Publication: Patrones de resistencias cruzadas a antibióticos
Loading...
Identifiers
Publication date
2021
Defense date
2021-09
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Export
Abstract
Este estudio busca realizar un análisis en profundidad sobre un conjunto de datos extraídos de la base de datos ATLAS, que es una colección de datos recogidos por Pfizer, que almacena información sobre cepas bacterianas extraidas a miles de pacientes entre los años 2014 y 2017, y los niveles de resistencia que estas presentan a distintos anitibióticos.
Con dicho análisis, lo que se pretende es encontrar pares de antibióticos que para una bacteria determinada presenten sensibilidad colateral, fenómeno por el cuál si una especie adquiere resistencia a un antibiótico puede perderla a otro.
Este fenómeno implica que para pacientes que hayan desarrollado resistencia a distintos tratamientos, podrían existir tratamientos alternativos que revirtieran dicha resistencia, lo que podría salvar miles de vida en todo el mundo.
Para ello, basándonos en una metodología CRISP-DM, la más utilizada en las tareas de análisis de datos actualmente, analizaremos y profundizaremos en el entendimiento del negocio y los datos, y estudiaremos las cuatro bacterias con mayor número de entradas en la base de datos ATLAS y los cinco países más representativos.
Trabajando sobre un cuaderno Jupyter, realizaremos regresiones entre los distintos pares de antibióticos y analizaremos las pendientes y correlaciones entre ellos, para encontrar aquellos pares que presentan sensibilidad colateral, y el grado en que este fenómeno se presenta, representándolos en grafos y mapas de calor, para facilitar su visualización.
Realizaremos comparaciones entre los distintos países, recogiendo los resultados obtenidos y evaluando hasta qué punto se podrían desarrollar tratamientos a nivel global, o, por el contrario, si sería cada país quien debería continuar la investigación individualmente.
This study seeks to perform an in-depth analysis on a set of data extracted from the ATLAS database, which is a collection of data that stores information on bacterial strains extracted from thousands of patients between 2014 and 2017, and the levels of resistance that these present to different anitibiotics. With this analysis, the aim is to find pairs of antibiotics that for a given bacterium present collateral sensitivity, a phenomenon whereby if a species acquires resistance to one antibiotic it can lose it to another. This phenomenon implies that for patients who have developed resistance to different treatments, there could be alternative treatments that could reverse this resistance, which could save thousands of lives around the world. To this end, based on a CRISP-DM methodology, the most widely used in data analysis tasks today, we will analyze and deepen our understanding of the business and the data, and study the four bacteria with the highest number of entries in the ATLAS database and the five most representative countries. Working on a Jupyter notebook, we will perform regressions between the different pairs of antibiotics and analyze the correlations between them, to find those pairs that present collateral sensitivity, and the degree to which this phenomenon is present, representing them in graphs and heat maps, to facilitate their visualization. We will make comparisons between the different countries, collecting the results obtained and evaluating to what extent treatments could be developed at a global level, or, on the contrary, if it would be each country who should continue the research individually.
This study seeks to perform an in-depth analysis on a set of data extracted from the ATLAS database, which is a collection of data that stores information on bacterial strains extracted from thousands of patients between 2014 and 2017, and the levels of resistance that these present to different anitibiotics. With this analysis, the aim is to find pairs of antibiotics that for a given bacterium present collateral sensitivity, a phenomenon whereby if a species acquires resistance to one antibiotic it can lose it to another. This phenomenon implies that for patients who have developed resistance to different treatments, there could be alternative treatments that could reverse this resistance, which could save thousands of lives around the world. To this end, based on a CRISP-DM methodology, the most widely used in data analysis tasks today, we will analyze and deepen our understanding of the business and the data, and study the four bacteria with the highest number of entries in the ATLAS database and the five most representative countries. Working on a Jupyter notebook, we will perform regressions between the different pairs of antibiotics and analyze the correlations between them, to find those pairs that present collateral sensitivity, and the degree to which this phenomenon is present, representing them in graphs and heat maps, to facilitate their visualization. We will make comparisons between the different countries, collecting the results obtained and evaluating to what extent treatments could be developed at a global level, or, on the contrary, if it would be each country who should continue the research individually.
Description
Keywords
Análisis de datos, Antibióticos