Application of machine learning methods on SARS-Cov-2: Mortality prediction by using health and nutritional factors and prediction of recurrent mutations

Identificador:  TDX:4416

Handle: https://hdl.handle.net/20.500.11797/TDX4416

Autores:  Saldivar Espinoza, Bryan Percy

Resumen:
In 2019 we witnessed the emergence of a new pandemic that has made society, healthcare systems and economy to tremble worldwide, unveiling how unprepared we were in terms of readiness, knowledge and protocols to minimize its negative impact. The pandemic was caused by the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), a positive single stranded RNA betacoronavirus. During the pandemic, it was seen that the disease COVID-19 affected people differently, depending on health, socioeconomic, nutritional and other factors. In addition, the virus showed an elevated propensity to mutate, generating uncertainty about the efficacy of treatments to fight it. In this regard, this thesis uses Machine Learning models to analyze the main factors affecting COVID-19 mortality and predict SARS-CoV-2 recurrent mutations. The predictive model developed for COVID-19 mortality at the US county level integrates health, socioeconomic, and nutritional data, achieving a notable correlation of 0.715. The analysis of influential variables revealed that the proportion of primary care physicians and other health providers relative to the population, along with socioeconomic indicators such as median household income and rates of physical inactivity and poverty, significantly impact COVID-19 mortality rates