Measuring the Response Performance of U.S. States against COVID-19 Using an Integrated DEA, CART, and Logistic...

Area of Science:

• Public Health
• Health Policy Analysis
• Data Science in Healthcare

Background:

• Assessing the effectiveness of the U.S. COVID-19 response is crucial for healthcare policy.
• Existing methods may not fully capture the complex interplay of factors influencing pandemic performance.

Purpose of the Study:

• To evaluate the efficiency of the U.S. COVID-19 response across states.
• To identify key environmental factors impacting response performance using machine learning.

Main Methods:

• Integrated Data Envelopment Analysis (DEA) with four machine learning techniques: Classification and Regression Tree (CART), Boosted Tree (BT), Random Forest (RF), and Logistic Regression (LR).
• DEA utilized inputs (tested, funding, healthcare employees, hospital beds) and outputs (recovered, confirmed cases).
• Machine learning models predicted performance based on social distancing, health policy, and socioeconomic factors.

Main Results:

• Twenty-three out of fifty U.S. states achieved efficiency, with an average score of 0.97.
• Boosted Tree (BT) and Random Forest (RF) models demonstrated superior prediction accuracy.
• Urban factors, physical inactivity, testing rates, population density, and hospital bed availability were identified as key efficiency drivers.

Conclusions:

• The study provides a robust framework for evaluating pandemic response efficiency.
• Machine learning models, particularly BT and RF, are effective tools for predicting COVID-19 response performance.
• Policy interventions should consider socioeconomic and healthcare infrastructure factors for improved pandemic preparedness.