Abstract This article explores the design and application of machine learning techniques to enhance traditional approaches for solving NP-hard optimization problems. Specifically, it focuses on the Last-Mile Routing Research Challenge (LMRRC), supported by Amazon and MIT, which sought innovative solutions for cargo routing optimization. While the challenge provided travel times and zone identifiers, the dependency on these factors raises concerns about the algorithms’ generalizability to different contexts and regions with standard delivery services registries. To address these concerns, this study proposes personalized cost matrices that incorporate both distance and time models, along with the relationships between delivery stops. Additionally, it presents an improved approach to sequencing stops by combining exact and approximate algorithms, utilizing a customized regression technique alongside fine-tuned metaheuristics and heuristics refinements. The resulting methodology achieves competitive scores on the LMRRC validation dataset, which comprises routes from the USA. By carefully delineating route characteristics, the study enables the selection of specific technique combinations for each route, considering its geometrical and geographical attributes. Furthermore, the proposed methodologies are successfully applied to real-case scenarios of last-mile deliveries in Montevideo (Uruguay), demonstrating similar average scores and accuracy on new testing routes. This research contributes to the advancement of last-mile delivery optimization by leveraging personalized cost matrices and algorithmic refinements. The findings highlight the potential for improving existing approaches and their adaptability to diverse geographic contexts, paving the way for more efficient and effective delivery services in the future.

Resumen Este artículo explora el diseño y aplicación de técnicas de aprendizaje automático para mejorar los enfoques tradicionales y así resolver problemas de optimización NP-hard. En particular, se enfoca en el Last-Mile Routing Research Challenge (LMRRC), apoyado por Amazon y MIT, que buscaba soluciones innovadoras para la optimización de rutas de carga. Si bien el desafío proporcionó tiempos de viaje e identificadores de zona, la dependencia de estos factores plantea preocupaciones sobre la generalización de los algoritmos a diferentes contextos y regiones con registros de servicios de entrega estándar. Para abordar estas interrogantes, este estudio propone matrices de costos personalizadas que incorporan modelos de distancia y tiempo, junto con las relaciones entre las paradas de entrega. Además, presenta enfoques mejorados para la secuenciación de paradas mediante la combinación de algoritmos exactos y aproximados, utilizando técnicas de regresión personalizada junto con metaheurísticas y refinamientos heurísticos ajustados. La metodología resultante logra puntajes competitivos en el conjunto de datos de validación LMRRC, que usa rutas de EE. UU. Al delinear cuidadosamente las características de la ruta, el estudio permite la selección de combinaciones de técnicas específicas para cada ruta, considerando sus atributos geométricos y geográficos. Además, las metodologías propuestas se aplican con éxito a escenarios de casos reales de entregas en Montevideo (Uruguay), demostrando puntajes promedio y precisión similares en nuevas rutas de prueba. Esta investigación contribuye al avance de la optimización de la entrega de última milla al aprovechar matrices de costos personalizadas y refinamientos algorítmicos. Los hallazgos resaltan el potencial para mejorar los enfoques existentes y su adaptabilidad a diversos contextos geográficos, allanando el camino para servicios de entrega más eficientes y efectivos en el futuro.

Abstract Introduction : Autism Spectrum Disorder (ASD) is a neurodevelopmental condition which traditional as sessment procedures encounter certain limitations. The current ASD research field is exploring and endorsing innovative methods to assess the disorder early on, based on the automatic detection of biomarkers. How ever, many of these procedures lack ecological validity in their measurements. In this context, virtual reality (VR) shows promise for objectively recording biosignals while users experience ecological situations. Methods : This study outlines a novel and playful VR procedure for the early assessment of ASD, relying on multimodal biosignal recording. During a VR experience featuring 12 virtual scenes, eye gaze, motor skills, elec trodermal activity and behavioural performance were measured in 39 children with ASD and 42 control peers. Machine learning models were developed to identify digital biomarkers and classify autism. Results : Biosignals reported varied performance in detecting ASD, while the combined model resulting from the combination of specific-biosignal models demon strated the ability to identify ASD with an accuracy of 83% (SD = 3%) and an AUC of 0.91 (SD = 0.04). Discussion : This screening tool may support ASD diagnosis by reinforcing the outcomes of traditional assessment procedures.

Resumen Introducción : El Trastorno del Espectro Autista (TEA) es un trastorno del neurodesarrollo, y sus procedimien tos tradicionales de evaluación encuentran ciertas li mitaciones. El actual campo de investigación sobre TEA está explorando y respaldando métodos innovadores para evaluar el trastorno tempranamente, basándose en la detección automática de biomarcadores. Sin embargo, muchos de estos procedimientos carecen de validez ecológica en sus mediciones. En este contexto, la reali dad virtual (RV) presenta un prometedor potencial para registrar objetivamente bioseñales mientras los usuarios experimentan situaciones ecológicas. Métodos : Este estudio describe un novedoso y lúdi co procedimiento de RV para la evaluación temprana del TEA, basado en la grabación multimodal de bio señales. Durante una experiencia de RV con 12 esce nas virtuales, se midieron la mirada, las habilidades motoras, la actividad electrodermal y el rendimiento conductual en 39 niños con TEA y 42 compañeros de control. Se desarrollaron modelos de aprendizaje automático para identificar biomarcadores digitales y clasificar el autismo. Resultados : Las bioseñales reportaron un rendimien to variado en la detección del TEA, mientras que el modelo resultante de la combinación de los modelos de las bioseñales demostró la capacidad de identificar el TEA con una precisión del 83% (DE = 3%) y un AUC de 0.91 (DE = 0.04). Discusión : Esta herramienta de detección pue de respaldar el diagnóstico del TEA al reforzar los resultados de los procedimientos tradicionales de evaluación.

Resumen: Las enfermedades mentales constituyen una de las principales causas de angustia en la vida de las personas a nivel individual, y repercuten en la salud y el bienestar de la sociedad. Para captar estas complejas asociaciones, las ciencias computacionales y la comunicación, a través del uso de métodos de procesamiento del lenguaje natural (NLP) en datos recolectados en redes sociales, han aportado prometedores avances para potenciar la atención sanitaria mental proactiva y ayudar al diagnóstico precoz. Por ello, se realizó una revisión sistemática de la literatura acerca de la detección de alteraciones mentales a través de redes sociales, mediante el uso de NLP en los últimos 5 años, que permitió identificar métodos, tendencias y orientaciones futuras, a través del análisis de 73 estudios, de 509 que arrojó la revisión de documentos extraídos de bases de datos científicas. El estudio reveló que, los fenómenos más comúnmente estudiados, correspondieron a Depresión e Ideación suicida, identificados a través del uso de algoritmos como el LIWC, CNN, LSTM, RF y SVM, en datos extraídos principalmente de Reddit y Twitter. Este estudio, finalmente proporciona algunas recomendaciones sobre las metodologías de NLP para la detección de enfermedades mentales, que pueden ser adoptadas en el ejercicio de profesionales interesados en la salud mental, y algunas reflexiones sobre el uso de estas tecnologías.

Abstract: Mental illness is a major cause of distress in people's lives at the individual level and impacts the health and well-being of society. To capture these complex associations, computational science and communication, through the use of natural language processing (NLP) methods on data collected in social networks, have provided promising advances to enhance proactive mental health care and aid in early diagnosis. Therefore, a systematic review of the literature on the detection of mental disorders through social networks, using NLP in the last 5 years, was carried out, which allowed identifying methods, trends and future directions, through the analysis of 73 studies, out of 509 that resulted from the review of documents extracted from scientific databases. The study revealed that the most commonly studied phenomena corresponded to Depression and Suicidal Ideation, identified through the use of algorithms such as LIWC, CNN, LSTM, RF and SVM on data extracted mainly from Reddit and Twitter. This study finally provides some recommendations on NLP methodologies for mental illness detection that can be adopted in the practice of professionals interested in mental health and some reflections on the use of these technologies.

A methodology for building pragmatic physics-based models is here adapted to predict the erosion of ladle linings in the steel industry, in order to support operators when deciding whether the lining can be used safely for another heat. A defective lining may allow 140 t of molten steel to spill, with disastrous consequences for workers and plant. The adopted work flow for the development, challenges faced, and some model results are presented. One key learning outcome is that model development should allow time for maturing the process understanding, as well as for many iterations by 'questions-responses and actions' at various stages in the model development. Good interaction between the development team and industry case owner is an important success factor. Combining or extending the model with the use of machine learning and cognition-related methods, such as knowledge graphs and self-adaptive algorithms, is discussed.

SYNOPSIS Ferrochrome (FeCr) is a vital ingredient in stainless steel production and is commonly produced by smelting chromite ores in submerged arc furnaces. Silicon (Si) is a componrnt of the FeCr alloy from the smelting process. Being both a contaminant and an indicator of the state of the process, its content needs to be kept within a narrow range. The complex chemistry of the smelting process and interactions between various factors make Si prediction by fundamental models infeasible. A data-driven approach offers an alternative by formulating the model based on historical data. This paper presents a systematic development of a data-driven model for predicting Si content. The model includes dimensionality reduction, regularized linear regression, and a boosting method to reduce the variability of the linear model residuals. It shows a good performance on testing data (R² = 0.63). The most significant predictors, as determined by linear model analysis and permutation testing, are previous Si content, carbon and titanium in the alloy, calcium oxide in the slag, resistance between electrodes, and electrode slips. Further analysis using thermodynamic data and models, links these predictors to electrode control and slag chemistry. This analysis lays the foundation for implementing Si content control on a ferrochrome smelter.

SYNOPSIS The large scale of cave mines leads to many challenges, including operational logistics and geomechanics design. In current practice, pillar stability assessment relies almost exclusively on stress analysis. However, stability is also affected by other factors including those related to operational aspects of the mining method, the effects of which are difficult to account for during the design stages. In this paper we present a case study of the application of a machine learning approach to evaluate the influence of these operational factors on pillar stability at the Chuquicamata underground cave mine in northern Chile. Due to the likely multi-factorial damage process leading to collapses and considering the different pillar conditions, a tree-based machine learning method was used and analysed to improve the understanding of the relative importance of the various contributing factors. Unlike stress analysis methods, it does not require any a priori knowledge of failure mechanisms, nor the calibration of associated controlling parameters. The proposed random forest model predicted pillar collapses with 80% accuracy despite limited samples to model from. The main contributing factors to collapses were found to be related to available pillar volume, cave front geometry, and time under abutment stress conditions. The effects and interactions of such factors were also studied, showing that careful and improved control over operational conditions can significantly reduce the likelihood of pillar collapses. These conclusions could not have been obtained from stress analysis alone, illustrating the complementary nature of conventional stress analysis and machine learning approaches.

Problem: Preterm birth is the leading cause of death and can result in significant long-term loss of physical and psychological capacity among survivors.Background: An estimated 15 million babies are born preterm every year. Prediction models based on machine learning methods have reported promising results.Aims: To identify risk factors associated with preterm birth and to develop and validate a prediction model for this outcome in a Brazilian birth cohort.Methods: Cross-sectional study of all births that occurred in Ribeirão Preto-SP and of one in three births that occurred in São Luís-MA, Brazil, in 2010. Questionnaires were applied to obtain pregnancy-related data. Multivariate adaptive regression splines were used to determine the independent variables. Preterm birth, defined as birth before 37 weeks gestational age, was the dependent variable. A random forest model was developed and its performance was evaluated by ROC analysis.Findings: The preterm birth rates were 12.7% (RP) and 14.1% (SL). The prediction and validation accuracies of the RF-based model were 91.3% and 85.5% respectively. The model can be applied starting in the third month of gestation and is more effective in identifying preterm infants with GA<31 weeks and 6 days (AUC=0.98).Discussion: It was possible to build a prediction model based on easily accessible low-cost data, without the need for complementary tests, providing results similar to those of other studies.Conclusions: Previous preterm birth and prenatal care were determinants. The use of an application for individualized patient monitoring an early stage can have positive effects on the quality of life of mother and child.

Abstract Soil moisture content can be used to predict drought impact on agricultural yield better than precipitation. Remote sensing is viable source of soil moisture data in instrument-scarce areas. However, space-based soil moisture estimates lack suitability for daily and high-resolution agricultural, hydrological, and environmental applications. This study aimed to assess the potential of the random forest machine learning technique to enhance the spatial resolution of remote soil moisture products from the SMAP satellite. Models were built using random forest for spatial downscaling of SMAP-L3-E, then visually and statistically evaluated for disaggregation quality. The impact of topography, soil properties, and precipitation on the downscaled soil moisture was examined. The relationship between downscaled soil moisture and in-situ soil moisture was analyzed. The results indicate that the proposed method demonstrated spatial and hydrological coherence, along with a satisfactory downscaling quality. Statistical validation indicated suitable generalization error for scientific and practical use (RMSE < 0.05 cm3 cm-3). Random forest effectively achieved spatial downscaling of SMAP-L3-E in the study area. Principal component and spatial analysis revealed dependence of downscaled soil moisture on elevation, soil organic carbon content, clay content, and saturated hydraulic conductivity, mainly under near-saturation conditions. Regarding validation against in-situ data, downscaled soil moisture explained in-situ soil moisture well under low soil water content (ρ = 0.624). Downscaling performance deteriorates for water contents between 0.40 to 0.50 cm3 cm-3, suggesting inadequacy under near saturation conditions at a daily temporal frequency. However, coarser temporal aggregations (7 to 10 days) yielded an average 0.98 correlation coefficient, regardless of saturation conditions. These results could potentially be applied in irrigation planning, soil physics studies and hydrology monitoring, to forecasting the occurrence of droughts, leaching of contaminants, surface runoff modeling, carbon cycle studies, soil's capacity to store and provide nutrients. Our results could mainly be applied to understanding the impact of droughts on crop yield.

Resumo O objetivo deste estudo é realizar uma análise de artigos científicos sobre inteligência artificial (IA) no contexto educacional. Esses artigos foram localizados utilizando o Google Scholar e o Science Direct e foram escritos em português, inglês e espanhol a partir de 2021. Em 2022, foi lançado o ChatGPT, uma tecnologia que faz parte do conceito chamado inteligência artificial generativa, criada a partir de técnicas de aprendizado de máquina. Esse tipo de ferramenta, que tem sido alvo de inúmeras discussões, também tem ganhado espaço nas instituições educacionais. Como resultado, governos de muitos países, preocupados com os impactos da IA, estão tentando regulamentar seu uso. Entre os artigos encontrados, destacam-se estudos sobre alfabetização em IA, formação do corpo docente, a necessidade de abordar o tema de forma interdisciplinar e começar desde o ensino fundamental.

Abstract The aim of this research is to analyze scientific articles about artificial intelligence (AI) in an educational context. These articles were located using Google Scholar and Science Direct and were written in Portuguese, English, and Spanish from 2021 onwards. In 2022, ChatGPT was launched, a technology that is part of the concept called generative AI, created by machine learning techniques. This type of tool, which has been the subject of many discussions, has also gained traction in educational institutions. As a result, governments in many countries, concerned about AI’s impacts, are attempting to regulate its use. Among these articles, studies about AI literacy, the training of teaching staff, the need to address the topic in an interdisciplinary manner, and starting from elementary school, stand out.

Resumen El objetivo de este estudio es realizar un recorrido analítico a través de artículos científicos sobre inteligencia artificial (IA) en el contexto educativo, localizados en Google Scholar y Science Direct, escritos en portugués, inglés y español desde 2021 en adelante. En 2022 se lanzó ChatGPT, tecnología que forma parte del concepto denominado inteligencia artificial generativa, creada mediante técnicas de aprendizaje automático. Este tipo de herramientas ha ganado espacio en las instituciones escolares y ha sido objeto de numerosas discusiones. A raíz de ello, los gobiernos de muchos países, preocupados por sus impactos, intentan regular el uso de la IA. Entre los resultados obtenidos se destaca la presencia de estudios sobre alfabetización en IA, la formación docente, la necesidad de abordar la temática interdisciplinariamente y desde niveles iniciales, entre otros.

ABSTRACT: In this work we use Machine Learning (Randon Forest) as a tool to classify biomass and calculate vegetation indices seeking to identify the characteristics of the vegetation cover at the head of the Atacama Desert. The aim is to establish the correlation between vegetation indices and precipitation, in order to know their reliability on the climatology in this region. The geospatial analysis based on Google Earth Engine (GEE) and the processing of Landsat 5 ETM and Landsat 8 OLI/TIRS images was important, for the period 1985 - 2022, which made it possible to characterize climate change. The NDVI, SAVI, GVI and RVI have been tested and validated in arid systems. The NDVI responds positively to precipitation in the wet season and weakly in the winter rainy season. It is confirmed that the high NDVI corresponds to summer, after a prolonged drought. Towards the years 2020 and 2022, an increase in vegetation cover is recorded in places with higher temperatures, evidencing climate change and reflected in biomass indices.

RESUMEN: En este trabajo usamos Machine Learning (Randon Forest) como herramienta para clasificar la biomasa y calcular los índices de vegetación buscando identificar las características de la cobertura vegetal en la cabecera del desierto Atacama. Se busca establecer la correlación entre los índices de vegetación y la precipitación, a fin de conocer su confiabilidad sobre la climatología en esta región. Fue importante el análisis geoespacial basado en Google Earth Engine (GEE) y el procesamiento de imágenes Landsat 5 ETM y Landsat 8 OLI/TIRS, para el período 1985 - 2022, lo que permitió caracterizar el cambio climático. El NDVI, SAVI, GVI y RVI han sido probados y validados en sistemas áridos. El NDVI responde positivamente a la precipitación en temporada húmeda y en forma débil en la temporada de lluvias invernales. Se confirma que el NDVI alto corresponde al verano, después de una sequía prolongada. Hacia los años 2020 y 2022, se registra un aumento de cobertura vegetal en lugares de mayor temperatura, evidenciando cambio climático y reflejado en los índices de biomasa.

Resumo Embora a acessibilidade ao ensino superior tenha melhorado em países de baixa e média renda (PBMR), a evasão persiste, especialmente entre estudantes socioeconomicamente desfavorecidos. Apesar dos avanços em modelos de Machine Learning para compreender esse desafio, muitos estudos negligenciam fatores institucionais específicos dos PBMR ou se concentram em cursos específicos, limitando sua aplicabilidade e relevância política. Para abordar isso, criamos uma base de dados usando registros administrativos e censitários para prever o desempenho acadêmico na Pontifícia Universidade Católica do Peru (PUCP). Os modelos mais eficazes, incluindo o Random Forest, destacaram preditores como desempenho prévio e pontuações em testes de admissão. Apresentamos um modelo eficiente com dez características que pode prever o desempenho futuro e assim contribuir para a redução da evasão na PUCP.

Abstract Despite improved higher education accessibility in low and middle-income countries (LMICs), challenges persist in student drop-out, especially for socio-economically disadvantaged students. While machine learning models have enhanced our understanding of this challenge by predicting academic performance, many studies overlook LMIC-specific institutional factors or focus on specific courses, limiting their generalizability and policy uses. To address these issues, the authors compiled a comprehensive database using administrative and census data to predict undergraduate academic performance at the Pontifical Catholic University of Peru (PUCP). The study found that the most effective models were tree-based ensembles, particularly Random Forest, with key predictors including prior secondary school performance and university admission test scores. They present a high-performing model using only ten features that can predict future academic performance and potentially aid in reducing student drop-out at PUCP.

Resumen Aunque la accesibilidad a la educación superior ha mejorado en países de renta baja y media (PRMB), persiste el abandono, especialmente entre estudiantes socioeconómicamente desfavorecidos. A pesar de los avances en modelos de Machine Learning para entender este desafío, muchos estudios descuidan factores institucionales específicos de los PRMB o se centran en cursos específicos, limitando su aplicabilidad y relevancia política. Para abordar esto, creamos una base de datos usando registros administrativos y censales para predecir el rendimiento académico en la Pontificia Universidad Católica del Perú (PUCP). Los modelos más efectivos, entre ellos Random Forest, destacaron predictores como el rendimiento previo y puntuaciones en pruebas de admisión. Presentamos un modelo eficiente con diez características que puede predecir el rendimiento futuro y así aportar a la reducción de la deserción en PUCP.

ABSTRACT When the only information available is the issue parameters, and the intended outwards, machine learning techniques like ANN (Artificial Neural Networks) and ANFIS (Adaptive neuro-fuzzy inference system) been proven to address the complex problems without duplicating the phenomena under investigation. The main prompting characteristics are the height-to-the-thicknesses ratio of prisms and the strength under compression of prisms and mortar were analyzed. As inputs, the prototypes are used as blocks and mortars. Both prototypes were accomplished and evaluated. Thirty-six data sets were gathered for testing in addition to verified technical and subsequently comparison with other empirical computation methods served to validate. The outwards show that the suggested prototypes have good forecast capabilities with negligible error rates. To assess and compare the structural behavior of structural completion of AAC block with the other types. At last, both the machine learning tools are good application and dependability.

Abstract: Tomatoes are widely cultivated, both by family farmers and corporate producers. During the tomato growth cycle, several diseases can affect the plant. The identification of these diseases through short-range images is significant, and computer vision techniques are commonly used to identify diseases in plant leaves. In this paper, a hybrid model that combines a convolutional neural network (CNN) and a Random Forest (RF) decision tree is used for foliar spot detection in tomato leaves. High-level features learned and extracted from CNN are used as input for the RF classifier. To evaluate the proposed model’s performance for plant disease identification, a case study of 2480 low-cost digital RGB images collected in actual field conditions, under different intensities of light exposure, were used, including healthy tomato leaves and leaves with visible symptoms of powdery mildew fungus, which attacks the tomato leaf. The results were compared with six conventional machine learning classifiers: Logistic Regression (LR), Linear Discriminant Analysis (LDA), K- Nearest Neighbors (KNN), Naive Bayes (NB), Support Vector Machine (SVM) and Random Forest (RF). The results show that the proposed model outperformed conventional classifiers, reaching an accuracy of 98%. The results highlight the importance of fusing models to improve the detection plant´s diseases.

Abstract Longitudinal study, whose objective was to present a better strategy and statistical methods, and demonstrate its use with the data across the 2013-2015 period in schoolchildren aged 7 to 11 years, covered with the same food questionnaire (WebCAAFE) survey in Florianopolis, southern Brazil. Six meals/snacks and 32 foods/beverages yielded 192 possible combinations denominated meal/snack-Specific Food/beverage item (MSFIs). LASSO algorithm (LASSO-logistic regression) was used to determine the MSFIs predictive of overweight/obesity, and then binary (logistic) regression was used to further analyze a subset of these variables. Late breakfast, lunch and dinner were all associated with increased overweight/obesity risk, as was an anticipated lunch. Time-of-day or meal-tagged food/beverage intake result in large number of variables whose predictive patterns regarding weight status can be analyzed by machine learning such as LASSO, which in turn may identify the patterns not amenable to other popular statistical methods such as binary logistic regression.

Resumo Estudo longitudinal cujo objetivo foi apresentar melhores estratégia e métodos estatísticos e demonstrar sua utilização com os dados do período 2013-2015 em escolares de 7 a 11 anos, contemplados com o mesmo questionário alimentar (WebCAAFE) em Florianópolis, Sul do Brasil. Seis refeições/lanches e 32 alimentos/bebidas resultaram em 192 combinações possíveis denominadas item refeição/lanche-alimentos/bebidas específicos (MSFIs). O algoritmo LASSO (LASSO-regressão logística) foi usado para determinar os MSFIs preditivos de sobrepeso/obesidade e, em seguida, a regressão binária (logística) foi usada para analisar um subconjunto dessas variáveis. Café da manhã, almoço e jantar tardios foram todos associados ao aumento do risco de sobrepeso/obesidade, assim como um almoço antecipado. O consumo de alimentos/bebidas marcados na hora do dia ou na refeição resulta em um grande número de variáveis cujos padrões preditivos em relação ao status do peso podem ser analisados por LASSO. Essa análise pode identificar os padrões não passíveis de outros métodos estatísticos populares, como a regressão logística binária.

Abstract: Severe acute respiratory infection (SARI) outbreaks occur annually, with seasonal peaks varying among geographic regions. Case notification is important to prepare healthcare networks for patient attendance and hospitalization. Thus, health managers need adequate resource planning tools for SARI seasons. This study aims to predict SARI outbreaks based on models generated with machine learning using SARI hospitalization notification data. In this study, data from the reporting of SARI hospitalization cases in Brazil from 2013 to 2020 were used, excluding SARI cases caused by COVID-19. These data were prepared to feed a neural network configured to generate predictive models for time series. The neural network was implemented with a pipeline tool. Models were generated for the five Brazilian regions and validated for different years of SARI outbreaks. By using neural networks, it was possible to generate predictive models for SARI peaks, volume of cases per season, and for the beginning of the pre-epidemic period, with good weekly incidence correlation (R2 = 0.97; 95%CI: 0.95-0.98, for the 2019 season in the Southeastern Brazil). The predictive models achieved a good prediction of the volume of reported cases of SARI; accordingly, 9,936 cases were observed in 2019 in Southern Brazil, and the prediction made by the models showed a median of 9,405 (95%CI: 9,105-9,738). The identification of the period of occurrence of a SARI outbreak is possible using predictive models generated with neural networks and algorithms that employ time series.

Resumen: Brotes de síndrome respiratorio agudo grave (SRAG) ocurren todos los años, con picos estacionales que varían entre regiones geográficas. La notificación de los casos es importante para preparar las redes de atención a la salud para el cuidado y hospitalización de los pacientes. Por lo tanto, los gestores de salud deben tener herramientas adecuadas de planificación de recursos para las temporadas de SRAG. Este estudio tiene el objetivo de predecir brotes de SRAG con base en modelos generados con aprendizaje automático utilizando datos de hospitalización por SRAG. Se incluyeron datos sobre casos de hospitalización por SRAG en Brasil desde 2013 hasta 2020, salvo los casos causados por la COVID-19. Se prepararon estos datos para alimentar una red neural configurada para generar modelos predictivos para series temporales. Se implementó la red neural con una herramienta de canalización. Se generaron los modelos para las cinco regiones brasileñas y se validaron para diferentes años de brotes de SRAG. Con el uso de redes neurales, se pudo generar modelos predictivos para los picos de SRAG, el volumen de casos por temporada y para el inicio del periodo pre-epidémico, con una buena correlación de incidencia semanal (R2 = 0,97; IC95%: 0,95-0,98, para la temporada de 2019 en la Región Sudeste). Los modelos predictivos tuvieron una buena predicción del volumen de casos notificados de SRAG; así, se observaron 9.936 casos en 2019 en la Región Sur, y la predicción de los modelos mostró una mediana de 9.405 (IC95%: 9.105-9.738). La identificación del periodo de ocurrencia de un brote de SRAG es posible a través de modelos predictivos generados con el uso de redes neurales y algoritmos que aplican series temporales.

Resumo: Surtos de síndrome respiratória aguda grave (SRAG) ocorrem anualmente, com picos sazonais variando entre regiões geográficas. A notificação dos casos é importante para preparar as redes de atenção à saúde para o atendimento e internação dos pacientes. Portanto, os gestores de saúde precisam ter ferramentas adequadas de planejamento de recursos para as temporadas de SRAG. Este estudo tem como objetivo prever surtos de SRAG com base em modelos gerados com aprendizado de máquina usando dados de internação por SRAG. Foram incluídos dados sobre casos de hospitalização por SRAG no Brasil de 2013 a 2020, excluindo os casos causados pela COVID-19. Estes dados foram preparados para alimentar uma rede neural configurada para gerar modelos preditivos para séries temporais. A rede neural foi implementada com uma ferramenta de pipeline. Os modelos foram gerados para as cinco regiões brasileiras e validados para diferentes anos de surtos de SRAG. Com o uso de redes neurais, foi possível gerar modelos preditivos para picos de SRAG, volume de casos por temporada e para o início do período pré-epidêmico, com boa correlação de incidência semanal (R2 = 0,97; IC95%: 0,95-0,98, para a temporada de 2019 na Região Sudeste). Os modelos preditivos obtiveram uma boa previsão do volume de casos notificados de SRAG; dessa forma, foram observados 9.936 casos em 2019 na Região Sul, e a previsão feita pelos modelos mostrou uma mediana de 9.405 (IC95%: 9.105-9.738). A identificação do período de ocorrência de um surto de SRAG é possível por meio de modelos preditivos gerados com o uso de redes neurais e algoritmos que aplicam séries temporais.