A groundbreaking study has unveiled a new method that greatly improves the precision of epidemiological estimates for infectious diseases such as COVID-19.
A groundbreaking study has unveiled a new method that greatly improves the precision of epidemiological estimates for infectious diseases such as COVID-19. This study, named “Overcoming Bias in Estimating Epidemiological Parameters with Realistic History-Dependent Disease Spread Dynamics,” appeared recently in Nature Communications.
The research team, led by Professor KIM Jae Kyoung from KAIST and Chief Investigator of the Biomedical Mathematics Group at the Institute for Basic Science (IBS), along with Dr. CHOI Sunhwa from the National Institute for Mathematical Sciences (NIMS), and Professor CHOI Boseung from Korea University, tackled a longstanding issue in infectious disease modeling. Traditionally, models have relied on history-independent dynamics, which presume that transitioning between various disease stages occurs with a constant probability, regardless of the time elapsed since exposure. This method can introduce significant bias when estimating vital parameters like the reproduction number (R), latent period, and infectious period.
In contrast, the new technique developed by the team employs a history-dependent model, where the chances of moving between stages of the disease vary with time. This realistic approach corrects the biases seen in older methods and enables more precise forecasts of disease transmission, even when based only on confirmed case data. This is essential for assessing the efficacy of intervention measures such as social distancing and vaccination efforts.
Professor KIM Jae Kyoung stated, “Our research signifies a major shift in how epidemiological parameters are estimated. By overcoming previous models’ limitations, we can now provide health officials with more accurate data regarding disease dynamics. This will foster more effective intervention strategies, ultimately aiding in better management of infectious disease outbreaks.”
Dr. CHOI Boseung from Korea University, a corresponding author of the study, noted, “The new method allows for accurate estimation of the infectious period distribution, even as this period fluctuates due to different intervention measures and the disease’s evolution. This adaptability in parameter estimation was not feasible with traditional models. Our findings will considerably influence how epidemiologists and health authorities respond to future pandemics.”
Using early COVID-19 outbreak data from Seoul, South Korea, the team showed that their method yields far more accurate estimates of the reproduction number compared to conventional techniques. They discovered that traditional approaches might overstate the reproduction number by as much as twofold, which could lead to misinformed policy choices.
Dr. CHOI Sunhwa emphasized, “This research represents a significant leap in our understanding of infectious disease dynamics. The new method provides health officials with more trustworthy data, fostering better-informed decisions during pandemics.”
The team also created a user-friendly computational package called IONISE (Inference Of Non-Markovian SEIR model), which simplifies the application of their sophisticated inference method. IONISE accommodates various epidemiological models, making it versatile for different infectious diseases and intervention contexts.
Dr. HONG Hyukpyo believes that this methodology will transform the arena of infectious disease modeling and epidemiological parameter estimation, setting the stage for more effective public health responses and strategies in future pandemics.
About the Research Team
The study was conducted by a collaborative research team from the Department of Mathematical Sciences at KAIST, the Biomedical Mathematics Group at IBS, NIMS, and the Division of Big Data Science at Korea University. With profound expertise in mathematical modeling and epidemiology, the team strives to solve critical challenges in predicting and controlling infectious diseases through advanced mathematical frameworks and computational methods.
The research received financial backing from grants provided by the National Research Foundation of Korea, the Ministry of Education, the Samsung Science and Technology Foundation, and the Institute for Basic Science.
