Remote sensing can contribute to early warning for diseases with environmental drivers, such as flooding for leptospirosis. In this study we assessed whether and which remotely-sensed flooding indicator could be used in Cambodia to study any disease for which flooding has already been identified as an important driver, using leptospirosis as a case study. The performance of six potential flooding indicators was assessed by ground truthing. The Modified Normalized Difference Water Index (MNDWI) was used to estimate the Risk Ratio (RR) of being infected by leptospirosis when exposed to floods it detected, in particular during the rainy season. Chi-square tests were also calculated. Another variable—the time elapsed since the first flooding of the year—was created using MNDWI values and was also included as explanatory variable in a generalized linear model (GLM) and in a boosted regression tree model (BRT) of leptospirosis infections, along with other explanatory variables. Interestingly, MNDWI thresholds for both detecting water and predicting the risk of leptospirosis seroconversion were independently evaluated at -0.3. Value of MNDWI greater than -0.3 was significantly related to leptospirosis infection (RR = 1.61 [1.10–1.52]; χ2 = 5.64, p-value = 0.02, especially during the rainy season (RR = 2.03 [1.25–3.28]; χ2 = 8.15, p-value = 0.004). Time since the first flooding of the year was a significant risk factor in our GLM model (p-value = 0.042). These results suggest that MNDWI may be useful as a risk indicator in an early warning remote sensing tool for flood-driven diseases like leptospirosis in South East Asia.

Age-stratified serosurvey data are often used to understand spatiotemporal trends in disease incidence and exposure through estimating the Force-of-Infection (FoI). Typically, median or mean FoI estimates are used as the response variable in predictive models, often overlooking the uncertainty in estimated FoI values when fitting models and evaluating their predictive ability. To assess how this uncertainty impact predictions, we compared three approaches with three levels of uncertainty integration. We propose a performance indicator to assess how predictions reflect initial uncertainty.In Colombia, 76 serosurveys (1980–2014) conducted at municipality level provided age-stratified Chagas disease prevalence data. The yearly FoI was estimated at the serosurvey level using a time-varying catalytic model. Environmental, demographic and entomological predictors were used to fit and predict the FoI at municipality level from 1980 to 2010 across Colombia.A stratified bootstrap method was used to fit the models without temporal autocorrelation at the serosurvey level. The predictive ability of each model was evaluated to select the best-fit models within urban, rural and (Amerindian) indigenous settings. Model averaging, with the 10 best-fit models identified, was used to generate predictions.Our analysis shows a risk of overconfidence in model predictions when median estimates of FoI alone are used to fit and evaluate models, failing to account for uncertainty in FoI estimates. Our proposed methodology fully propagates uncertainty in the estimated FoI onto the generated predictions, providing realistic assessments of both central tendency and current uncertainty surrounding exposure to Chagas disease.

Dengue is a national priority disease in Cambodia. The Cambodian National Dengue Surveillance System is based on passive surveillance of dengue-like inpatients reported by public hospitals and on a sentinel, pediatric hospital-based active surveillance system. This system works well to assess trends but the sensitivity of the early warning and time-lag to usefully inform hospitals can be improved. During The ECOnomic development, ECOsystem MOdifications, and emerging infectious diseases Risk Evaluation (ECOMORE) project’s knowledge translation platforms, Cambodian hospital staff requested an early warning tool to prepare for major outbreaks. Our objective was therefore to find adapted tools to improve the early warning system and preparedness. Dengue data was provided by the National Dengue Control Program (NDCP) and are routinely obtained through passive surveillance. The data were analyzed at the provincial level for eight Cambodian provinces during 2008–2015. The R surveillance package was used for the analysis. We evaluated the effectiveness of Bayesian algorithms to detect outbreaks using count data series, comparing the current count to an expected distribution obtained from observations of past years. The analyses bore on 78,759 patients with dengue-like syndromes. The algorithm maximizing sensitivity and specificity for the detection of major dengue outbreaks was selected in each province. The overall sensitivity and specificity were 73% and 97%, respectively, for the detection of significant outbreaks during 2008–2015. Depending on the province, sensitivity and specificity ranged from 50% to 100% and 75% to 100%, respectively. The final algorithm meets clinicians’ and decisionmakers’ needs, is cost-free and is easy to implement at the provincial level.

In Cambodia, dengue outbreaks occur each rainy season (May–October) but vary in magnitude. Using national surveillance data, we designed a tool that can predict 90% of the variance in peak magnitude by April, when typically <10% of dengue cases have been reported. This prediction may help hospitals anticipate excess patients.

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.