Integration of a Kalman filter in the geographically weighted regression for modeling the transmission of hand, foot and mouth disease

Abstract

Background: Hand, foot and mouth disease (HFMD) is a common infectious disease whose mechanism of transmission continues to remain a puzzle for researchers. The measurement and prediction of the HFMD incidence can be combined to improve the estimation accuracy, and provide a novel perspective to explore the spatiotemporal patterns and determinant factors of an HFMD epidemic. Methods: In this study, we collected weekly HFMD incidence reports for a total of 138 districts in Shandong province, China, from May 2008 to March 2009. A Kalman filter was integrated with geographically weighted regression (GWR) to estimate the HFMD incidence. Spatiotemporal variation characteristics were explored and potential risk regions were identified, along with quantitatively evaluating the influence of meteorological and socioeconomic factors on the HFMD incidence. Results: The results showed that the average error covariance of the estimated HFMD incidence by district was reduced from 0.3841 to 0.1846 compared to the measured incidence, indicating an overall improvement of over 50% in error reduction. Furthermore, three specific categories of potential risk regions of HFMD epidemics in Shandong were identified by the filter processing, with manifest filtering oscillations in the initial, local and long-term periods, respectively. Amongst meteorological and socioeconomic factors, the temperature and number of hospital beds per capita, respectively, were recognized as the dominant determinants that influence HFMD incidence variation. Conclusions: The estimation accuracy of the HFMD incidence can be significantly improved by integrating a Kalman filter with GWR and the integration is effective for exploring spatiotemporal patterns and determinants of an HFMD epidemic. Our findings could help establish more accurate HFMD prevention and control strategies in Shandong. The present study demonstrates a novel approach to exploring spatiotemporal patterns and determinant factors of HFMD epidemics, and it can be easily extended to other regions and other infectious diseases similar to HFMD.

Spatiotemporal cluster patterns of hand, foot, and mouth disease at the province level in mainland China, 2011–2018

“…Moreover, compared with population density and GDP per capita, hospital beds per 10,000 persons ha a greater explanatory power for HFMD incidence. The result is consistent with the findings in Shandong Province [ 23 ]. It suggests that the healthcare level may play an essential role in controlling the spread of HFMD.…”

“…Most of the studies were conducted in a single province or local area, while the influence of meteorological and socioeconomic factors is more complex in the vast area of mainland China. In addition, a study found that the number of hospital beds per capita was also one of the dominant determinants that influence HFMD incidence in Shandong [ 23 ]. The number of hospital beds per capita and the number of health technicians per capita are critical indicators to measure the health resources of a region, but most studies ignored the influence of this factor.…”

Using a population-based Kalman estimator to model the COVID-19 epidemic in France: estimating associations between disease transmission and non-pharmaceutical interventions

“…This is a major challenge because the available data are very sparse and noisy, and the parametric form of transmission rates is also unknown. By assimilating data across multiple geographic regions and coupling public data with a dynamic mechanistic model, smooth transmission rates can be estimated using a Kalman filter approach (32; 33) – as already used in epidemiology for COVID-19 spread (34; 35; 36) or for other epidemics with regional variability (37). More specifically, we develop a sophisticated method to address this difficult problem based on two important methodological innovations: (1) in the model with the introduction of time-varying dynamics for the transmission rate, including a Wiener process that accounts for modeling errors, and (2) in the way the population is integrated, as we use a new method from the Kalman filter that is compatible with population approaches.…”

Analyzing spatial variations of heart disease and type-2 diabetes: A multi-scale geographically weighted regression approach

“…It explores the spatial changes and related driving factors of the research object in a certain scale by establishing a local regression equation of each point in the spatial range. GWR has been utilized to understand how spatial determinants vary across space in health and diseaserelated researches, such as targeting the spatial context of obesity determinants (Oshan et al 2020) and modeling the transmission of hand, foot, and mouth disease (Hu et al 2020). Some of the research work also focused on using GWR in modeling diabetes and heart disease.…”

“…While much literature focuses on the applications of GWR in diabetes and heart disease (Hu et al, 2020), none studied the spatiotemporal variation of leading factors that may cause the diseases. Furthermore, the traditional GWR method assumes that all of the regression processes are in the same spatial scale, constraining local relationships within each model to vary at the same spatial scale (Yang 2014).…”