The impact of social confinement measures on the control of COVID-19 in Portugal
Data
2022-07-12
Autores
Título da revista
ISSN da revista
Título do Volume
Editora
Resumo
Este trabalho foca-se em dois objetivos principais. A análise de
dados portugueses de hospitalização, de forma a compreender melhor a dinâmica de
hospitalização do COVID-19 e assim criar parâmetros a serem utilizados em modelação
matemática. E a construção de um aplicativo para facilitar o uso e divulgação dos modelos
desenvolvidos pela equipa do projeto COVID-19 in-CTRL.
Para a análise dos dados de internamento, os dados de Grupos de Diagnósticos
Homogéneos (DRG) portugueses, relativos a um período de um ano (março 2020 a março
2021), foram tratatos e transformados de forma a permitir uma análise exploratória
em que o tempo de permanência em hospital, tempo até a morte, taxas de letalidade,
percentagem de pacientes que necessitaram de tratamento em Unidades de Cuidados
Intensivos (UCI) e outras métricas relativas a hospitalização por COVID foram estudadas,
para pacientes UCI e não UCI, de acordo com vários fatores como a idade, o género e
a Administação Regional de Saúde (ARS) e também ao longo do tempo. Contagens de
entradas, fatalidades, ocupação e outras métricas de evolução ao longo do tempo também
foram analisadas.
Durante a pandemia COVID-19 em Portugal, o grupo COVID-19 in-CTRL teve um
importante papel na criação de modelos para prever o impacto de vários tipos de medidas
de saúde pública para prevenir a propagação da pandemia e assim ajudar as autoridades
portuguesas nas suas tomadas de decisão. Com o final do projeto o autor pretende
preservar todo o trabalho realizado ao reimplementar, na linguagem de programação R,
o motor utilizado para resolver os sistemas de equações diferenciais ordinárias (EDO) dos
modelos matemáticos, de forma a permitir que futuros trabalhos, utilizando os modelos
desenvolvidos ou criando novos modelos, apenas fornecendo o sistema de EDO do modelo,
realizem os cálculos de forma rápida e modular e, se assim o desejarem, numa população
heterogênea sem necessidade de alterar as equações diferenciais.
Também foi construída uma aplicação mais complexa na qual os usuários podem
facilmente realizar cirar cenários com os modelos fornecidos. Para utilizar o aplicativo
o usuário fornece, sequencialmente, o modelo com o qual deseja utilizar, em seguida os
valores iniciais e os parâmetros iniciais, adiciona os breakpoints desejados, sendo estes
datas em que o valor de parâmetros muda, e o aplicativo garante que a linha do tempo
é construída corretamente e valida os valores fornecidos pelo utilizador. Os resultados
podem ser expressos numa tabela onde as faixas etárias são agrupadas ou num grafico
simples. Desta forma o usuário pode facilmente criar seus próprios cenários, ver os
resultados e extrair os dados para usar noutras aplicações.
This work focuses on two main objectives. The analysis of Portuguese hospitalization data, as to better understand the disease hospitalization dynamics and create parameters to be used in mathematical modeling of COVID-19. And the construction of an application to allow the easy usage and disclosure of the models developed by the COVID-19 in-CTRL project team. For the analysis of the hospitalization data, Portuguese diagnosis-related groups (DRGs) data, corresponding to a one year period (March 2020 - March 2021), was treated and transformed to allow an exploratory analysis, in which COVID-19 hospitalization related length of stay, time until death, fatality rates, percentage of patients requiring treatment in intensive care units (ICU) and other metrics were studied, for non-ICU and ICU patients, according to factors as age, gender and health region and also over time. Counts of entries, fatalities, occupancy and other metrics evolution with time were also analysed. During the COVID-19 pandemic in Portugal, the COVID-19 in-CTRL group played an important role in creating models to predict the impact of various types of public health care measures to prevent the spread of the pandemic and help the Portuguese authorities in their decision making. With the end of the project the author intends to preserve all the work done by re-implementing, in the R programming language, the engine used to solve the mathematical models system of ordinary differential equations (ODE), in a way that allows future works with the models built or, by just providing the ODEs systems perform the calculations in a fast and modular way and, if they so wish, in a heterogeneous population, without changing the differential equations. A more complex application was also built, in which users can easily perform modulation with the provided models. To use the application, the user supplies, sequentially, the model to be used, then the initial values and initial parameters, add the desired breakpoints, that is, dates at which the value of parameters change, and the application makes sure the timeline is constructed and that the provided inputs were values. The results can be expressed in a table where age groups are grouped or not, and a plot functionality is also provided. In this way, the user can easily create their own scenarios, see the results and extract the data to use in other applications.
This work focuses on two main objectives. The analysis of Portuguese hospitalization data, as to better understand the disease hospitalization dynamics and create parameters to be used in mathematical modeling of COVID-19. And the construction of an application to allow the easy usage and disclosure of the models developed by the COVID-19 in-CTRL project team. For the analysis of the hospitalization data, Portuguese diagnosis-related groups (DRGs) data, corresponding to a one year period (March 2020 - March 2021), was treated and transformed to allow an exploratory analysis, in which COVID-19 hospitalization related length of stay, time until death, fatality rates, percentage of patients requiring treatment in intensive care units (ICU) and other metrics were studied, for non-ICU and ICU patients, according to factors as age, gender and health region and also over time. Counts of entries, fatalities, occupancy and other metrics evolution with time were also analysed. During the COVID-19 pandemic in Portugal, the COVID-19 in-CTRL group played an important role in creating models to predict the impact of various types of public health care measures to prevent the spread of the pandemic and help the Portuguese authorities in their decision making. With the end of the project the author intends to preserve all the work done by re-implementing, in the R programming language, the engine used to solve the mathematical models system of ordinary differential equations (ODE), in a way that allows future works with the models built or, by just providing the ODEs systems perform the calculations in a fast and modular way and, if they so wish, in a heterogeneous population, without changing the differential equations. A more complex application was also built, in which users can easily perform modulation with the provided models. To use the application, the user supplies, sequentially, the model to be used, then the initial values and initial parameters, add the desired breakpoints, that is, dates at which the value of parameters change, and the application makes sure the timeline is constructed and that the provided inputs were values. The results can be expressed in a table where age groups are grouped or not, and a plot functionality is also provided. In this way, the user can easily create their own scenarios, see the results and extract the data to use in other applications.
Descrição
Dissertation submeted to
UNIVERSITY OF TRÁS-OS-MONTES AND ALTO DOURO
for obtaining the degree of
MASTER
in Bioinformatics and Applications to Life Sciences, in accordance with
DR – I série–A, Decreto-Lei n.º 74/2006 from 24th of March and the
Regulation of Postgraduate Studies at UTAD
DR, 2.ª série – Deliberation n.º 2391/2007
Palavras-chave
Modelação matemática , Análise de dados