Modeling Approaches to Predicting Persistent Hotspots in SCORE Studies for Gaining Control of Schistosomiasis Mansoni in Kenya and Tanzania

Development of prediction models to identify hotspots of schistosomiasis in endemic regions to guide mass drug administration

Schistosomiasis is a neglected tropical disease affecting over 150 million people. Hotspots of Schistosoma transmission-communities where infection prevalence does not decline adequately with mass drug administration-present a key challenge in eliminating schistosomiasis. Current approaches to identify hotspots require evaluation 2-5 y after a baseline survey and subsequent mass drug administration. Here, we develop statistical models to predict hotspots at baseline prior to treatment comparing three common hotspot definitions, using epidemiologic, survey-based, and remote sensing data. In a reanalysis of randomized trials in 589 communities in five endemic countries, a regression model predicts whether Schistosoma mansoni infection prevalence will exceed the WHO threshold of 10% in year 5 ("prevalence hotspot") with 86% sensitivity, 74% specificity, and 93% negative predictive value (NPV; assuming 30% hotspot prevalence), and a regression model for Schistosoma haematobium achieves 90% sensitivity, 90% specificity, and 96% NPV. A random forest model predicts whether S. mansoni moderate and heavy infection prevalence will exceed a public health goal of 1% in year 5 ("intensity hotspot") with 92% sensitivity, 79% specificity, and 96% NPV, and a boosted trees model for S. haematobium achieves 77% sensitivity, 95% specificity, and 91% NPV. Baseline prevalence is a top predictor in all models. Prediction is less accurate in countries not represented in training data and for a third hotspot definition based on relative prevalence reduction over time ("persistent hotspot"). These models may be a tool to prioritize high-risk communities for more frequent surveillance or intervention against schistosomiasis, but prediction of hotspots remains a challenge.

Defining schistosomiasis hotspots based on literature and shareholder interviews

The World Health Organization (WHO) recently proposed a new operational definition which designates communities with ≥10% prevalence of Schistosoma spp. infection as a persistent hotspot, when, after at least two rounds of high-coverage annual preventive chemotherapy, there is a lack of appropriate reduction. However, inconsistencies and challenges from both biological and operational perspectives remain, making the prescriptive use of this definition difficult. Here, we present a comprehensive analysis of the use of the term 'hotspot' across schistosomiasis research over time, including both literature searches and opinions from a range of stakeholders, to assess the utility and generalisability of the new WHO definition of a persistent hotspot. Importantly, we propose an updated definition based on our analyses.

Predicting future community-level ocular Chlamydia trachomatis infection prevalence using serological, clinical, molecular, and geospatial data

Trachoma is an infectious disease characterized by repeated exposures to Chlamydia trachomatis (Ct) that may ultimately lead to blindness. Efficient identification of communities with high infection burden could help target more intensive control efforts. We hypothesized that IgG seroprevalence in combination with geospatial layers, machine learning, and model-based geostatistics would be able to accurately predict future community-level ocular Ct infections detected by PCR. We used measurements from 40 communities in the hyperendemic Amhara region of Ethiopia to assess this hypothesis. Median Ct infection prevalence among children 0-5 years old increased from 6% at enrollment, in the context of recent mass drug administration (MDA), to 29% by month 36, following three years without MDA. At baseline, correlation between seroprevalence and Ct infection was stronger among children 0-5 years old (ρ = 0.77) than children 6-9 years old (ρ = 0.48), and stronger than the correlation between active trachoma and Ct infection (0-5y ρ = 0.56; 6-9y ρ = 0.40). Seroprevalence was the strongest concurrent predictor of infection prevalence at month 36 among children 0-5 years old (cross-validated R2 = 0.75, 95% CI: 0.58-0.85), though predictive performance declined substantially with increasing temporal lag between predictor and outcome measurements. Geospatial variables, a spatial Gaussian process, and stacked ensemble machine learning did not meaningfully improve predictions. Serological markers among children 0-5 years old may be an objective tool for identifying communities with high levels of ocular Ct infections, but accurate, future prediction in the context of changing transmission remains an open challenge.

Prediction of antischistosomal small molecules using machine learning in the era of big data

Schistosomiasis is a neglected tropical disease caused by helminths of the Schistosoma genus. Despite its high morbidity and socio-economic burden, therapeutics are just a handful with praziquantel being the main drug. Praziquantel is an old drug registered for human use in 1982 and has since been administered en masse for chemotherapy, risking the development of resistance, thus the need for new drugs with different mechanisms of action. This review examines the use of machine learning (ML) in this era of big data to aid in the prediction of novel antischistosomal molecules. It first discusses the challenges of drug discovery in schistosomiasis. Explanations are then offered for big data, its characteristics and then, some open databases where large biochemical data on schistosomiasis can be obtained for ML model development are examined. The concepts of artificial intelligence, ML, and deep learning and their drug applications are explored in schistosomiasis. The use of binary classification in predicting antischistosomal compounds and some algorithms that have been applied including random forest and naive Bayesian are discussed. For this review, some deep learning algorithms (deep neural networks) are proposed as novel algorithms for predicting antischistosomal molecules via binary classification. Databases specifically designed for housing bioactivity data on antischistosomal molecules enriched with functional genomic datasets and ontologies are thus urgently needed for developing predictive ML models. This shows the application of machine learning techniques for the discovery of novel antischistosomal small molecules via binary classification in the era of big data.

Discovering, Defining, and Summarizing Persistent Hotspots in SCORE Studies

The Schistosomiasis Consortium for Operational Research and Evaluation (SCORE) conducted large field studies on schistosomiasis control and elimination in Africa. All of these studies, carried out in low-, moderate-, and high-prevalence areas, resulted in a reduction in prevalence and intensity of Schistosoma infection after repeated mass drug administration (MDA). However, in all studies, there were locations that experienced minimal or no decline or even increased in prevalence and/or intensity. These areas are termed persistent hotspots (PHS). In SCORE studies in medium- to high-prevalence areas, at least 30% of study villages were PHS. There was no consistent relationship between PHS and the type or frequency of intervention, adequacy of reported MDA coverage, and prevalence or intensity of infection at baseline. In a series of small studies, factors that differed between PHS and villages that responded to repeated MDA as expected included sources of water for personal use, sanitation, and hygiene. SCORE studies comparing PHS with villages that responded to MDA suggest the potential for PHS to be identified after a few years of MDA. However, additional studies in different social-ecological settings are needed to develop generalizable approaches that program managers can use to identify and address PHS. This is essential if goals for schistosomiasis control and elimination are to be achieved.

Challenges in Protocol Development and Interpretation of the Schistosomiasis Consortium for Operational Research and Evaluation Intervention Studies

In 2010, the Schistosomiasis Consortium for Operational Research and Evaluation (SCORE) began the design of randomized controlled trials to compare different strategies for praziquantel mass drug administration, whether for gaining or sustaining control of schistosomiasis or for approaching local elimination of Schistosoma transmission. The goal of this operational research was to expand the evidence base for policy-making for regional and national control of schistosomiasis in sub-Saharan Africa. Over the 10-year period of its research programs, as SCORE operational research projects were implemented, their scope and scale posed important challenges in terms of research performance and the final interpretation of their results. The SCORE projects yielded valuable data on program-level effectiveness and strengths and weaknesses in performance, but in most of the trials, a greater-than-expected variation in community-level responses to assigned schedules of mass drug administration meant that identification of a dominant control strategy was not possible. This article critically reviews the impact of SCORE’s cluster randomized study design on performance and interpretation of large-scale operational research such as ours.