Linear and Machine Learning modelling for spatiotemporal disease predictions: Force-of-Infection of Chagas disease

Abstract

Background Chagas disease is a long-lasting disease with a prolonged asymptomatic period. Cumulative indices of infection such as prevalence do not shed light on the current epidemiological situation, as they integrate infection over long periods. Instead, metrics such as the Force-of-Infection (FoI) provide information about the rate at which susceptible people become infected and permit sharper inference about temporal changes in infection rates. FoI is estimated by fitting (catalytic) models to available age-stratified serological (ground-truth) data. Predictive FoI modelling frameworks are then used to understand spatial and temporal trends indicative of heterogeneity in transmission and changes effected by control interventions. Ideally, these frameworks should be able to propagate uncertainty and handle spatiotemporal issues. Methodology/principal findings We compare three methods in their ability to propagate uncertainty and provide reliable estimates of FoI for Chagas disease in Colombia as a case study: two Machine Learning (ML) methods (Boosted Regression Trees (BRT) and Random Forest (RF)), and a Linear Model (LM) framework that we had developed previously. Our analyses show consistent results between the three modelling methods under scrutiny. The predictors (explanatory variables) selected, as well as the location of the most uncertain FoI values, were coherent across frameworks. RF was faster than BRT and LM, and provided estimates with fewer extreme values when extrapolating to areas where no ground-truth data were available. However, BRT and RF were less efficient at propagating uncertainty. Conclusions/significance The choice of FoI predictive models will depend on the objectives of the analysis. ML methods will help characterise the mean behaviour of the estimates, while LM will provide insight into the uncertainty surrounding such estimates. Our approach can be extended to the modelling of FoI patterns in other Chagas disease-endemic countries and to other infectious diseases for which serosurveys are regularly conducted for surveillance.

From serological surveys to disease burden: a modelling pipeline for Chagas disease

“…The process of estimating the burden of Chagas disease using local serosurveys (figure 1) involves three steps: (1) Step 1: Local seroprevalence information is used to estimate local trends in temporal exposure (quantified by the FoI). (2) Step 2: Exposure estimates from various surveys are used to predict spatio-temporal trends across larger geographical areas (using Random Forest (RF) models, as in [20]). (3) Step 3: Predicted exposure estimates are used at a fine spatial resolution to predict disease burden based on a disease progression model.…”

“…The process of estimating the burden of Chagas disease using local serosurveys (figure 1) involves three steps: Step 1: Local seroprevalence information is used to estimate local trends in temporal exposure (quantified by the FoI).Step 2: Exposure estimates from various surveys are used to predict spatio-temporal trends across larger geographical areas (using Random Forest (RF) models, as in [20]).Step 3: Predicted exposure estimates are used at a fine spatial resolution to predict disease burden based on a disease progression model.

Figure 1Modelling pipeline, from local serosurveys to sub-national yearly disease burden estimates. Using local point seroprevalence age-stratified data ( a ), the modelling pipeline uses a Force-of-Infection (FoI) catalytic model ( b ) to estimate yearly FoI local point estimates ( c ).

…”

“…Step 2: Exposure estimates from various surveys are used to predict spatio-temporal trends across larger geographical areas (using Random Forest (RF) models, as in [ 20 ]).…”

“…These age-stratified serosurveys were used to fit a Bayesian catalytic model and obtain yearly estimates of the local FoI (in the ‘catchment areas’), from the birth of the oldest participant to the year of the survey. (We refer to those municipalities where at least one serosurvey was conducted as municipalities ‘in catchment areas’ [ 20 ].) Electronic supplementary material, text S1.1–S1.5 describes the catalytic models used.…”

“…outside the catchment areas), to obtain yearly FoI predictions across the entire country, at the municipality level (electronic supplementary material, text S2). Informed by previous studies [17,20], and in order to build a pipeline that could be applied in other countries, the predictors selected are available across Latin America and include characteristics of the serosurveys as well as spatio-temporal, climatic, environmental, demographic and socio-economic predictors as described below:…”