A Holistic Approach to Comparing Infectious Disease Forecasting Methods

• Activity Number: 140 - Disease Outbreak and Modeling Applications in Defense and National Security
• Type: Topic Contributed
• Date/Time: Monday, August 8, 2022 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistics in Defense and National Security
• Author(s): Karl Pazdernik* and Samuel Dixon and Ravikiran Keshava Murthy and Brent Daniel and Andrew Stevens and Lauren Charles
• Companies: Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory and Pacific Northwest National Laboratory
• Keywords: Regression Trees; Encoder-decoder; Transformer; Mechanistic models; GLARMA; COVID-19

Abstract

Accurate infectious disease forecasting can inform efforts to prevent outbreaks and mitigate adverse impacts. This study compares the performance of statistical, mechanistic, machine learning (ML), and deep learning (DL) approaches in forecasting infectious disease incidences across different countries and time intervals. We forecasted case counts for a diverse set of diseases that cover different modes of transmission using a wide variety of features from public datasets, e.g., landscape, climate, and socioeconomic factors. We compared methods in the traditional sense, through forms of accuracy and computational cost, but also explore their interpretability and generalizability. We hope that this comparison not only provides guidance on the best methods for disease prediction, but also can become a template for how to holistically compare statistical and machine learning algorithms.