from 350 to 2500 nm. From this analysis, 28 spectral bands were extracted, seeking to distinguish the spectral signature for each severity level with the least input dataset. Data was subjected to machine learning analysis using Artificial Neural Network (ANN), REPTree (DT) and J48 decision trees, Random Forest (RF), and Support Vector Machine (SVM) algorithms, as well as a traditional classification method (Logistic Regression - LR). Two different input datasets were tested for each algorithm: the full spectrum (ALL) provided by the sensor and the 28 spectral bands (SB). Tests with different sample sizes were also conducted to investigate the algorithms' ability to detect severity levels with a reduced sample size. Our findings indicate differences between the spectral curves for the severity levels assessed, which makes it possible to differentiate between healthy plants with low and high severity using hyperspectral sensing. SVM was the most accurate algorithm for classifying severity levels by using all the spectral information as input. This algorithm also provided high classification accuracy when using smaller leaf samples. This study reveals that hyperspectral sensing and the use of ML algorithms provide an accurate classification of different levels of Asian rust severity, and can be powerful tools for a more efficient disease monitoring process.

Assuntos

RESUMO

Abstract Objective: To develop a natural language processing application capable of automatically identifying benign gallbladder diseases that require surgery, from radiology reports. Materials and Methods: We developed a text classifier to classify reports as describing benign diseases of the gallbladder that do or do not require surgery. We randomly selected 1,200 reports describing the gallbladder from our database, including different modalities. Four radiologists classified the reports as describing benign disease that should or should not be treated surgically. Two deep learning architectures were trained for classification: a convolutional neural network (CNN) and a bidirectional long short-term memory (BiLSTM) network. In order to represent words in vector form, the models included a Word2Vec representation, with dimensions of 300 or 1,000. The models were trained and evaluated by dividing the dataset into training, validation, and subsets (80/10/10). Results: The CNN and BiLSTM performed well in both dimensional spaces. For the 300- and 1,000-dimensional spaces, respectively, the F1-scores were 0.95945 and 0.95302 for the CNN model, compared with 0.96732 and 0.96732 for the BiLSTM model. Conclusion: Our models achieved high performance, regardless of the architecture and dimensional space employed.

Resumo Objetivo: Desenvolver uma aplicação de processamento de linguagem natural capaz de identificar automaticamente doenças cirúrgicas benignas da vesícula biliar a partir de laudos radiológicos. Materiais e Métodos: Desenvolvemos um classificador de texto para classificar laudos como contendo ou não doenças cirúrgicas benignas da vesícula biliar. Selecionamos aleatoriamente 1.200 laudos com descrição da vesícula biliar de nosso banco de dados, incluindo diferentes modalidades. Quatro radiologistas classificaram os laudos como doença benigna cirúrgica ou não. Duas arquiteturas de aprendizagem profunda foram treinadas para a classificação: a rede neural convolucional (convolutional neural network - CNN) e a memória longa de curto prazo bidirecional (bidirectional long short-term memory - BiLSTM). Para representar palavras de forma vetorial, os modelos incluíram uma representação Word2Vec, com dimensões variando de 300 a 1000. Os modelos foram treinados e avaliados por meio da divisão do conjunto de dados entre treinamento, validação e teste (80/10/10). Resultados: CNN e BiLSTM tiveram bom desempenho em ambos os espaços dimensionais. Relatamos para 300 e 1000 dimensões, respectivamente, as pontuações F1 de 0,95945 e 0,95302 para o modelo CNN e de 0,96732 e 0,96732 para a BiLSTM. Conclusão: Nossos modelos alcançaram alto desempenho, independentemente de diferentes arquiteturas e espaços dimensionais.

RESUMO

Given rising traders and consumers concerns, the global food industry is increasingly demanding authentic and traceable products. Consequently, there is a heightened focus on verifying geographical authenticity as food quality assurance. In this work, we assessed pattern recognition approaches based on elemental predictors to discern the provenance of mandarin juices from three distinct citrus-producing zones located in the Northeast region of Argentina. A total of 202 samples originating from two cultivars were prepared through microwave-assisted acid digestion and analyzed by microwave plasma atomic emission spectroscopy (MP-AES). Later, we applied linear discriminant analysis (LDA), k-nearest neighbor (k-NN), support vector machine (SVM), and random forest (RF) to the element data obtained. SVM accomplished the best classification performance with a 95.1% success rate, for which it was selected for citrus samples authentication. The proposed method highlights the capability of mineral profiles in accurately identifying the genuine origin of mandarin juices. By implementing this model in the food supply chain, it can prevent mislabeling fraud, thereby contributing to consumer protection.

RESUMO

By observing the actions taken by operators, it is possible to determine the risk level of a work task. One method for achieving this is the recognition of human activity using biosignals and inertial measurements provided to a machine learning algorithm performing such recognition. The aim of this research is to propose a method to automatically recognize physical exertion and reduce noise as much as possible towards the automation of the Job Strain Index (JSI) assessment by using a motion capture wearable device (MindRove armband) and training a quadratic support vector machine (QSVM) model, which is responsible for predicting the exertion depending on the patterns identified. The highest accuracy of the QSVM model was 95.7%, which was achieved by filtering the data, removing outliers and offsets, and performing zero calibration; in addition, EMG signals were normalized. It was determined that, given the job strain index's purpose, physical exertion detection is crucial to computing its intensity in future work.

Assuntos

RESUMO

Background: Identifying species, particularly small metazoans, remains a daunting challenge and the phylum Nematoda is no exception. Typically, nematode species are differentiated based on morphometry and the presence or absence of certain characters. However, recent advances in artificial intelligence, particularly machine learning (ML) algorithms, offer promising solutions for automating species identification, mostly in taxonomically complex groups. By training ML models with extensive datasets of accurately identified specimens, the models can learn to recognize patterns in nematodes' morphological and morphometric features. This enables them to make precise identifications of newly encountered individuals. Implementing ML algorithms can improve the speed and accuracy of species identification and allow researchers to efficiently process vast amounts of data. Furthermore, it empowers non-taxonomists to make reliable identifications. The objective of this study is to evaluate the performance of ML algorithms in identifying species of free-living marine nematodes, focusing on two well-known genera: Acantholaimus Allgén, 1933 and Sabatieria Rouville, 1903. Methods: A total of 40 species of Acantholaimus and 60 species of Sabatieria were considered. The measurements and identifications were obtained from the original publications of species for both genera, this compilation included information regarding the presence or absence of specific characters, as well as morphometric data. To assess the performance of the species identification four ML algorithms were employed: Random Forest (RF), Stochastic Gradient Boosting (SGBoost), Support Vector Machine (SVM) with both linear and radial kernels, and K-nearest neighbor (KNN) algorithms. Results: For both genera, the random forest (RF) algorithm demonstrated the highest accuracy in correctly classifying specimens into their respective species, achieving an accuracy rate of 93% for Acantholaimus and 100% for Sabatieria, only a single individual from Acantholaimus of the test data was misclassified. Conclusion: These results highlight the overall effectiveness of ML algorithms in species identification. Moreover, it demonstrates that the identification of marine nematodes can be automated, optimizing biodiversity and ecological studies, as well as turning species identification more accessible, efficient, and scalable. Ultimately it will contribute to our understanding and conservation of biodiversity.

Assuntos

RESUMO

Pisco is an alcoholic beverage obtained from grape juice distillation. Considered the flagship drink of Peru, it is produced following strict and specific quality standards. In this work, sensing results for volatile compounds in pisco, obtained with an electronic nose, were analyzed through the application of machine learning algorithms for the differentiation of pisco varieties. This differentiation aids in verifying beverage quality, considering the parameters established in its Designation of Origin". For signal processing, neural networks, multiclass support vector machines and random forest machine learning algorithms were implemented in MATLAB. In addition, data augmentation was performed using a proposed procedure based on interpolation-extrapolation. All algorithms trained with augmented data showed an increase in performance and more reliable predictions compared to those trained with raw data. From the comparison of these results, it was found that the best performance was achieved with neural networks.

Assuntos

RESUMO

BACKGROUND AND AIMS: The actual classification of breast tumors in subtypes represents an attempt to stratify patients into clinically cohesive groups, nevertheless, clinicians still lack reproducible and reliable protein biomarkers for breast cancer subtype discrimination. In this study, we aimed to access the differentially expressed proteins between these tumors and its biological implications, contributing to the subtype's biological and clinical characterization, and with protein panels for subtype discrimination. METHODS: In our study, we applied high-throughput mass spectrometry, bioinformatic, and machine learning approaches to investigate the proteome of different breast cancer subtypes. RESULTS: We identified that each subtype depends on different protein expression patterns to sustain its malignancy, and also alterations in pathways and processes that can be associated with each subtype and its biological and clinical behaviors. Regarding subtype biomarkers, our panels achieved performances with at least 75% of sensibility and 92% of specificity. In the validation cohort, the panels obtained acceptable to outstanding performances (AUC = 0.740 to 1.00). CONCLUSIONS: In general, our results expand the accuracy of breast cancer subtypes' proteomic landscape and improve the understanding of its biological heterogeneity. In addition, we identified potential protein biomarkers for the stratification of breast cancer patients, improving the repertoire of reliable protein biomarkers. SIGNIFICANCE: Breast cancer is the most diagnosed cancer type worldwide and the most lethal cancer in women. As a heterogeneous disease, breast cancer tumors can be classified into four major subtypes, each presenting particular molecular alterations, clinical behaviors, and treatment responses. Thus, a pivotal step in patient management and clinical decisions is accurately classifying breast tumor subtypes. Currently, this classification is made by the immunohistochemical detection of four classical markers (estrogen receptor, progesterone receptor, HER2 receptor, and the Ki-67 index); however, it is known that these markers alone do not fully discriminate the breast tumor subtypes. Also, the poor understanding of the molecular alterations of each subtype leads to a challenging decision-making process regarding treatment choice and prognostic determination. This study, through high-throughput label-free mass-spectrometry data acquisition and downstream bioinformatic analysis, advances in the proteomic discrimination of breast tumors and achieves an in-depth characterization of the subtype's proteomes. Here, we indicate how the variations in the subtype's proteome can influence the tumor's biological and clinical differences, highlighting the variation in the expression pattern of oncoproteins and tumor suppressor proteins between subtypes. Also, through our machine-learning approach, we propose multi-protein panels with the potential to discriminate the breast cancer subtypes. Our panels achieved high classification performance in our cohort and in the independent validation cohort, demonstrating their potential to improve the current tumor discrimination system as complements to the classical immunohistochemical classification.

Assuntos

RESUMO

The blood diagnosis of diabetes mellitus (DM) is highly accurate; however, it is an invasive, high-cost, and painful procedure. In this context, the combination of ATR-FTIR spectroscopy and machine learning techniques in other biological samples has been used as an alternative tool to develop a non-invasive, fast, inexpensive, and label-free diagnostic or screening platform for several diseases, including DM. In this study, we used the ATR-FTIR tool associated with linear discriminant analysis (LDA) and a support vector machine (SVM) classifier in order to identify changes in salivary components to be used as alternative biomarkers for the diagnosis of type 2 DM. The band area values of 2962 cm-1, 1641 cm-1, and 1073 cm-1 were higher in type 2 diabetic patients than in non-diabetic subjects. The best classification of salivary infrared spectra was by SVM, showing a sensitivity of 93.3% (42/45), specificity of 74% (17/23), and accuracy of 87% between non-diabetic subjects and uncontrolled type 2 DM patients. The SHAP features of infrared spectra indicate the main salivary vibrational modes of lipids and proteins that are responsible for discriminating DM patients. In summary, these data highlight the potential of ATR-FTIR platforms coupled with machine learning as a reagent-free, non-invasive, and highly sensitive tool for screening and monitoring diabetic patients.

RESUMO

The sanitary quality of seed is essential in agriculture. This is because pathogenic fungi compromise seed physiological quality and prevent the formation of plants in the field, which causes losses to farmers. Multispectral images technologies coupled with machine learning algorithms can optimize the identification of healthy peanut seeds, greatly improving the sanitary quality. The objective was to verify whether multispectral images technologies and artificial intelligence tools are effective for discriminating pathogenic fungi in tropical peanut seeds. For this purpose, dry peanut seeds infected by fungi (A. flavus, A. niger, Penicillium sp., and Rhizopus sp.) were used to acquire images at different wavelengths (365 to 970 nm). Multispectral markers of peanut seed health quality were found. The incubation period of 216 h was the one that most contributed to discriminating healthy seeds from those containing fungi through multispectral images. Texture (Percent Run), color (CIELab L*) and reflectance (490 nm) were highly effective in discriminating the sanitary quality of peanut seeds. Machine learning algorithms (LDA, MLP, RF, and SVM) demonstrated high accuracy in autonomous detection of seed health status (90 to 100%). Thus, multispectral images coupled with machine learning algorithms are effective for screening peanut seeds with superior sanitary quality.

RESUMO

Motor imagery brain-computer interface (MI-BCI) is one of the most used paradigms in EEG-based brain-computer interface (BCI). The current state-of-the-art in BCI involves tuning classifiers to subject-specific training data, acquired over several sessions, in order to perform calibration prior to actual use of the so-called subject-specific BCI system (SS-BCI). Herein, the goal is to provide a ready-to-use system requiring minimal effort for setup. Thus, our challenge was to design a subject-independent BCI (SI-BCI) to be used by any new user without the constraint of individual calibration. Outcomes from other studies with the same purpose were used to undertake comparisons and validate our findings. For the EEG signal processing, we used a combination of the delta (0.5-4 Hz), alpha (8-13 Hz), and beta+gamma (13-40 Hz) bands at a stage prior to feature extraction. Next, we extracted features from the 27-channel EEG using common spatial pattern (CSP) and performed binary classification (MI of right- and left-hand) with linear discriminant analysis (LDA) and support vector machine (SVM) classifiers. These analyses were done for both the SS-BCI and SI-BCI models. We employed "leave-one-subject-out" (LOSO) arrangement and 10-fold cross-validation to evaluate our SI-BCI and SS-BCI systems, respectively. Compared with other two studies, our work was the only one that showed higher accuracy for the LDA classifier in SI-BCI as compared to SS-BCI. On the other hand, LDA accuracy was lower than accuracy achieved with SVM in both conditions (SI-BCI and SS-BCI). Our SS-BCI accuracy reached 76.85% using LDA and 94.20% using SVM and for SI-BCI we got 80.30% with LDA and 83.23% with SVM. We conclude that SI-BCI may be a feasible and relevant option, which can be used in scenarios where subjects are not able to submit themselves to long training sessions or to fast evaluation of the so called "BCI illiteracy." Comparatively, our strategy proved to be more efficient, giving us the best result for SI-BCI when faced against the classification performances of other three studies, even considering the caveat that different datasets were used in the comparison of the four studies.

Assuntos

RESUMO

Several approaches to assess the authenticity of food products have been developed, given that fraudulent products may impact consumers' confidence, affect commercial trades and lead to health risks. This paper proposes an approach to identify the chemical elements that optimally discriminate rice samples according to their producing region in the South of Brazil, the largest rice producer outside Asia. A combinatorial procedure on the concentration of 26 elements determined using inductively coupled mass spectrometry (ICP-MS) and liquid chromatography hyphenated with ICP-MS from 640 rice samples was coupled with Support Vector Machine. The assessed elements included nonmetal and metal elements of 3 types of rice collected from 5 rice-producing regions. The framework selected Mn, Fe, Cu, Zn, Ni, Mo, Cd, Cs, As, Rb, Se, and iAs as the most informative elements for tracking samples' origin. The concentration of such elements is strongly affected by fertilization procedures and soil composition.

Assuntos

RESUMO

Major depressive disorder (MDD) is characterized by a series of disabling symptoms like anhedonia, depressed mood, lack of motivation for daily tasks and self-extermination thoughts. The monoamine deficiency hypothesis states that depression is mainly caused by a deficiency of monoamine at the synaptic cleft. Thus, major efforts have been made to develop drugs that inhibit serotonin (SERT), norepinephrine (NET) and dopamine (DAT) transporters and increase the availability of these monoamines. Current gold standard treatment of MDD uses drugs that target one or more monoamine transporters. Triple reuptake inhibitors (TRIs) can target SERT, NET, and DAT simultaneously, and are believed to have the potential to be early onset antidepressants. Quantitative structure-activity relationship models were developed using machine learning algorithms in order to predict biological activities of a series of triple reuptake inhibitor compounds that showed in vitro inhibitory activity against multiple targets. The results, using mostly interpretable descriptors, showed that the internal and external predictive ability of the models are adequate, particularly of the DAT and NET by Random Forest and Support Vector Machine models. The current work shows that models developed from relatively simple, chemically interpretable descriptors can predict the activity of TRIs with similar structure in the applicability domain using ML methods.Communicated by Ramaswamy H. Sarma.

Assuntos

RESUMO

This letter aims to reply to Bratchenko and Bratchenko's comment on our paper "Feasibility of Raman spectroscopy as a potential in vivo tool to screen for pre-diabetes and diabetes." Our paper analyzed the feasibility of using in vivo Raman measurements combined with machine learning techniques to screen diabetic and prediabetic patients. We argued that this approach yields high overall accuracy (94.3%) while retaining a good capacity to distinguish between diabetic (area under the receiver-operating curve [AUC] = 0.86) and control classes (AUC = 0.97) and a moderate performance for the prediabetic class (AUC = 0.76). Bratchenko and Bratchenko's comment focuses on the possible overestimation of the proposed classification models and the absence of information on the age of participants. In this reply, we address their main concerns regarding our previous manuscript.

Assuntos

RESUMO

ABSTRACT: The extension of the area occupied by the inter tussock stratum and tussock stratum in natural pastures is essential for the productive performance of grazing animals. Images obtained from unmanned remote sensors can provide useful information, especially because they have a high spatial resolution. Thus, this study evaluated the performance of the supervised adaptive classification applied to aerial images obtained from an onboard drone camera to map the area covered by tussocks in a natural pasture of the Pampa biome. The study was carried out in a natural pasture area managed since 1986 under different forage allowances, considering treatments of 8, 12, and 16 kg of dry matter per 100 kg live weight (% LW). An aerial image from September 2017, obtained with a Canon S100 camera onboard a drone at an altitude of 120 m, with a spatial resolution of 5 cm, was used. The random forest and support vector machine classifiers were tested associated with specific classification rules. False-color images showed considerable visual similarity in the large patterns of the vegetation distribution and the validation performed with independent samples when compared to the classified images. The tested classifiers were able to measure the area covered by the tussock stratum, which could be an indicator of the quality vegetation in a natural grassland of the Pampa biome.

RESUMO: A quantidade de área ocupada por estrato inferior e superior em pastagens naturais tem grande importância sobre o desempenho produtivo dos animais em pastejo. Imagens obtidas de sensores remotos não tripulados podem fornecer informações úteis, especialmente por possuírem alta resolução espacial. O objetivo deste trabalho foi avaliar o desempenho de classificação supervisionada adaptativa aplicada a imagem aérea obtida por câmera a bordo de drone, no mapeamento da área coberta por touceiras em pastagem natural do bioma Pampa. O estudo foi realizado em área de pastagem natural, manejada desde 1986 sob diferentes ofertas de forragem, tendo sido considerados os tratamentos 8, 12 e 16 kg de matéria seca por 100 kg de peso vivo (% PV). Foi utilizada uma imagem aérea, de setembro de 2017, obtida com uma câmera Canon S100, a bordo de um drone a 120 m de altitude, correspondendo a resolução espacial de 5 cm. Foram testados dois classificadores, Random Forest e Support Vector Machine associados a regras específicas de classificação. As imagens de falsa cor, quando comparadas às imagens classificadas, apresentaram considerável semelhança visual nos grandes padrões de distribuição da vegetação, bem como na validação feita com amostras independentes. Os classificadores testados foram capazes de mensurar a área coberta por estrato superior, podendo ser um indicador da qualidade da vegetação, em pastagem natural do bioma Pampa.

RESUMO

The extension of the area occupied by the inter tussock stratum and tussock stratum in natural pastures is essential for the productive performance of grazing animals. Images obtained from unmanned remote sensors can provide useful information, especially because they have a high spatial resolution. Thus, this study evaluated the performance of the supervised adaptive classification applied to aerial images obtained from an onboard drone camera to map the area covered by tussocks in a natural pasture of the Pampa biome. The study was carried out in a natural pasture area managed since 1986 under different forage allowances, considering treatments of 8, 12, and 16 kg of dry matter per 100 kg live weight (% LW). An aerial image from September 2017, obtained with a Canon S100 camera onboard a drone at an altitude of 120 m, with a spatial resolution of 5 cm, was used. The random forest and support vector machine classifiers were tested associated with specific classification rules. False-color images showed considerable visual similarity in the large patterns of the vegetation distribution and the validation performed with independent samples when compared to the classified images. The tested classifiers were able to measure the area covered by the tussock stratum, which could be an indicator of the quality vegetation in a natural grassland of the Pampa biome.

A quantidade de área ocupada por estrato inferior e superior em pastagens naturais tem grande importância sobre o desempenho produtivo dos animais em pastejo. Imagens obtidas de sensores remotos não tripulados podem fornecer informações úteis, especialmente por possuírem alta resolução espacial. O objetivo deste trabalho foi avaliar o desempenho de classificação supervisionada adaptativa aplicada a imagem aérea obtida por câmera a bordo de drone, no mapeamento da área coberta por touceiras em pastagem natural do bioma Pampa. O estudo foi realizado em área de pastagem natural, manejada desde 1986 sob diferentes ofertas de forragem, tendo sido considerados os tratamentos 8, 12 e 16 kg de matéria seca por 100 kg de peso vivo (% PV). Foi utilizada uma imagem aérea, de setembro de 2017, obtida com uma câmera Canon S100, a bordo de um drone a 120 m de altitude, correspondendo a resolução espacial de 5 cm. Foram testados dois classificadores, Random Forest e Support Vector Machine associados a regras específicas de classificação. As imagens de falsa cor, quando comparadas às imagens classificadas, apresentaram considerável semelhança visual nos grandes padrões de distribuição da vegetação, bem como na validação feita com amostras independentes. Os classificadores testados foram capazes de mensurar a área coberta por estrato superior, podendo ser um indicador da qualidade da vegetação, em pastagem natural do bioma Pampa.

Assuntos

RESUMO

Electromyography (EMG) processing is a fundamental part of medical research. It offers the possibility of developing new devices and techniques for the diagnosis, treatment, care, and rehabilitation of patients, in most cases non-invasively. However, EMG signals are random, non-stationary, and non-linear, making their classification difficult. Due to this, it is of vital importance to define which factors are helpful for the classification process. In order to improve this process, it is possible to apply algorithms capable of identifying which features are most important in the categorization process. Algorithms based on metaheuristic methods have demonstrated an ability to search for suitable subsets of features for optimization problems. Therefore, this work proposes a methodology based on genetic algorithms for feature selection to find the parameter space that offers the slightest classification error in 250 ms signal segments. For classification, a support vector machine is used. For this work, two databases were used, the first corresponding to the right upper extremity and the second formed by movements of the right lower extremity. For both databases, a feature space reduction of over 65% was obtained, with a higher average classification efficiency of 91% for the best subset of parameters. In addition, particle swarm optimization (PSO) was applied based on right upper extremity data, obtaining an 88% average error and a 46% reduction for the best subset of parameters. Finally, a sensitivity analysis was applied to the characteristics selected by PSO and genetic algorithms for the database of the right upper extremity, obtaining that the parameters determined by the genetic algorithms show greater sensitivity for the classification process.

RESUMO

An accurate model of a proton-exchange membrane fuel cell (PEMFC) is important for understanding this fuel cell's dynamic process and behavior. Among different large-scale energy storage systems, fuel cell technology does not have geographical requirements. To provide an effective operation estimation of PEMFC, this paper proposes a support vector machine (SVM) based model. The advantages of the SVM, such as the ability to model nonlinear systems and provide accurate estimations when nonlinearities and noise appear in the system, are the main motivations to use the SVM method. This model can capture the static and dynamic voltage-current characteristics of the PEMFC system in the three operating regions. The validity of the proposed SVM model has been verified by comparing the estimated voltage with the real measurements from the Ballard Nexa® 1.2 kW fuel cell (FC) power module. The obtained results have shown high accuracy between the proposed model and the experimental operation of the PEMFC. A statistical study is developed to evaluate the effectiveness and superiority of the proposed SVM model compared with the diffusive global (DG) model and the evolution strategy (ES)-based model.

RESUMO

Most clinical neurofeedback studies based on functional magnetic resonance imaging use the patient's own neural activity as feedback. The objective of this study was to create a subject-independent brain state classifier as part of a real-time fMRI neurofeedback (rt-fMRI NF) system that can guide patients with depression in achieving a healthy brain state, and then to examine subsequent clinical changes. In a first step, a brain classifier based on a support vector machine (SVM) was trained from the neural information of happy autobiographical imagery and motor imagery blocks received from a healthy female participant during an MRI session. In the second step, 7 right-handed female patients with mild or moderate depressive symptoms were trained to match their own neural activity with the neural activity corresponding to the "happiness emotional brain state" of the healthy participant. The training (4 training sessions over 2 weeks) was carried out using the rt-fMRI NF system guided by the brain-state classifier we had created. Thus, the informative voxels previously obtained in the first step, using SVM classification and Effect Mapping, were used to classify the Blood-Oxygen-Level Dependent (BOLD) activity of the patients and converted into real-time visual feedback during the neurofeedback training runs. Improvements in the classifier accuracy toward the end of the training were observed in all the patients [Session 4-1 Median = 6.563%; Range = 4.10-27.34; Wilcoxon Test (0), 2-tailed p = 0.031]. Clinical improvement also was observed in a blind standardized clinical evaluation [HDRS CE2-1 Median = 7; Range 2 to 15; Wilcoxon Test (0), 2-tailed p = 0.016], and in self-report assessments [BDI-II CE2-1 Median = 8; Range 1-15; Wilcoxon Test (0), 2-tailed p = 0.031]. In addition, the clinical improvement was still present 10 days after the intervention [BDI-II CE3-2_Median = 0; Range -1 to 2; Wilcoxon Test (0), 2-tailed p = 0.50/ HDRS CE3-2 Median = 0; Range -1 to 2; Wilcoxon Test (0), 2-tailed p = 0.625]. Although the number of participants needs to be increased and a control group included to confirm these findings, the results suggest a novel option for neural modulation and clinical alleviation in depression using noninvasive stimulation technologies.