Is it possible to estimate the number of patients with COVID-19 admitted to intensive care units and general wards using clinical and telemedicine data?

ABSTRACT Objective: To develop and validate predictive models to estimate the number of COVID-19 patients hospitalized in the intensive care units and general wards of a private not-for-profit hospital in São Paulo, Brazil. Methods: Two main models were developed. The first model calculated hospital occupation as the difference between predicted COVID-19 patient admissions, transfers between departments, and discharges, estimating admissions based on their weekly moving averages, segmented by general wards and intensive care units. Patient discharge predictions were based on a length of stay predictive model, assessing the clinical characteristics of patients hospitalized with COVID-19, including age group and usage of mechanical ventilation devices. The second model estimated hospital occupation based on the correlation with the number of telemedicine visits by patients diagnosed with COVID-19, utilizing correlational analysis to define the lag that maximized the correlation between the studied series. Both models were monitored for 365 days, from May 20th, 2021, to May 20th, 2022. Results: The first model predicted the number of hospitalized patients by department within an interval of up to 14 days. The second model estimated the total number of hospitalized patients for the following 8 days, considering calls attended by Hospital Israelita Albert Einstein's telemedicine department. Considering the average daily predicted values for the intensive care unit and general ward across a forecast horizon of 8 days, as limited by the second model, the first and second models obtained R² values of 0.900 and 0.996, respectively and mean absolute errors of 8.885 and 2.524 beds, respectively. The performances of both models were monitored using the mean error, mean absolute error, and root mean squared error as a function of the forecast horizon in days. Conclusion: The model based on telemedicine use was the most accurate in the current analysis and was used to estimate COVID-19 hospital occupancy 8 days in advance, validating predictions of this nature in similar clinical contexts. The results encourage the expansion of this method to other pathologies, aiming to guarantee the standards of hospital care and conscious consumption of resources.

Métodos de aprendizaje automático para predecir el comportamiento epidemiológico de enfermedades arbovirales: revisión estructurada de literatura

Introducción: Los métodos de aprendizaje automático permiten manejar datos estructurados y no estructurados para construir modelos predictivos y apoyar la toma de decisiones. Objetivo: Identificar los métodos de aprendizaje automático aplicados para predecir el comportamiento epidemiológico de enfermedades arbovirales utilizando datos de vigilancia epidemiológica. Metodología: Se realizó búsqueda en EMBASE y PubMed, análisis bibliométrico y síntesis de la información. Resultados: Se seleccionaron 41 documentos, todos publicados en la última década. La palabra clave más frecuente fue dengue. La mayoría de los autores (88,3 %) participó en un artículo de investigación. Se encontraron 16 métodos de aprendizaje automático, el más frecuente fue Red Neuronal Artificial, seguido de Máquinas de Vectores de Soporte. Conclusiones: En la última década se incrementó la publicación de trabajos que pretenden predecir el comportamiento epidemiológico de arbovirosis por medio de diversos métodos de aprendizaje automático que incorporan series de tiempo de los casos, variables climatológicas, y otras fuentes de información de datos abiertos.

Introduction: Machine learning methods allow to manipulate structured and unstructured data to build predictive models and support decision-making. Objective: To identify machine learning methods applied to predict the epidemiological behavior of vector-borne diseases using epidemiological surveillance data. Methodology: A literature search in EMBASE and PubMed, bibliometric analysis, and information synthesis were performed. Results: A total of 41 papers were selected, all of them were published in the last decade. The most frequent keyword was dengue. Most authors (88.3 %) participated in a research article. Sixteen machine learning methods were found, the most frequent being Artificial Neural Network, followed by Support Vector Machines. Conclusions: In the last decade there has been an increase in the number of articles that aim to predict the epidemiological behavior of vector-borne diseases using by means of various machine learning methods that incorporate time series of cases, climatological variables, and other sources of open data information.

Rendimiento en los Juegos Panamericanos de Santiago 2023: Predicción mediante método de suavización exponencial simple

El objetivo del presente trabajo es analizar el desempeño deportivo de la delegación chilena en los Juegos Panamericanos celebrados entre los años 1951 y 2023, haciendo uso de datos retrospectivos y proyectivos a través de series temporales de tiempo. Para esto se empleó un diseño cuantitativo, no experimental y longitudinal de tendencias y un método de suavización exponencial simple, que utiliza promedios históricos y que permite realizar una predicción o comportamiento futuro basado en una media ponderada de los valores actuales y de los pasados. A partir de los resultados obtenidos, fue posible concluir que, en las últimas décadas, la ubicación de Chile en el ranking de los Juegos Panamericanos se ha estabilizado en torno a un onceavo lugar, posición pronosticada para Santiago 2023. Manteniéndose condiciones similares, el desempeño deportivo general y específico no tendría un quiebre exponencial de la tendencia y los resultados no resultan favorables, específicamente en lo que respecta a la obtención de medallas de oro y la posición general de la delegación.

The objective of this paper is to analyze the sports performance of the Chilean delegation in the Pan American Games held between 1951 and 2023, using retrospective and projective data through time series. For this purpose, a quantitative, non-experimental and longitudinal design of trends and a simple exponential smoothing method was used, which uses historical averages and allows a prediction or future behavior based on a weighted average of current and past values. From the results obtained, it was possible to conclude that, in recent decades, Chile's position in the Pan American Games ranking has stabilized around eleventh place, a position predicted for Santiago 2023. Maintaining similar conditions, the general and specific sporting performance would not have an exponential break in the trend and the results are not favorable, specifically in terms of obtaining gold medals and the overall position of the delegation.

O objetivo deste artigo é analisar o desempenho esportivo da delegação chilena nos Jogos Pan-Americanos realizados entre 1951 e 2023, usando dados retrospectivos e projetivos por meio de séries temporais. Para isso, foi utilizado um desenho quantitativo, não experimental e longitudinal de tendências e um método de suavização exponencial simples, que utiliza médias históricas e permite uma previsão do comportamento futuro com base em uma média ponderada dos valores atuais e passados. Com base nos resultados obtidos, foi possível concluir que, nas últimas décadas, a posição do Chile no ranking dos Jogos Pan-Americanos se estabilizou em torno do 11º lugar, posição prevista para Santiago 2023. Mantendo-se condições semelhantes, o desempenho esportivo geral e específico não teria uma quebra exponencial na tendência e os resultados não são favoráveis, especificamente em termos de conquista de medalhas de ouro e posição geral da delegação.

Análisis temporal de la mortalidad por causas evitables en las primeras 24 horas de vida, 2000-2021 / Temporal analysis of mortality from preventable causes in the first 24 hours of life, 2000-2021 / Análise temporal da mortalidade por causas evitáveis nas primeiras 24 horas de vida, 2000-2021

Objetivo: analizar el patrón temporal y estimar las tasas de mortalidad en las primeras 24 horas de vida y por causas evitables en el estado de Pernambuco en el período de 2000 a 2021. Método: estudio ecológico, teniendo como unidad de análisis el trimestre. La fuente de datos se constituyó por el Sistema de Informaciones sobre Mortalidad y el Sistema de Informaciones sobre Nacidos Vivos. El modelado de series temporales se realizó según el Modelo Autorregresivo Integrado de Promedio Móvil. Resultados: se registraron 14.462 óbitos en las primeras 24 horas de vida, siendo 11.110 (el 76,8%) evitables. Se observa para los pronósticos ( forecasts) que la tasa de mortalidad en las primeras 24 horas de vida registro una variación de 3,3 a 2,4 por 1.000 nacidos vivos, y la tasa de mortalidad por causas evitables de 2,3 a 1,8 por 1.000 nacidos vivos. Conclusión: la predicción sugirió avances en la reducción de la mortalidad en las primeras 24 horas de vida en el estado y por causas evitables. Los modelos ARIMA presentaron estimaciones satisfactorias para las tasas de mortalidad y por causas evitables en las primeras 24 horas de vida.

Objective: to analyze the temporal pattern and estimate mortality rates in the first 24 hours of life and from preventable causes in the state of Pernambuco from 2000 to 2021. Method: an ecological study, using the quarter as the unit of analysis. The data source was made up of the Mortality Information System and the Live Birth Information System. The time series modeling was conducted according to the Autoregressive Integrated Moving Average Model. Results: 14,462 deaths were recorded in the first 24 hours of life, 11,110 (76.8%) of which being preventable. It is observed from the forecasts that the mortality rate in the first 24 hours of life ranged from 3.3 to 2.4 per 1,000 live births, and the mortality rate from preventable causes ranged from 2.3 to 1.8 per 1,000 live births. Conclusion: the prediction suggested progress in reducing mortality in the first 24 hours of life in the state and from preventable causes. The ARIMA models presented satisfactory estimates for mortality rates and preventable causes in the first 24 hours of life.

Objetivo: analisar o padrão temporal e estimar as taxas de mortalidade nas primeiras 24 horas de vida e por causas evitáveis no estado de Pernambuco no período de 2000 a 2021. Método: estudo ecológico, tendo como unidade de análise o trimestre. A fonte de dados foi constituída pelo Sistema de Informações sobre Mortalidade e pelo Sistema de Informações sobre Nascidos Vivos. A modelagem da série temporal foi conduzida segundo o Modelo Autorregressivo Integrado de Médias Móveis. Resultados: foram registrados 14.462 óbitos nas primeiras 24 horas de vida, sendo 11.110 (76,8%) evitáveis. Observa-se para os forecasts que a taxa de mortalidade nas primeiras 24 horas de vida variou de 3,3 a 2,4 por 1.000 nascidos vivos, e a taxa de mortalidade por causas evitáveis variou de 2,3 a 1,8 por 1.000 nascidos vivos. Conclusão: a previsão sugeriu avanços na redução da mortalidade nas primeiras 24 horas de vida no estado e por causas evitáveis. Os modelos ARIMA apresentaram estimativas satisfatórias para as taxas de mortalidade e por causas evitáveis nas primeiras 24 horas de vida.

Evaluación de la disfunción del endotelio para el pronóstico de riesgo cardiovascular / Evaluation of endothelial dysfunction for cardiovascular risk prediction i: en

Introducción: Las enfermedades cardiovasculares constituyen la primera causa de muerte en el mundo, por lo que la identificación y modificación de los factores de riesgo asociados a ellas constituyen estrategias priorizadas por la Organización Mundial de la Salud. Contar con un modelo de predicción del riesgo cardiovascular enriquecido con la evaluación de la disfunción endotelial influiría positivamente en estas metas. Objetivos: Identificar la presencia de disfunción endotelial en pacientes con enfermedades cardiovasculares o sin estas y determinar la asociación entre ambas. Métodos: Se realizó un estudio observacional y descriptivo, de serie de casos, en el Centro de Cardiología y Cirugía Cardiovascular del Hospital Provincial Docente Clínico-Quirúrgico Saturnino Lora de Santiago de Cuba, desde enero del 2022 hasta igual mes del 2023, donde se analizaron como variables los factores de riesgo cardiovascular tradicionales y los biomarcadores de disfunción endotelial. Secundariamente, se llevó a cabo un estudio analítico de casos y controles en el cual se aplicó la regresión logística binaria multivariada. Resultados: Se confirmó la presencia de disfunción endotelial asociada a la aparición de las enfermedades cardiovasculares, lo que se evaluó a través del índice de vasodilatación, mediado por el flujo de la arteria braquial y las concentraciones plasmáticas de fibrinógeno. Conclusiones: Las características epidemiológicas y clínicas de los pacientes con enfermedades cardiovasculares o sin estas no difirieron de lo registrado en la literatura especializada acerca de la base de identificación de los factores de riesgo tradicionales.

Introduction: Cardiovascular diseases constitute the first death cause worldwide, reason why the identification and modification of associated risk factors constitute prioritized strategies by the World Health Organization. To have a prediction model of cardiovascular risk enriched with the evaluation of the endothelial dysfunction would influence positively in these goals. Objectives: To identify the presence of endothelial dysfunction in patients with or without cardiovascular diseases and to determine the association between them. Methods: An observational and descriptive cases series study was carried out in the Cardiology and Cardiovascular Surgery Center at Saturnino Lora Teaching Clinical Surgical Provincial Hospital in Santiago de Cuba, from January, 2022 to the same month, 2023, where the traditional cardiovascular risk factors and endothelial dysfunction biomarkers were analyzed as variables. Secondarily, an analytic case-control study was carried out in which multivariate binary logistic regression was applied. Results: The presence of endothelial dysfunction associated with the onset of cardiovascular diseases was confirmed, what was evaluated through the vasodilatation index, mediated by the brachial artery flow and the fibrinogen plasmatic concentrations. Conclusions: The clinical and epidemiological pattern of patients with or without cardiovascular diseases did not differ from that reported in the specialized literature on the base of the identification of traditional risk factors.

Evaluación cuantitativa de un modelo SEIR para predecir casos de COVID-19 / Quantitative evaluation of a SEIR model for forecasting COVID-19 cases

Resumen Introducción : Los modelos epidemiológicos han sido ampliamente utilizados durante la pandemia de COVID-19, aunque la evaluación de su desempeño ha sido limitada. El objetivo del presente trabajo fue evaluar de forma retrospectiva un modelo SEIR para la predicción de casos a corto plazo (1 a 3 semanas), cuantificando su desempeño real y potencial, me diante la optimización de los parámetros del modelo. Métodos : Se realizaron proyecciones para cada día de la primera ola de casos (31 de julio de 2020 al 11 de marzo de 2021) en el municipio de General Pueyrredón (Argentina), cuantificando el desempeño del modelo en términos de incertidumbre, inexactitud e imprecisión. La evaluación se realizó con los parámetros originales del modelo (utilizados en proyecciones que fueron oportunamente publicadas), y luego variando distintos parámetros a fin de identificar valores óptimos. Resultados : El análisis del desempeño del modelo mostró que valores alternativos de algunos parámetros, y la corrección de los valores de entrada utilizando un filtro de "media móvil" para eliminar las variaciones semanales en los reportes de casos, habrían otorgado mejores resultados. El modelo con los parámetros opti mizados logró disminuir desde casi 40% a menos de 15% la incertidumbre, con valores similares de inexactitud, y con una imprecisión levemente mayor. Discusión : Modelos epidemiológicos sencillos, sin grandes requerimientos para su implementación, pue den ser de utilidad para la toma de decisiones rápi das en localidades pequeñas o con recursos limitados, siempre y cuando se tenga en cuenta la importancia de su evaluación y la consideración de sus alcances y limitaciones.

Abstract Introduction : Epidemiological models have been widely used during the COVID-19 pandemic, although performance evaluation has been limited. The objec tive of this work was to thoroughly evaluate a SEIR model used for the short-term (1 to 3 weeks) predic tion of cases, quantifying its actual past performance, and its potential performance by optimizing the model parameters. Methods : Daily case forecasts were obtained for the first wave of cases (July 31, 2020 to March 11, 2021) in the district of General Pueyrredón (Argentina), quantifying the model performance in terms of uncertainty, inac curacy and imprecision. The evaluation was carried out with the original parameters of the model (used in the forecasts that were published), and also varying different parameters in order to identify optimal values. Results : The analysis of the model performance showed that alternative values of some parameters, and the correction of the input values using a "mov ing average" filter to eliminate the weekly variations in the case reports, would have yielded better results. The model with the optimized parameters was able to reduce the uncertainty from almost 40% to less than 15%, with similar values of inaccuracy, and with slightly greater imprecision. Discussion : Simple epidemiological models, without large requirements for their implementation, can be very useful for making quick decisions in small cities or cities with limited resources, as long as the importance of their evaluation is taken into account and their scope and limitations are considered.

Modelo de inteligencia artificial para la detección temprana de diabetes

Introducción. La diabetes es una enfermedad crónica que se caracteriza por el aumento de la concentración de la glucosa en sangre. Puede generar complicaciones que afectan la calidad de vida y aumentan los costos de la atención en salud. En los últimos años, las tasas de prevalencia y mortalidad han aumentado en todo el mundo. El desarrollo de modelos con gran desempeño predictivo puede ayudar en la identificación temprana de la enfermedad. Objetivo. Desarrollar un modelo basado en la inteligencia artificial para apoyar la toma de decisiones clínicas en la detección temprana de la diabetes. Materiales y métodos. Se llevó a cabo un estudio de corte transversal, utilizando un conjunto de datos que incluía edad, signos y síntomas de pacientes con diabetes y de individuos sanos. Se utilizaron técnicas de preprocesamiento para los datos. Posteriormente, se construyó el modelo basado en mapas cognitivos difusos. El rendimiento se evaluó mediante tres parámetros: exactitud, especificidad y sensibilidad. Resultados. El modelo desarrollado obtuvo un excelente desempeño predictivo, con una exactitud del 95 %. Además, permitió identificar el comportamiento de las variables involucradas usando iteraciones simuladas, lo que proporcionó información valiosa sobre la dinámica de los factores de riesgo asociados con la diabetes. Conclusiones. Los mapas cognitivos difusos demostraron ser de gran valor para la identificación temprana de la enfermedad y en la toma de decisiones clínicas. Los resultados sugieren el potencial de estos enfoques en aplicaciones clínicas relacionadas con la diabetes y respaldan su utilidad en la práctica médica para mejorar los resultados de los pacientes.

Introduction. Diabetes is a chronic disease characterized by a high blood glucose level. It can lead to complications that affect the quality of life and increase the costs of healthcare. In recent years, prevalence and mortality rates have increased worldwide. The development of models with high predictive performance can help in the early identification of the disease. Objective. To develope a model based on artificial intelligence to support clinical decision-making in the early detection of diabetes. Materials and methods. We conducted a cross-sectional study, using a dataset that contained age, signs, and symptoms of patients with diabetes and of healthy individuals. Pre-processing techniques were applied to the data. Subsequently, we built the model based on fuzzy cognitive maps. Performance was evaluated with three metrics: accuracy, specificity, and sensitivity. Results. The developed model obtained an excellent predictive performance with an accuracy of 95%. In addition, it allowed to identify the behavior of the variables involved using simulated iterations, which provided valuable information about the dynamics of the risk factors associated with diabetes. Conclusions. Fuzzy cognitive maps demonstrated a high value for the early identification of the disease and in clinical decision-making. The results suggest the potential of these approaches in clinical applications related to diabetes and support their usefulness in medical practice to improve patient outcomes.

Modelo explicativo de ansiedad generalizada y bullying como predictores de autolesiones no suicidas en adolescentes peruanos

Fundamento las autolesiones no suicidas se consideran un problema de salud pública y social durante la última década, el cual afecta en mayor medida a los adolescentes. La ansiedad generalizada y el bullying pueden ser factores desencadenantes para su desarrollo. Objetivo analizar un modelo explicativo de la ansiedad generalizada y el bullying como predictores de autolesiones no suicidas en adolescentes peruanos. Métodos estudio de diseño explicativo, transversal, con participación de 1 249 adolescentes peruanos, de edad promedio de 15 años (desviación estándar = 1,49) quienes respondieron escalas sobre ansiedad generalizada, bullying y autolesiones no suicidas. Para el análisis de datos, se aplicó la potencia estadística, la correlación y un modelo de regresión estructural basado en covarianzas para confirmar el modelo explicativo. Resultados las variables psicológicas se correlacionaron de manera positiva y estadísticamente significativa. El modelo propuesto presentó índices de ajuste adecuados (CFI = 0,94; RMSEA = 0,03 [IC del 90 %: 0,02-0,03] y SRMR = 0,04) y se evidenció que la ansiedad generalizada (β = 0,26, p = 0,001) y las dimensiones del bullying, como la agresión (β = 0,25, p = 0,001) y la victimización (β = 0,21, p = 0,003) predijeron de manera estadísticamente significativa las autolesiones no suicidas. Conclusiones los hallazgos sugieren que tanto la ansiedad generalizada como el bullying predicen las autolesiones no suicidas en adolescentes. La evidencia proporciona información útil para desarrollar y evaluar programas de prevención basados en estas variables psicológicas, con vistas a disminuir el riesgo de las autolesiones no suicidas.

Background non-suicidal self-harm has been considered a public and social health problem during the last decade, which affects adolescents to a greater extent. Generalized anxiety and bullying can be trigger factors for its development. Objective to analyze a generalized anxiety and bullying explanatory model as non-suicidal self-harm predictors in Peruvian adolescents. Methods cross-sectional, explanatory design study, with 1,249 Peruvian adolescents, average age 15 years old (standard deviation = 1.49), who answered scales on generalized anxiety, bullying, and non-suicidal self-harm. For data analysis, statistical power, correlation, and a structural regression model based on covariances were applied to confirm the explanatory model. Results the psychological variables were positively and statistically significantly correlated. The proposed model had adequate fit indices (CFI = 0.94; RMSEA = 0.03 [90% CI: 0.02-0.03] and SRMR = 0.04) and it was evidenced that generalized anxiety (β = 0.26, p = 0.001) and bullying dimensions such as aggression (β = 0.25, p = 0.001) and victimization (β = 0.21, p = 0.003) statistically significantly predicted self-harm not suicidal. Conclusions The findings suggest that both generalized anxiety and bullying predict non suicidal self-harm in adolescents. The evidence provides useful information for developing and evaluating prevention programs based on these psychological variables, to reduce the non-suicidal self-harm risks.

Effect of Soil and Climatic Conditions on Brown Spot Occurrence in Rice Lowland across Four Agro-climatic Zones of Côte d’Ivoire

Aims: To evaluate interaction of soil pH and acidity with weather on Rice Brown spot (BS) occurrence in rice lowlands. Study Design: Cross sectional study. Place and Duration of Study: Four distinct rice lowlands belonging to different climatic zones (forest, transitional and savanna) of Côte d’Ivoire during cropping seasons of 2021. Methodology: BS characterization were done in different farmer fields where soil samples were also collected during dry and rainy seasons. Soil silicon and acidity were determined in those samples and rice grain yield at harvest time were recorded in different sites. Weather data related to sites and seasons were used to find out correlations. Results: Occurrence of BS was found in forest zones with scores of 4 and 3 compared to 1 and 2 in savanna and transitional zones, respectively, with seasonal variation. Both rice production and the occurrence of BS were explained by soil parameters in conjunction with climatic parameters. Rainfall (R=0.38) and relative humidity (R=0.64) leaded BS occurrence and decrease of yield. Wind speed (R=0.62) and air maximum temperature (R= 0.63) were the determinant factors affecting rice yields. Si was found to be a component of sustainable soil management that interferes with soil pH in all climatic zones. Combined with Temperature, both soil parameters predicted BS occurrence over 50%. Conclusion: Temperature decrease BS pathogens occurrence whereas high humidity increases its spread. Those parameters combined with silicon which interferes with pH could leads sustainable solutions in BS control. Furthermore, having a deep understanding with rice varietal considerations can significantly improve strategies related to rice cultivation and protection.

Diseño y validación de riesgo de síndrome bajo gasto cardíaco en la cirugía valvular

Introducción: La incidencia del síndrome bajo gasto cardíaco postoperatorio es variable en las distintas series publicadas, desde 4 % hasta 15 %, con una mortalidad que se aproxima al 20 %. Si bien en enfermos mayores de 70 años el síndrome puede estar presente hasta en un 63 %, a pesar del desarrollo de mejores técnicas de cardioprotección y cuidados postoperatorios, la incidencia de este síndrome en poblaciones de alto riesgo no se ha modificado en una proporción significativa. Objetivo: Diseñar y validar un modelo predictivo de síndrome de bajo gasto cardíaco postoperatorio a través de factores de riesgo. Métodos: Se realizó un estudio analítico de casos y controles en pacientes con síndrome de bajo gasto cardíaco posoperatorio, atendidos en el Centro de Cardiología y Cirugía Cardiovascular del Hospital Provincial Docente Saturnino Lora de la provincia Santiago de Cuba, en un periodo 2019-2021. Se empleó la regresión logística con ajustes para obtener el modelo. Resultados: Los factores de riesgo predictores de mayor valor fue edad > 65 años, función de ventrículo derecho disminuida, el tiempo de pinzamiento aórtico, sangrado postoperatorio que fueron los que arrojó el modelo de regresión logística. Se realizó su validación interna por división de datos. Conclusiones: El modelo predictivo elaborado a partir de la regresión logística quedó compuesto por los predictores: edad > 65 años, el tiempo de pinzamiento aórtico > 90 minutos y el sangrado posoperatorio prolongado; presentó buen ajuste y poder discriminante, sobre todo valor predictivo positivo.

Introduction: The incidence of postoperative low cardiac output syndrome is variable in the different published series, from 4% to 15%, with a mortality approaching 20%. Although in patients over 70 years of age the syndrome may be present in up to 63%, despite the development of better cardioprotection techniques and postoperative care, the incidence of this syndrome in high-risk populations has not changed in a significant proportion. Objective: To design and validate a predictive model of postoperative low cardiac output syndrome through risk factors. Methods: An analytical, case-control study was conducted in patients with postoperative low cardiac output syndrome attended at the Center for Cardiology and Cardiovascular Surgery of the Saturnino Lora Teaching Provincial Hospital in Santiago de Cuba, in the period 2019-2021. Logistic regression with adjustments to obtain the model was used. Results: The highest value predictor risk factors are: age > 65 years, decreased right ventricular function, aortic clamping time, postoperative bleeding, which were the ones that yielded the logistic regression model. Internal validation was performed by data division. Conclusions: The predictive model developed from logistic regression was composed of the predictors: age > 65 years, aortic clamping time > 90 minutes and prolonged postoperative bleeding. It presented good fit and discriminant power, especially positive predictive value.

Web App for prediction of hospitalisation in Intensive Care Unit by covid-19 / Web App para la predicción de hospitalización em la Unidad de Cuidados Intensivos por covid-19 / Web App para a predição de internação em Unidade de Terapia Intensiva por covid-19

ABSTRACT Objective: To develop a Web App from a predictive model to estimate the risk of Intensive Care Unit (ICU) admission for patients with covid-19. Methods: An applied technological production research was carried out with the development of Streamlit using Python, considering the decision tree model that presented the best performance (AUC 0.668). Results: Based on the variables associated with Precision Nursing, Streamlit stratifies patients admitted to clinical units who are most likely to be admitted to the Intensive Care Unit, serving as a decision-making support tool for healthcare professionals. Final considerations: The performance of the model may have been influenced by the start of vaccination during the data collection period, however, the Web App via Streamlit proved to be a feasible tool for presenting research results, due to the ease of understanding by nurses and its potential for supporting clinical decision-making.

RESUMEN Objetivo: Desarrollar una Web App a partir de un modelo predictivo para estimar el riesgo de ingreso a la Unidad de Cuidados Intensivos (UCI) para pacientes con covid-19. Métodos: Se realizó una investigación de producción tecnológica aplicada con el desarrollo de Streamlit utilizando Python, considerando el modelo de árbol de decisiones que presentó el mejor rendimiento (AUC 0.668). Resultados: Basado en las variables asociadas con la Enfermería de Precisión, Streamlit estratifica a los pacientes ingresados en unidades clínicas que tienen más probabilidades de ser admitidos en la Unidad de Cuidados Intensivos, sirviendo como una herramienta de apoyo para la toma de decisiones para los profesionales de la salud. Consideraciones finales: El rendimiento del modelo puede haber sido influenciado por el inicio de la vacunación durante el período de recolección de datos. La Web App a través de Streamlit demostró ser una herramienta factible para presentar los resultados, debido a la facilidad de comprensión y su potencial para apoyar la toma de decisiones clínicas.

RESUMO Objetivo: Desenvolver um Web App a partir de um modelo preditivo para estimar o risco de internação de pacientes com covid-19 em UTI. Métodos: Realizou-se uma pesquisa aplicada de produção tecnológica com o desenvolvimento do Streamlit a partir do Python, considerando o modelo de árvore de decisão que apresentou o melhor desempenho (AUC 0.668). Resultados: A partir das variáveis associadas à Enfermagem de Precisão, o Streamlit estratifica os pacientes internados nas unidades clínicas com maior probabilidade de internação em Unidade de Terapia Intensiva, funcionando como uma ferramenta de apoio à tomada de decisão dos profissionais de saúde. Considerações finais: A performance do modelo pode ter sido influenciada pelo início da vacinação no período de coleta de dados, no entanto, o Web App via Streamlit mostrou-se uma ferramenta viável para a apresentação dos resultados de pesquisa, devido à facilidade de entendimento por parte dos enfermeiros e pelo potencial de apoio à decisão clínica.

Scientific writing with artificial intelligence: key considerations and alerts / Escrita científica com inteligência artificial: principais considerações e alertas

A integração de geradores de texto de inteligência artificial (IA) em relatórios científicos exige uma avaliação cuidadosa de considerações éticas específicas. Embora essas tecnologias de IA ofereçam suporte à geração de texto, abordar as implicações éticas é fundamental. Este editorial destaca a necessidade de uma abordagem ponderada e responsável, enfatizando o estabelecimento de diretrizes e melhores práticas por parte de pesquisadores e comunidades científicas. Esforços colaborativos entre desenvolvedores de IA, pesquisadores e comitês éticos podem garantir a integração perfeita das tecnologias de IA, ao mesmo tempo em que mantêm a integridade, qualidade e padrões éticos da divulgação científica. Este texto oferece um resumo abrangente considerações-chave ao se utilizar geradores de texto de inteligência artificial em relatórios científicos (AU)

The integration of artificial intelligence (AI) text generators in scientific reports demands careful evaluation of specific ethical considerations. While these AI technologies offer text generation support, addressing the ethical implications is vital. This editorial highlights the need for a thoughtful and responsible approach, emphasizing the establishment of guidelines and best practices by researchers and scientific communities. Collaborative efforts between AI developers, researchers, and ethical committees can ensure the seamless integration of AI technologies while upholding the integrity, quality, and ethical standards of scientific reporting. This text comprehensively summarizes the key considerations to be followed when utilizing artificial intelligence text generators in scientific reports.(AU)

Uso de técnicas de aprendizado de máquina para classificação de fatores que influenciam a ocorrência de dermatites ocupacionais / Using machine learning techniques to classify factors that influence the occurrence of occupational dermatitis

Resumo Introdução: realizar a predição de doenças relacionadas ao trabalho é um desafio às organizações e ao poder público. Com as técnicas de aprendizado de máquina (AM), é possível identificar fatores determinantes para a ocorrência de uma doença ocupacional, visando direcionar ações mais efetivas à proteção dos trabalhadores. Objetivo: predizer, a partir da comparação de técnicas de AM, os fatores com maior influência para a ocorrência de dermatite ocupacional. Métodos: desenvolveu-se um código em linguagem R e uma análise descritiva dos dados e identificaram-se os fatores de influência de acordo com a técnica de AM que demonstrou melhor desempenho. O banco de dados foi disponibilizado pelo Serviço de Dermatologia Ocupacional da Fundação Oswaldo Cruz e contém informações de trabalhadores que apresentaram alterações cutâneas sugestivas de dermatite ocupacional no período de 2000-2014. Resultados: as técnicas com melhor desempenho foram: neural network, random forest, support vector machine e naive Bayes. As variáveis sexo, escolaridade e profissão foram as mais adequadas para os modelos de previsão de dermatite ocupacional. Conclusão: as técnicas de AM possibilitam predizer os fatores que influenciam a segurança e a saúde dos trabalhadores, os parâmetros que subsidiam a implantação de procedimentos e as políticas mais efetivas para prevenir a dermatite ocupacional.

Abstract Introduction: to predict work related diseases is a challenge for organizations and the governmental authorities. By means of machine learning (ML) techniques it is possible to identify factors that determine the occurrence of an occupational disease, aiming at taking more effective actions to protect workers. Objective: to predict, by comparing ML techniques, the factors which highly influence the occurrence of occupational dermatitis. Methods: we developed a code in R language and a descriptive analysis of the data and identified the influence factors according to the ML technique that presented the best performance. The database was made available by the Occupational Dermatology Service of Oswaldo Cruz Foundation and assembles information of the workers who experienced cutaneous alterations suggestive of occupational dermatitis between 2000-2014. Results: the techniques which presented the best performance were: neural network, random forest, support vector machine, and naive Bayes. Sex, schooling, and profession were the most adequate variables for the occupational dermatitis prediction models. Conclusion: ML techniques allowed to predict the factors that influence the workers' safety and health, as well as the parameters that subsidize the procedures implementation, and the most effective policies to prevent occupational dermatitis.

Predicting respiratory motion using an Informer deep learning network / 中华放射医学与防护杂志

Objective:To investigate a time series deep learning model for respiratory motion prediction.Methods:Eighty pieces of respiratory motion data from lung cancer patients were used in this study. They were divided into a training set and a test set at a ratio of 8∶2. The Informer deep learning network was employed to predict the respiratory motions with a latency of about 600 ms. The model performance was evaluated based on normalized root mean square errors (nRMSEs) and relative root mean square errors (rRMSEs).Results:The Informer model outperformed the conventional multilayer perceptron (MLP) and long short-term memory (LSTM) models. The Informer model yielded an average nRMSE and rRMSE of 0.270 and 0.365, respectively, at a prediction time of 423 ms, and 0.380 and 0.379, respectively, at a prediction time of 615 ms.Conclusions:The Informer model performs well in the case of a longer prediction time and has potential application value for improving the effects of the real-time tracking technology.

Comparative analysis of clinical diagnosis application of two intrauterine adhesion scoring criteria / 中华妇产科杂志

Objective:To explore the similarities and differences of China Society of Gynecology Endoscopy (CSGE) and American Fertility Society (AFS) intrauterine adhesion (IUA) scoring criteria on IUA grading and their predictive value of reproductive prognosis.Methods:From January 2016 to January 2019, a total of 1 249 patients were diagnosed with IUA by hysteroscopy at Beijing Obstetrics and Gynecology Hospital. Totally, 378 patients with complete clinical data were enrolled, and the results diagnosed by CSGT and AFS scoring criteria were compared and analyzed.And follow-up for 2 years, the pregnancy rate and live birth rate were statistical analysis.Results:(1) The grade of IUA according to AFS and CSGE scoring criteria was less consistent ( κ=0.295, P<0.001). Compared with AFS, the proportion of severe IUA cases diagnosed by CSGE was significantly lower [45.8% (173/378) vs 15.1% (57/378); RR=0.22, 95% CI: 0.15-0.30, P<0.01); the proportions of both mild and moderate IUA cases were significantly higher ( RR=4.16, 95% CI: 2.38-7.14; RR=2.38, 95% CI: 1.75-3.23; both P<0.01). (2) The pregnancy rates of mild, moderate and severe IUA diagnosed according to CSGE were 11/13, 64.5% (147/228), 31.8% (7/22), live birth rates were 11/13, 54.8% (125/228) and 22.7% (5/22), respectively; there were statistically significant differences between the groups (all P<0.01). The pregnancy rates of mild, moderate and severe IUA diagnosed based on AFS were 3/3, 66.9% (97/145) and 56.5% (65/115), respectively, with no statistically significant difference between the groups ( P>0.05). (3) IUA grades based on both CSGE and AFS criteria were significantly negatively correlated with pregnancy rates and live birth rates (CSGE: r=-0.210, r=-0.226; AFS: r=-0.130, r=-0.147; all P<0.05). Univariate logistic regression analysis showed that CSGE had higher OR for both pregnancy rates and live birth rates compared to AFS (3.889 vs 1.657, 3.983 vs 1.554, respectrvely). Conclusions:Compared with AFS, the IUA grade based on CSGE is better related with reproductive prognosis, suggesting that the CSGE standard might be more objective and comprehensive and has better predictive value for reproductive prognosis, thus avoiding overdiagnosis and overtreatment.

Predictive indicators of a prolonged course of fever in children with infectious mononucleosis / 中国基层医药

Objective:To explore the relevant predictive indicators of fever course > 7 days in children with infectious mononucleosis.Methods:The clinical data of 163 children with infectious mononucleosis who received treatment in Xi'an Children's Hospital from January 2018 to October 2020 were retrospectively analyzed. According to the heat duration, the children were divided into the fever course > 7 days group ( n = 55) and the fever course ≤ 7 days group ( n = 108). The clinical manifestations and laboratory indexes on admission were compared between the two groups. A logistic regression model was used to analyze the influential factors of fever course in children. A receiver operating curve was used to evaluate the predictive value of heat course > 7 days for infectious mononucleosis. Results:The majority of children with infectious mononucleosis had a heat course of 7 days (21.5%). There were no significant differences in clinical manifestations between the fever course > 7 days group and the fever course ≤ 7 days group (all P > 0.05). Neutrophil count, the proportion of monocytes, aspartate aminotransferase, and the proportion of suppressor T (Ts) cells in the fever course > 7 days group were (15.97 ± 7.60) × 10 9/L, 7.75 (4.93, 10.75)%, 53.00 (22.00, 91.50) U/L, 70.00 (57.00, 75.00)%, respectively, which were significantly higher than (15.21 ± 5.29) × 10 9/L, 5.40 (3.40, 9.60)%, 40.00 (30.00, 63.75) U/L, 63.50 (55.00,70.75)% in the fever course ≤ 7 days group ( t = -5.10, Z = -2.31, Z = -2.26, Z = -2.12, all P < 0.05). The proportion of helper T (Th) cells and the ratio of Th/Ts cells in the fever course > 7 days group were 13.00 (9.00, 17.00)% and 0.19 (0.12, 0.30)%, respectively, which were significantly lower than 16.00 (12.25, 20.75)%, 0.26 (0.18, 0.37)% in the fever course ≤ 7 days group ( Z = 2.44, 2.48, both P < 0.05). Multivariate logistic regression analysis showed that the increased proportion of Ts cells ( OR = 0.96, 95% CI 0.922-0.978, P < 0.05) was an influential factor of the prolonged course of fever. The area under the receiver operating characteristic curve of the proportion of Ts cells was 0.637. The cut-off value, sensitivity, and specificity were 67.50%, 61.3%, and 64.3%, respectively. Conclusion:Children with infectious mononucleosis with a longer heat course have more severe immune responses. The proportion of Ts cells > 67.5% can be used as a risk factor for the fever course > 7 days in children with infectious mononucleosis.

The value of fibroblast growth factor 2 and microRNA-206 in predicting postoperative delayed union of closed tibial shaft fractures / 中国基层医药

Objective:To investigate the value of fibroblast growth factor 2 (FGF-2) and microRNA-206 (miR-206) in predicting postoperative delayed union of closed tibial shaft fractures.Methods:The clinical data of 136 patients who underwent closed tibial shaft fracture surgery in Hospital of the 80 th Group Army of Chinese People's Liberation Army Ground Forces from May 2018 to May 2021 were retrospectively analyzed. Eighty-six patients who had delayed union of closed tibial shaft fractures were included in the observation group, and fifty patients who had normal union of closed tibial shaft fractures were included in the control group. Serum FGF-2 level was measured using the enzyme-linked immunosorbent assay, and serum miR-206 expression was detected using the real-time fluorescence polymerase chain reaction. The relationship between FGF-2 expression and miR-206 expression and closed tibial shaft fractures was analyzed. Results:At 1 day, 1, and 4 weeks after surgery, serum FGF-2 level was significantly lower in the observation group than the control group [(14.24 ± 2.15) ng/L vs. (20.36 ± 3.42) ng/L, (21.38 ± 3.27) ng/L vs. (30.45 ± 4.29) ng/L, (23.59 ± 4.36) ng/L vs. (36.67 ± 4.51) ng/L, t = 7.42, 8.42, 16.66, all P < 0.001]. Serum FGF-2 level gradually increased with time in each group. At 1 day after surgery, serum miR-206 expression was significantly lower in the observation group than the control group ( t = 7.50, P < 0.001). At 4 weeks after surgery, serum miR-206 expression was significantly higher in the observation group than the control group ( t = 17.24, P < 0.001). At 1 week after surgery, there was no significant difference in serum miR-206 expression between the two groups ( P > 0.05). Univariate analysis results showed that postoperative infection, FGF-2, and miR-206 were closely related to the delayed union of closed tibial shaft fractures after surgery (all P < 0.05). Multivariate logistic regression analysis results showed that postoperative infection ( OR = 1.93, 95% CI: 1.20-3.07), FGF-2 ( OR = 2.10, 95% CI: 1.31-3.36), miR-206 ( OR = 2.30, 95% CI: 1.35-3.89) were independent risk factors for delayed union of closed tibial shaft fractures after surgery (all P < 0.05). The receiver operating characteristic (ROC) curves plotting serum FGF-2 level and serum miR-206 expression after closed tibial shaft fractures showed that at 4 weeks after surgery, the optimal cut-off value of FGF-2 for predicting delayed union of closed tibial shaft fractures was 29.83 ng/L, with the area under the curve, sensitivity, and specificity of 0.76 (95% CI: 1.23-3.25), 79.34%, and 68.82%, respectively; at 4 weeks after surgery, the optimal cut-off value of miR-206 for predicting delayed union of closed tibial shaft fractures was 0.63, with the area under the curve, sensitivity and specificity of 0.72 (95% CI: 1.10-2.45), 75.33%, and 67.25%, respectively. The area under the curve, the sensitivity, and specificity of combined use of FGF-2 and miR-206 in predicting delayed union of closed tibial shaft fractures were 0.81 (95% CI: 1.35-3.26), sensitivity and specificity were 83.45% and 67.36% respectively. Conclusion:The decrease in serum FGF-2 level and the increase in serum miR-206 expression at 4 weeks after surgery are independent risk factors for delayed union of closed tibial shaft fractures. Combined use of FGF-2 and miR-206 can better predict the delayed union of closed tibial shaft fractures.

Analysis on optimal cut-off values for prediction of lumbar spinal stenosis using lumbar pedicle thickness / 中国基层医药

Objective:To investigate the optimal cut-off values for the prediction of lumbar spinal stenosis using lumbar pedicle thickness.Methods:The clinical data of 64 patients with lumbar spinal stenosis (patient group) admitted to Binzhou Center Hospital from November 2019 to April 2021 and 48 healthy volunteers (healthy control group) who concurrently received routine physical examination involving lumbar spine MRI examination in the same hospital were retrospectively analyzed. Lumbar pedicle thickness was measured on T 2 weighted images of the L 5 vertebral body in the axial projection. Lumbar pedicle thickness was compared between groups using the independent sample t-test. The relationship between lumbar pedicle thickness and age change was analyzed using a one-way analysis of variance. The efficacy of lumbar pedicle thickness in the diagnosis of lumbar spinal stenosis was evaluated using the receiver operating characteristic (ROC) curve, optimal cut-off values, sensitivity, specificity, and the area under the ROC curve. Results:There was no significant correlation between lumbar pedicle thickness and age change ( P > 0.05). Lumbar pedicle thickness of patients with lumbar spinal stenosis was significantly higher than that of healthy controls [(13.25 ± 1.73) mm vs. (8.54 ± 1.88) mm, t = 13.75, P < 0.05]. ROC curve results showed that the optimal cut-off value was 10.50 mm, with a sensitivity of 95.3% and a specificity of 85.4. The area under the ROC curve was 0.963 (95% CI 0.928-0.998). Conclusion:The increase in lumbar pedicle thickness is related to the increase in the incidence of lumbar spinal stenosis. Lumbar pedicle thickness is an accurate, objective, and clear morphological parameter for the prediction of lumbar spinal stenosis. Application of lumbar pedicle thickness to predict lumbar spinal stenosis is innovative and scientific.

The value of 18F-FDG PET-CT in predicting relapse of type 1 autoimmune pancreatitis / 中华胰腺病杂志

Objective:To investigate the predictive value of F-2-fluoro-2-deoxy-D-glucose ( 18F-FDG) PET-CT metabolic parameters for the recurrence of type 1 autoimmune pancreatitis (AIP). Methods:Eighty-six patients with type 1 AIP who met the International Consensus Diagnostic Criteria (ICDC) and underwent 18F-FDG PET-CT before interventional treatment at the PLA General Hospital between May 2009 and June 2021 were included and divided into recurrence group ( n=43) and no-recurrence group ( n=43) according to whether they recurred after treatment. The standard uptake value (SUV)≥2.5 fixed threshold was used to outline the pancreatic lesion volume of interest (VOI) in three dimensions, and the three-dimensional diameter of the lesion, maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), peak standardized uptake value (SUVpeak), metabolic tumor volume (MTV), total lesion glycolysis (TLG), target-to-bench ratio (TBR) and standardized uptake value ratio (SUVR) were measured to compare the clinical characteristics, biochemical indices and treatment of the two groups; univariate and multifactorial regression analysis were used to examine 18F-FDG PET/CT visual indices of pancreatic lesions and extra-pancreatic involved organs as well as metabolic parameters in the two groups. A recurrence prediction model was constructed and its predictive efficacy was assessed. Results:The proportion of patients receiving glucocorticoid maintenance therapy was significantly higher in the no-recurrence group than in the recurrence group (58% vs 23.3%), and the serum IgG4 levels before treatment were significantly higher in the recurrence group [(15 309±11 724) mg/L vs (8 816±7 169) mg/L]. The results of univariate analysis showed that the proportion of extra-pancreatic salivary gland involvement and VOI, SUVmax, SUVpeak, SUVR, TBR, MTV, and TLG were significantly higher in the recurrence group than in the no-recurrence group, and the differences were statistically significant (all P values <0.05); the results of multivariate analysis showed that VOI ( OR=1.012, 95% CI 1.001-1.023 ), SUV max ( OR=1.398, 95% CI 1.029-1.899), SUV peak ( OR=1.408, 95% CI1.002-1.978), SUVR ( OR=1.977, 95% CI1.036-3.771) and MTV ( OR=1.012, 95% CI1.000- 1.022) in the recurrence group were significantly higher than those in the no-recurrence group, and all differences were statistically significant (all P values <0.05). The prediction model was constructed by multifactorial binary logistic regression analysis of SUVR>2, MTV>36 cm 3, and IgG4>11 400 mg/L, which had an AUC of 0.800 (95% CI 0.704-0.897), sensitivity of 81.4% (95% CI 0.661-0.911), specificity of 74.4% (95% CI 0.585-0.860), and prediction accuracy of 77.9%. Conclusions:18F-FDG PET/CT metabolic parameters can be used as predictors of type 1 AIP recurrence; a multiparameter model constructed based on metabolic parameters SUVR, MTV and IgG4 has a good predictive efficacy for predicting type 1 AIP recurrence.

The predictive value of renal artery resistance index combined with cystatin C in contrast-induced nephropathy after coronary artery CT angiography / 中国医师进修杂志

Objective:To investigate the predictive value of serum cystatin C (Cys-C) and renal artery resistance index (RRI) 24 h before coronary CT angiography (CTA) examination in contrast-induced nephropathy(CIN).Methods:Sixty-four patients with coronary heart disease who received coronary CTA examination in Hebei Petro China Central Hospital from January 2020 to March 2021 were selected as the research subjects. According to the incidence of CIN after coronary CTA examination, they were divided into the case group (25 patients) and the normal group(39 patients). Serum Cys-C level was measured by automatic biochemical analyzer at 24 h before CTA examination, and RRI value was measured by color Doppler ultrasound. Risk factors of CIN after CTA examination were analyzed by Logistic regression. The receiver operating characteristic (ROC) curve was drawn to evaluate the predictive value of serum Cys-C, RRI and the combination of the two indexes in CIN.Results:Compared with the control group, the dosage of contrast agent, patients with hypertension, serum Cys-C level at 24 h before CTA examination and RRI value in the case group were higher than those in the normal group: (85.53 ± 16.27) ml vs. (64.37 ± 15.08) ml, 80.00%(20/25) vs. 56.41%(22/39), (1.36 ± 0.18)mg/L vs.(1.02 ± 0.21) mg/L, 0.743 ± 0.093 vs. 0.632 ± 0.081, there were statistical differences ( P<0.05). Multivariate Logistic regression analysis showed that the amount of contrast agent, hypertension, serum Cys-C level at 24 h before CTA examination and RRI value were independent risk factor for CIN after CTA examination ( P<0.05). The results of ROC curve analysis showed that serum Cys-C (>1.318 mg/L) combined with RRI value (>0.653) at 24 h before CTA examination predicted CIN with the highest area under the curve was 0.922, sensitivity was 92.5% and specificity was 81.6%. Conclusions:The incidence of CIN after CTA is related to the dosage of contrastant, hypertension, serum Cys-C level and RRI value at 24 h before CTA. The combination of Cys-C level and RRI value has a high predictive value for the occurrence of CIN.