Prediction of hospital mortality in intensive care unit patients from clinical and laboratory data: A machine learning approach.

BACKGROUND: Intensive care unit (ICU) patients demand continuous monitoring of several clinical and laboratory parameters that directly influence their medical progress and the staff's decision-making. Those data are vital in the assistance of these patients, being already used by several scoring systems. In this context, machine learning approaches have been used for medical predictions based on clinical data, which includes patient outcomes. AIM: To develop a binary classifier for the outcome of death in ICU patients based on clinical and laboratory parameters, a set formed by 1087 instances and 50 variables from ICU patients admitted to the emergency department was obtained in the "WiDS (Women in Data Science) Datathon 2020: ICU Mortality Prediction" dataset. METHODS: For categorical variables, frequencies and risk ratios were calculated. Numerical variables were computed as means and standard deviations and Mann-Whitney U tests were performed. We then divided the data into a training (80%) and test (20%) set. The training set was used to train a predictive model based on the Random Forest algorithm and the test set was used to evaluate the predictive effectiveness of the model. RESULTS: A statistically significant association was identified between need for intubation, as well predominant systemic cardiovascular involvement, and hospital death. A number of the numerical variables analyzed (for instance Glasgow Coma Score punctuations, mean arterial pressure, temperature, pH, and lactate, creatinine, albumin and bilirubin values) were also significantly associated with death outcome. The proposed binary Random Forest classifier obtained on the test set (n = 218) had an accuracy of 80.28%, sensitivity of 81.82%, specificity of 79.43%, positive predictive value of 73.26%, negative predictive value of 84.85%, F1 score of 0.74, and area under the curve score of 0.85. The predictive variables of the greatest importance were the maximum and minimum lactate values, adding up to a predictive importance of 15.54%. CONCLUSION: We demonstrated the efficacy of a Random Forest machine learning algorithm for handling clinical and laboratory data from patients under intensive monitoring. Therefore, we endorse the emerging notion that machine learning has great potential to provide us support to critically question existing methodologies, allowing improvements that reduce mortality.

Recognition of esophagitis in endoscopic images using transfer learning.

BACKGROUND: Esophagitis is an inflammatory and damaging process of the esophageal mucosa, which is confirmed by endoscopic visualization and may, in extreme cases, result in stenosis, fistulization and esophageal perforation. The use of deep learning (a field of artificial intelligence) techniques can be considered to determine the presence of esophageal lesions compatible with esophagitis. AIM: To develop, using transfer learning, a deep neural network model to recognize the presence of esophagitis in endoscopic images. METHODS: Endoscopic images of 1932 patients with a diagnosis of esophagitis and 1663 patients without any pathological diagnosis provenient from the KSAVIR and HyperKSAVIR datasets were splitted in training (80%) and test (20%) and used to develop and evaluate a binary deep learning classifier built using the DenseNet-201 architecture, a densely connected convolutional network, with weights pretrained on the ImageNet image set and fine-tuned during training. The classifier model performance was evaluated in the test set according to accuracy, sensitivity, specificity and area under the receiver operating characteristic curve (AUC). RESULTS: The model was trained using Adam optimizer with a learning rate of 0.0001 and applying binary cross entropy loss function. In the test set (n = 719), the classifier achieved 93.32% accuracy, 93.18% sensitivity, 93.46% specificity and a 0.96 AUC. Heatmaps for spatial predictive relevance in esophagitis endoscopic images from the test set were also plotted. In face of the obtained results, the use of dense convolutional neural networks with pretrained and fine-tuned weights proves to be a good strategy for predictive modeling for esophagitis recognition in endoscopic images. In addition, adopting the classification approach combined with the subsequent plotting of heat maps associated with the classificatory decision gives greater explainability to the model. CONCLUSION: It is opportune to raise new studies involving transfer learning for the analysis of endoscopic images, aiming to improve, validate and disseminate its use for clinical practice.

Screening Anti-inflammatory, Anticoagulant, and Respiratory Agents for SARS-CoV-2 3CLpro Inhibition from Chemical Fingerprints Through a Deep Learning Approach.

BACKGROUND: Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2), the etiologic agent of coronavirus disease 2019 (COVID-19), triggers a pathophysiological process linked not only to viral mechanisms of infectivity, but also to the pattern of host response. Drug repurposing is a promising strategy for rapid identification of treatments for SARS-CoV-2 infection, and several attractive molecular viral targets can be exploited. Among those, 3CL protease is a potential target of great interest. OBJECTIVE: The objective of the study was to screen potential 3CLpro inhibitors compounds based on chemical fingerprints among anti-inflammatory, anticoagulant, and respiratory system agents. METHODS: The screening was developed based on a drug property prediction framework, in which the evaluated property was the ability to inhibit the activity of the 3CLpro protein, and the predictions were performed using a dense neural network trained and validated on bioassay data. RESULTS: On the validation and test set, the model obtained area under the curve values of 98.2 and 76.3, respectively, demonstrating high specificity for both sets (98.5% and 94.7%). Regarding the 1278 compounds screened, the model indicated four anti-inflammatory agents, two anticoagulants, and one respiratory agent as potential 3CLpro inhibitors. CONCLUSIONS: Those findings point to a possible desirable synergistic effect in the management of patients with COVID-19 and provide potential directions for in vitro and in vivo research, which are indispensable for the validation of their results.

Perfil das internações por encefalite viral no Brasil em 2018 por unidade da federação, sexo e faixa etária / Profile of hospitalizations for viral encephalitis in Brazil in 2018 per state, sex, and age group

RESUMO: Introdução: A encefalite viral é uma condição com altas taxas de morbimortalidade, e um melhor entendimento de sua epidemiologia pode colaborar para a construção de estratégias de prevenção e controle. Diante disso, este estudo se propôs a traçar um perfil epidemiológico para a encefalite viral no Brasil no ano de 2018 a partir de dados de internações hospitalares no Sistema Único de Saúde (SUS). Métodos: Estudo ecológico de análise espacial. Os dados estudados foram relativos às internações hospitalares por encefalite viral no SUS em 2018, estratificadas por unidade da federação (UF), sexo e faixa etária. A distribuição geográfica foi abordada exploratoriamente, já as variáveis sexo e faixa etária foram abordadas analiticamente. Resultados: Foram registradas 2075 internações, com taxa de 0,99/105 habitantes. As taxas para cada UF foram representadas a partir de um mapa colorimétrico, enquanto as taxas para cada sexo e faixa etária foram representadas em uma tabela comparativa univariada. Discussão: Observou-se ampla variação numérica das taxas de internação dentre as UF, sendo Pernambuco o estado com maior incidência (4,13/105 habitantes) e Paraíba o estado com menor (0,29/105 habitantes). Foi constatada associação significativa com o risco de internação hospitalar por encefalite viral para o sexo masculino e para as faixas etárias de 1 a 4 anos (RR: 3,28) e menores de 1 ano (RR: 6,02). Conclusão: UF, gênero e faixa etária foram determinantes importantes da taxa de internação hospitalar por encefalite viral. Todavia, carecem de estudos atuais no Brasil e no mundo para a melhor caracterização da epidemiologia da encefalite viral. (AU)

ABSTRACT: Introduction: Viral encephalitis is a condition with high morbidity and mortality rates, and a better understanding of its epidemiology may contribute to the construction of prevention and control strategies. For this reason, this study aimed to draw an epidemiological profile for viral encephalitis in Brazil in 2018 from data on hospitalizations in the Unified Health System (SUS). Methods: Ecological study of spatial analysis. The data studied were hospi-talizations for viral encephalitis in SUS in 2018, stratified by federation unit (FU), gender, and age group. The geographical distribution was approached in an exploratory way, whereas gender and age variables were analytically addressed. Results: There were 2075 hospitalizations, with a rate of 0.99/105 inhabitants. The rates for each FU were represented in a colorimetric map, whereas the rates for each sex and age group were exemplified in a univariate comparative table. Discussion: There was a wide numerical variation in hospitalization rates among the FUs, with Pernambuco being the state with the highest incidence (4.13/105 inhabitants) and Paraíba with the lowest (0.29/105 inhabitants). A significant association was found with the risk of hospitalization for viral encephalitis for males and the ages from 1 to 4 years (RR: 3.28) and under one year (RR: 6.02). Conclusion: FU, gender, and age group were important determinants of the hospitalization rate due to viral encephalitis. However, current studies are needed in Brazil and worldwide to better characterize the epidemiology of viral encephalitis.(AU)

Study of the Phase Transitions in the Binary System NPG-TRIS for Thermal Energy Storage Applications.

Neopentylglycol (NPG) and tris(hydroxymethyl)aminomethane (TRIS) are promising phase change materials (PCMs) for thermal energy storage (TES) applications. These molecules undergo reversible solid-solid phase transitions that are characterized by high enthalpy changes. This work investigates the NPG-TRIS binary system as a way to extend the use of both compounds in TES, looking for mixtures that cover transition temperatures different from those of pure compounds. The phase diagram of NPG-TRIS system has been established by thermal analysis. It reveals the existence of two eutectoids and one peritectic invariants, whose main properties as PCMs (transition temperature, enthalpy of phase transition, specific heat and density) have been determined. Of all transitions, only the eutectoid at 392 K shows sufficiently high enthalpy of solid-solid phase transition (150-227 J/g) and transition temperature significantly different from that of the solid-state transitions of pure compounds (NPG: 313 K; TRIS: 407 K). Special attention has also been paid to the analysis of metastability issues that could limit the usability of NPG, TRIS and their mixtures as PCMs. It is proven that the addition of small amounts of expanded graphite microparticles contributes to reduce the subcooling phenomena that characterizes NPG and TRIS and solve the reversibility problems observed in NPG/TRIS mixtures.