Comparing and linking machine learning and semi-mechanistic models for the predictability of endemic measles dynamics

Abstract

Measles is one the best-documented and most-mechanistically-studied non-linear infectious disease dynamical systems. However, systematic investigation into the comparative performance of traditional mechanistic models and machine learning approaches in forecasting the transmission dynamics of this pathogen are still rare. Here, we compare one of the most widely used semi-mechanistic models for measles (TSIR) with a commonly used machine learning approach (LASSO), comparing performance and limits in predicting short to long term outbreak trajectories and seasonality for both regular and less regular measles outbreaks in England and Wales (E&W) and the United States. First, our results indicate that the proposed LASSO model can efficiently use data from multiple major cities and achieve similar short-to-medium term forecasting performance to semi-mechanistic models for E&W epidemics. Second, interestingly, the LASSO model also captures annual to biennial bifurcation of measles epidemics in E&W caused by susceptible response to the late 1940s baby boom. LASSO may also outperform TSIR for predicting less-regular dynamics such as those observed in major cities in US between 1932–45. Although both approaches capture short-term forecasts, accuracy suffers for both methods as we attempt longer-term predictions in highly irregular, post-vaccination outbreaks in E&W. Finally, we illustrate that the LASSO model can both qualitatively and quantitatively reconstruct mechanistic assumptions, notably susceptible dynamics, in the TSIR model. Our results characterize the limits of predictability of infectious disease dynamics for strongly immunizing pathogens with both mechanistic and machine learning models, and identify connections between these two approaches.

Informing policy via dynamic models: Cholera in Haiti

“…More broadly, they have useful roles throughout biology, especially when combined with statistical methods that properly account for stochasticity and nonlinearity [ 58 ]. In some situations, modern machine learning methods can outperform mechanistic models on epidemiological forecasting tasks [ 59 , 60 ]. The predictive skill of non-mechanistic models can reveal limitations in mechanistic models, but cannot readily replace the scientific understanding obtained by describing the biological dynamics of the system in a mathematical model [ 60 , 61 ].…”

Neural networks for endemic measles dynamics: comparative analysis and integration with mechanistic models

“…However, in general, these models have not primarily focused on generating long-term forecasting accuracy. While machine learning models, including a recent work leveraging the Least Absolute Shrinkage and Selection Operator (LASSO) [15], have shown improved forecasting skills for endemic measles dynamics, they generally lack deep mechanistic interpretability. These models also do not explicitly consider spatial interactions between locations which is a known driver for measles transmission, particularly, between less populous locations (e.g., small towns) and population centers (e.g., core cities) [16].…”

“…Following the widespread May 27, 2024 2/18 vaccination in the late 1960s, the epidemics shifted from highly regular cycles to largely irregular dynamics [12]. Due to its simple natural history and long time series of data, measles incidence in England and Wales has provided a fruitful testing ground for better understanding spatiotemporal nonlinear epidemiological dynamics, and developing semi-mechanistic statistical modeling approaches more broadly [13][14][15][16][17][18].…”