Machine Learning for Prognosis of Patients with COVID-19: An Early Days Analysis

Abstract

This work proposes a machine learning approach to predict the prognosis of patients with COVID-19. To assist in this task, a descriptive analysis and relative risk estimation were performed. In addition, the importance of variables in the perspective of machine learning algorithms was computed and discussed. The experiments were performed with large-scale nation-wide dataset from Brazil. The results reveal that the model developed was able to predict the patient's prognosis with an AUC = 0.8382. The results also point out that the chance of death is greater among patients over 60 years old, with comorbidities, and symptoms such as dyspnea and Oxygen saturation (< 95%), confirming results observed in other regions of the world.

Integration of Epidemiologic, Socioeconomic, and Sociodemographic Indicators to Predict Early COVID-19 In-Hospital Outcomes

“…In addition, most studies are based on information that characterizes the behavior of the disease during hospitalization [De Souza et al 2021, Figuerêdo et al 2021]. This information involves the use of invasive respiration, hospitalization in the Intensive Care Unit (ICU), and X-ray findings [Figuerêdo et al 2021]. This information helps identify more severe cases and facilitates the classification by machine learning models.…”

“…While the existing literature has provided valuable insights into the influence of socioeconomic and demographic indicators on COVID-19 outcomes, there remains a need for further investigation. Many studies have primarily focused on individual factors in isolation without comprehensively exploring the combined effects of multiple indicators [De Souza et al 2021, Figuerêdo et al 2021. Additionally, some studies have been conducted in specific regions, limiting the generalizability of the findings [De Souza et al 2021].…”

Towards Effective and Reliable Data-driven Prognostication: An Application to COVID-19

“…In addition, to better understand the relationship between the predictive variables and the model decisions, we enhance the machine learning decision support system with interpretability assets based on LIME technique [Ribeiro et al 2016]. It should be noted that this study is a significant extension of a previous work [Figuerêdo et al 2021]. We add new machine learning algorithms and added interpretability assets based on LIME in order to assist during the decision-making process.…”