Analysis of causal relations between vaccine hesitancy for COVID-19 vaccines and ideological orientations in Brazil.

This article presents a causal inference analysis of vaccine hesitancy for Coronavirus disease 2019 (COVID-19) vaccines based on socio-demographic data obtained via questionnaires applied to a sample of the Brazilian population. This data includes the respondents' political preferences, age group, education, salary range, country region, sex, believing fake news, vaccine confidence, and intention to get the COVID-19 vaccine. The research created a causal graph using these variables, seeking to answer questions about the probability of people getting vaccinated. The results of this research corroborate findings observed in the literature, also presenting unique findings: (i) The perception that the vaccine is safe is positively affected by age group and negatively by religion; (ii) The older the person, the greater the probability of considering the vaccine safe and, consequently, of getting vaccinated; (iii) The religion variable showed great importance in the model since it has a simultaneous causal effect on political preferences and the perception of vaccine safety; (iv) The data reveal that the probability of a person accepting the vaccination against COVID-19 is reduced given the fact that they believe fake news related to the vaccine. The methodology applied in this research can be replicated for populations from other countries so that it is possible to generate customized models. General causal models can be helpful for agencies dealing with vaccine hesitancy to decide which variables should be addressed to reduce this phenomenon.

Uso de redes neurais para a predição de diagnóstico de AVE: uma revisão sistemática / Use of neural networks for stroke diagnosis prediction: a systematic review neural networks and stroke prediction

Fundamentos: O Acidente Vascular Encefálico (AVE) é uma síndrome de déficit neurológico agudo atribuído à lesão vascular do Sistema Nervoso (SN). As técnicas de Inteligência Artificial (IA) na Medicina ­ como algoritmos de Redes Neurais Artificiais (RNAs) ­ têm ajudado na tomada de decisões clínicas voltadas para essa condição. Objetivo: o objetivo desta revisão será avaliar como as redes neurais artificiais estão sendo utilizadas para a predição de diagnóstico de AVE. Métodos: Trata-se de uma revisão sistemática de artigos indexados nas bases de dados PubMed, BVS, SciELO, Cochrane e SpringerLink, entre janeiro e fevereiro de 2022. Os critérios de inclusão e filtros para esse trabalho foram: artigos relacionados ao tema, estudos randomizados, coorte e ensaios clínicos, trabalhos em humanos, realizados nos últimos 5 anos, apenas nos idiomas Português, Inglês e Espanhol e com texto completo disponível gratuitamente. Os parâmetros de exclusão foram: artigos duplicados, fuga ao tema, artigos de revisão e trabalhos que não preenchiam todos os critérios de inclusão. Resultados: As RNAs estão sendo utilizadas, principalmente, para avaliação de áreas de lesões isquêmicas e hemorrágicas por métodos de segmentação e os exames mais utilizados para a modelagem dos programas têm sido Ressonância Magnética (RM) e Tomografia Computadorizada (TC). Além da TC e RM, a angiorressonância e angiotomografia também estão sendo utilizadas para o modelamento do algoritmo e são úteis por apresentarem maior sensibilidade para detecção de infartos. Conclusão: Algoritmos de segmentação e classificação aplicados nas RNAs fazem parte da medicina personalizada e servem de base para médicos na prática clínica.

Background: Stroke is an acute neurological deficit syndrome attributed to vascular injury to the Nervous System (NS). Artificial Intelligence (AI) techniques in Medicine ­ such as Artificial Neural Networks (ANNs) algorithms ­ have helped in making clinical decisions aimed at this condition. Objective: the objective of this review will be to evaluate how artificial neural networks are being used to predict the diagnosis of stroke. Methods: This is a systematic review of articles indexed in PubMed, VHL, SciELO, Cochrane and SpringerLink databases, between January and February 2022. The inclusion criteria and filters for this work were: articles related to the topic, studies randomized, cohort and clinical trials, studies in humans, carried out in the last 5 years, only in Portuguese, English and Spanish and with full text available free of charge. The exclusion parameters were: duplicate articles, escape from the topic, review articles and works that did not meet all the inclusion criteria. Results: ANNs are being used mainly for the evaluation of areas of ischemic and hemorrhagic lesions by segmentation methods and the most used exams for modeling the programs have been Magnetic Resonance Imaging (MRI) and Computed Tomography (CT). In addition to CT and MRI, magnetic resonance angiography and tomography angiography are also being used to model the algorithm and are useful because they have greater sensitivity for detecting infarctions. Conclusion: Segmentation and classification algorithms applied in ANNs are part of personalized medicine and serve as a basis for physicians in clinical practice.