Assessing the impact of case correction methods on the fairness of COVID-19 predictive models

Overview

This paper evaluates whether applying case correction methods to COVID-19 training data changes the fairness of county-level case prediction models across racial and ethnic groups.

Key Findings

• Case correction methods are not fairness-neutral and can either reduce or exacerbate prediction disparities.
• Correction approaches based only on death data can worsen bias in downstream forecasts.
• Auditing corrected versus official case data is important before deploying public health prediction models.

Recommended Citation

Smolyak, D., Abrar, S. M., Awasthi, N., & Frias-Martinez, V. (2026). Assessing the impact of case correction methods on the fairness of COVID-19 predictive models. EPJ Data Science. https://doi.org/10.1140/epjds/s13688-026-00656-z

Read the paper

Recommended citation: Smolyak, D., Abrar, S. M., Awasthi, N., & Frias-Martinez, V. (2026). Assessing the impact of case correction methods on the fairness of COVID-19 predictive models. EPJ Data Science. https://doi.org/10.1140/epjds/s13688-026-00656-z
Download Paper