What is it about?

This study uses machine learning techniques to explore the key risk factors associated with COVID-19 mortality in Mexico. Using a national database of confirmed COVID-19 cases from February 2020 to April 2022, we developed predictive models of mortality using the Extreme Gradient Boosting (XGBoost) algorithm. We analyzed the models using Shapley values (SHAP) to identify the most influential characteristics contributing to patient mortality in the first four epidemiologic waves. Our results highlight pneumonia, advanced age, and medical unit type as the most significant factors influencing COVID-19 outcomes.

Featured Image

Why is it important?

Understanding the risk factors associated with COVID-19 mortality is critical for improving public health interventions and healthcare resource allocation. This study provides a comprehensive analysis of how demographic variables, comorbidities, and access to health care influenced patient outcomes throughout the pandemic in Mexico. By identifying high-risk populations, policymakers can develop targeted strategies to mitigate mortality in future health crises. In addition, the use of machine learning techniques improves the accuracy and interpretability of predictive models, making them valuable tools for epidemiological assessment.

Perspectives

This study demonstrates the potential of decision tree-based algorithms such as XGBoost in handling large epidemiological data. The use of SHAP values for explainability ensures that models are not only accurate but also interpretable, addressing one of the key concerns in medical AI applications. Future research could explore deep learning techniques to further improve performance and integrate additional environmental or genetic data to refine risk assessment. In addition, addressing class imbalance in mortality prediction remains a challenge, highlighting the need for more robust sampling and augmentation techniques in health-related machine learning models.

Lorena Díaz-González
Universidad Autonoma del Estado de Morelos

Read the Original

This page is a summary of: Risk Factors Associated with COVID-19 Lethality: A Machine Learning Approach Using Mexico Database, Journal of Medical Systems, August 2023, Springer Science + Business Media,
DOI: 10.1007/s10916-023-01979-4.
You can read the full text:

Read

Resources

Contributors

The following have contributed to this page