Estimation of time-varying recovery and death rates from epidemiological data: A new approach

TL;DR

This paper introduces a novel method to estimate time-dependent recovery and death rates in epidemics using aggregate data, enhancing epidemic modeling accuracy without needing individual-level data.

Contribution

It presents a new approach employing the Nadaraya-Watson estimator to derive infection-dependent rates from aggregate epidemiological data, accommodating irregular reporting.

Findings

Validated with COVID-19 data

Applied to measles and typhoid cases

Improves epidemic progression modeling

Abstract

The time-to-recovery or time-to-death for various infectious diseases can vary significantly among individuals, influenced by several factors such as demographic differences, immune strength, medical history, age, pre-existing conditions, and infection severity. To capture these variations, time-since-infection dependent recovery and death rates offer a detailed description of the epidemic. However, obtaining individual-level data to estimate these rates is challenging, while aggregate epidemiological data (such as the number of new infections, number of active cases, number of new recoveries, and number of new deaths) are more readily available. In this article, a new methodology is proposed to estimate time-since-infection dependent recovery and death rates using easily available data sources, accommodating irregular data collection timings reflective of real-world reporting practices. The Nadaraya-Watson estimator is utilized to derive the number of new infections. This model improves the accuracy of epidemic progression descriptions and provides clear insights into recovery and death distributions. The proposed methodology is validated using COVID-19 data and its general applicability is demonstrated by applying it to some other diseases like measles and typhoid.