Development of a Measurement System Using Infrared Spectroscopy-Attenuated Total Reflectance, Principal Component Analysis and Artificial Intelligence for the Safe Quantification of the Nucleating Agent Sorbitol in Food Packaging.

Sorbitol derivatives and other additives are commonly used in various products, such as packaging or food packaging, to improve their mechanical, physical, and optical properties. To accurately and precisely evaluate the efficacy of adding sorbitol-type nucleating agents to these articles, their quantitative determination is essential. This study systematically investigated the quantification of sorbitol-type nucleating agents in food packaging made from impact copolymers of polypropylene (PP) and polyethylene (PE) using attenuated total reflectance infrared spectroscopy (ATR-FTIR) together with analysis of principal components (PCA) and machine learning algorithms. The absorption spectra revealed characteristic bands corresponding to the C-O-C bond and hydroxyl groups attached to the cyclohexane ring of the molecular structure of sorbitol, providing crucial information for identifying and quantifying sorbitol derivatives. PCA analysis showed that with the selected FTIR spectrum range and only the first two components, 99.5% of the variance could be explained. The resulting score plot showed a clear pattern distinguishing different concentrations of the nucleating agent, affirming the predictability of concentrations based on an impact copolymer. The study then employed machine learning algorithms (NN, SVR) to establish prediction models, evaluating their quality using metrics such as RMSE, R2, and RMSECV. Hyperparameter optimization was performed, and SVR showed superior performance, achieving near-perfect predictions (R2 = 0.9999) with an RMSE of 0.100 for both calibration and prediction. The chosen SVR model features two hidden layers with 15 neurons each and uses the Adam algorithm, balanced precision, and computational efficiency. The innovative ATR-FTIR coupled SVR model presented a novel and rapid approach to accurately quantify sorbitol-type nucleating agents in polymer production processes for polymer research and in the analysis of nucleating agent derivatives. The analytical performance of this method surpassed traditional methods (PCR, NN).

Methodology for the identification of relevant loci for milk traits in dairy cattle, using machine learning algorithms.

Machine learning methods were considered efficient in identifying single nucleotide polymorphisms (SNP) underlying a trait of interest. This study aimed to construct predictive models using machine learning algorithms, to identify loci that best explain the variance in milk traits of dairy cattle. Further objectives involved validating the results by comparison with reported relevant regions and retrieving the pathways overrepresented by the genes flanking relevant SNPs. Regression models using XGBoost (XGB), LightGBM (LGB), and Random Forest (RF) algorithms were trained using estimated breeding values for milk production (EBVM), milk fat content (EBVF) and milk protein content (EBVP) as phenotypes and genotypes on 40417 SNPs as predictor variables. To evaluate their efficiency, metrics for actual vs. predicted values were determined in validation folds (XGB and LGB) and out-of-bag data (RF). Less than 4500 relevant SNPs were retrieved for each trait. Among the genes flanking them, signaling and transmembrane transporter activities were overrepresented. The models trained:•Predicted breeding values for animals not included in the dataset.•Were efficient in identifying a subset of SNPs explaining phenotypic variation. The results obtained using XGB and LGB algorithms agreed with previous results. Therefore, the method proposed could be applied for future association studies on milk traits.

Modelos estadísticos para las predicciones de la COVID-19 en Cuba

RESUMEN Introducción: Los estudios basados en modelos estadísticos juegan un papel importante para las predicciones sobre la COVID-19. Objetivo: Realizar un análisis de modelación estadística combinando 6 modelos de pronósticos para predecir la aparición de casos positivos diarios, activos y fallecidos por COVID-19 en Cuba. Métodos: Se utilizaron los datos reportados diariamente del 11 de marzo al 25 de mayo publicados en el sitio web CUBADEBATE. A los modelos propuestos se les calculó el desempeño mediante los estadísticos: MAE, RMSE, MAPE y ME así como el análisis de residuales. Resultados: Los modelos A y B dan una tendencia constante de 8 y 9 casos positivos respectivamente para el día 22 de julio. El modelo C indica una ligera disminución de los casos con 4 ese mismo día y el modelo D una tendencia al aumento con 19 casos. . El modelo E refleja un mínimo de 126 casos el día 3 de junio y luego un aumento de los casos hasta alcanzar el 22 de julio un valor de 374 casos activos hospitalizados. En el modelo F se apreció una tendencia a mantenerse constante el número de fallecidos por encima de 80 casos en la primera quincena de julio. Conclusiones: Los 6 modelos estudiados cumplen con las pruebas estadísticas, de desempeño y residuales. Sus datos proporcionan un pronóstico para la COVID-2019, representando una herramienta válida.

ABSTRACT Introduction: Studies based on statistical models play an important role for predictions about COVID-19. Objective: To carry out a statistical modeling analysis combining 6 forecast models to predict the appearance of daily positive cases, active and deceased by COVID-19 in Cuba. Method: Data reported daily from March 11 to May 25 from the CUBADEBATE website were used, which were processed and analyzed. The performance of the models was calculated: Mean absolute error (MAE), root of the mean square error (RMSE), percent of mean absolute error (MAPE) and the mean error (ME) as well as the residual analysis. Results: Models A and B gave a constant trend between 8 and 9 cases of until July 22. Model C indicated a decrease in cases with 4 that same day and model D indicated a raise to 19 cases. Model E indicated a minimum of 126 cases on June E and then a raise to 374 hospitalized cases. Deceases cases had a constant tendency in deceases numbers above of 80 cases in first 15 days of July. Conclusions: The 6 models studied meet the statistical , performance and residual tests. Their data provides a forecast for COVID2019, representing a valid tool.

Predicting Posttraumatic Stress Symptom Prevalence and Local Distribution after an Earthquake with Scarce Data.

BACKGROUND: After a major earthquake, the assignment of scarce mental health emergency personnel to different geographic areas is crucial to the effective management of the crisis. The scarce information that is available in the aftermath of a disaster may be valuable in helping predict where are the populations that are in most need. OBJECTIVE: The objectives of this study were to derive algorithms to predict posttraumatic stress (PTS) symptom prevalence and local distribution after an earthquake and to test whether there are algorithms that require few input data and are still reasonably predictive. METHODS: A rich database of PTS symptoms, informed after Chile's 2010 earthquake and tsunami, was used. Several model specifications for the mean and centiles of the distribution of PTS symptoms, together with posttraumatic stress disorder (PTSD) prevalence, were estimated via linear and quantile regressions. The models varied in the set of covariates included. RESULTS: Adjusted R2 for the most liberal specifications (in terms of numbers of covariates included) ranged from 0.62 to 0.74, depending on the outcome. When only including peak ground acceleration (PGA), poverty rate, and household damage in linear and quadratic form, predictive capacity was still good (adjusted R2 from 0.59 to 0.67 were obtained). CONCLUSIONS: Information about local poverty, household damage, and PGA can be used as an aid to predict PTS symptom prevalence and local distribution after an earthquake. This can be of help to improve the assignment of mental health personnel to the affected localities. Dussaillant F , Apablaza M . Predicting posttraumatic stress symptom prevalence and local distribution after an earthquake with scarce data. Prehosp Disaster Med. 2017;32(4):357-367.

Resting energy expenditure in obese women: comparison between measured and estimated values.

Assessing energy requirements is a fundamental activity in clinical dietetic practice. The aim of this study was to investigate which resting energy expenditure (REE) predictive equations are the best alternatives to indirect calorimetry before and after an interdisciplinary therapy in Brazilian obese women. In all, twelve equations based on weight, height, sex, age, fat-free mass and fat mass were tested. REE was measured by indirect calorimetry. The interdisciplinary therapy consisted of nutritional, physical exercise, psychological and physiotherapy support during the course of 1 year. The average differences between measured and predicted REE, as well as the accuracy at the ±10 % level, were evaluated. Statistical analysis included paired t tests, intraclass correlation coefficients and Bland-Altman plots. Validation was based on forty obese women (BMI 30-39·9 kg/m2). Our major findings demonstrated a wide variation in the accuracy of REE predictive equations before and after weight loss in non-morbid, obese women. The equations reported by Harris-Benedict and FAO/WHO/United Nations University (UNU) were the only ones that did not show significant differences compared with indirect calorimetry and presented a bias <5 %. The Harris-Benedict equation provided 40 and 47·5 % accurate predictions before and after therapy, respectively. The FAO equation provided 35 and 47·5 % accurate predictions. However, the Bland-Altman analysis did not show good agreement between these equations and indirect calorimetry. Therefore, the Harris-Benedict and FAO/WHO/UNU equations should be used with caution for obese women. The need to critically re-assess REE data and generate regional and more homogeneous REE databases for the target population is reinforced.