Leveraging mathematical models of disease dynamics and machine learning to improve development of novel malaria interventions.

BACKGROUND: Substantial research is underway to develop next-generation interventions that address current malaria control challenges. As there is limited testing in their early development, it is difficult to predefine intervention properties such as efficacy that achieve target health goals, and therefore challenging to prioritize selection of novel candidate interventions. Here, we present a quantitative approach to guide intervention development using mathematical models of malaria dynamics coupled with machine learning. Our analysis identifies requirements of efficacy, coverage, and duration of effect for five novel malaria interventions to achieve targeted reductions in malaria prevalence. METHODS: A mathematical model of malaria transmission dynamics is used to simulate deployment and predict potential impact of new malaria interventions by considering operational, health-system, population, and disease characteristics. Our method relies on consultation with product development stakeholders to define the putative space of novel intervention specifications. We couple the disease model with machine learning to search this multi-dimensional space and efficiently identify optimal intervention properties that achieve specified health goals. RESULTS: We apply our approach to five malaria interventions under development. Aiming for malaria prevalence reduction, we identify and quantify key determinants of intervention impact along with their minimal properties required to achieve the desired health goals. While coverage is generally identified as the largest driver of impact, higher efficacy, longer protection duration or multiple deployments per year are needed to increase prevalence reduction. We show that interventions on multiple parasite or vector targets, as well as combinations the new interventions with drug treatment, lead to significant burden reductions and lower efficacy or duration requirements. CONCLUSIONS: Our approach uses disease dynamic models and machine learning to support decision-making and resource investment, facilitating development of new malaria interventions. By evaluating the intervention capabilities in relation to the targeted health goal, our analysis allows prioritization of interventions and of their specifications from an early stage in development, and subsequent investments to be channeled cost-effectively towards impact maximization. This study highlights the role of mathematical models to support intervention development. Although we focus on five malaria interventions, the analysis is generalizable to other new malaria interventions.

Use cases for genetic epidemiology in malaria elimination.

BACKGROUND: While traditional epidemiological approaches have supported significant reductions in malaria incidence across many countries, higher resolution information about local and regional malaria epidemiology will be needed to efficiently target interventions for elimination. The application of genetic epidemiological methods for the analysis of parasite genetics has, thus far, primarily been confined to research settings. To illustrate how these technical methods can be used to advance programmatic and operational needs of National Malaria Control Programmes (NMCPs), and accelerate global progress to eradication, this manuscript presents seven use cases for which genetic epidemiology approaches to parasite genetic data are informative to the decision-making of NMCPs. METHODS: The use cases were developed through a highly iterative process that included an extensive review of the literature and global guidance documents, including the 2017 World Health Organization's Framework for Malaria Elimination, and collection of stakeholder input. Semi-structured interviews were conducted with programmatic and technical experts about the needs and opportunities for genetic epidemiology methods in malaria elimination. RESULTS: Seven use cases were developed: Detect resistance, Assess drug resistance gene flow, Assess transmission intensity, Identify foci, Determine connectivity of parasite populations, Identify imported cases, and Characterize local transmission chains. The method currently used to provide the information sought, population unit for implementation, the pre-conditions for using these approaches, and post-conditions intended as a product of the use case were identified for each use case. DISCUSSION: This framework of use cases will prioritize research and development of genetic epidemiology methods that best achieve the goals of NMCPs, and ultimately, inform the establishment of normative policy guidance for their uses. With significant engagement of stakeholders from malaria endemic countries and collaboration with local programme experts to ensure strategic implementation, genetic epidemiological approaches have tremendous potential to accelerate global malaria elimination efforts.

Role of mass drug administration in elimination of Plasmodium falciparum malaria: a consensus modelling study.

BACKGROUND: Mass drug administration for elimination of Plasmodium falciparum malaria is recommended by WHO in some settings. We used consensus modelling to understand how to optimise the effects of mass drug administration in areas with low malaria transmission. METHODS: We collaborated with researchers doing field trials to establish a standard intervention scenario and standard transmission setting, and we input these parameters into four previously published models. We then varied the number of rounds of mass drug administration, coverage, duration, timing, importation of infection, and pre-administration transmission levels. The outcome of interest was the percentage reduction in annual mean prevalence of P falciparum parasite rate as measured by PCR in the third year after the final round of mass drug administration. FINDINGS: The models predicted differing magnitude of the effects of mass drug administration, but consensus answers were reached for several factors. Mass drug administration was predicted to reduce transmission over a longer timescale than accounted for by the prophylactic effect alone. Percentage reduction in transmission was predicted to be higher and last longer at lower baseline transmission levels. Reduction in transmission resulting from mass drug administration was predicted to be temporary, and in the absence of scale-up of other interventions, such as vector control, transmission would return to pre-administration levels. The proportion of the population treated in a year was a key determinant of simulated effectiveness, irrespective of whether people are treated through high coverage in a single round or new individuals are reached by implementation of several rounds. Mass drug administration was predicted to be more effective if continued over 2 years rather than 1 year, and if done at the time of year when transmission is lowest. INTERPRETATION: Mass drug administration has the potential to reduce transmission for a limited time, but is not an effective replacement for existing vector control. Unless elimination is achieved, mass drug administration has to be repeated regularly for sustained effect. FUNDING: Bill & Melinda Gates Foundation.

Operational strategies of anti-malarial drug campaigns for malaria elimination in Zambia's southern province: a simulation study.

BACKGROUND: Malaria elimination requires reducing both the potential of mosquitoes to transmit parasites to humans and humans to transmit parasites to mosquitoes. To achieve this goal in Southern province, Zambia a mass test and treat (MTAT) campaign was conducted from 2011-2013 to complement high coverage of long-lasting insecticide-treated nets (LLIN). To identify factors likely to increase campaign effectiveness, a modelling approach was applied to investigate the simulated effect of alternative operational strategies for parasite clearance in southern province. METHODS: OpenMalaria, a discrete-time, individual-based stochastic model of malaria, was parameterized for the study area to simulate anti-malarial drug administration for interruption of transmission. Simulations were run for scenarios with a range of artemisinin-combination therapies, proportion of the population reached by the campaign, targeted age groups, time between campaign rounds, Plasmodium falciparum test protocols, and the addition of drugs aimed at preventing onward transmission. A sensitivity analysis was conducted to assess uncertainty of simulation results. Scenarios were evaluated based on the reduction in all-age parasite prevalence during the peak transmission month one year following the campaign, compared to the currently-implemented strategy of MTAT 19 % population coverage at pilot and 40 % coverage during the first year of implementation in the presence of 56 % LLIN use and 18 % indoor residual spray coverage. RESULTS: Simulation results suggest the most important determinant of success in reducing prevalence is the population coverage achieved in the campaign, which would require more than 1 year of campaign implementation for elimination. The inclusion of single low-dose primaquine, which acts as a gametocytocide, or ivermectin, which acts as an endectocide, to the drug regimen did not further reduce parasite prevalence one year following the campaign compared to the currently-implemented strategy. Simulation results indicate a high proportion of low-density infections were missed by rapid diagnostic tests that would be treated and cleared with mass drug administration (MDA). CONCLUSIONS: The optimal implementation strategy for MTAT or MDA will vary by background level of prevalence, by rate of infections imported to the area, and by ability to operationally achieve high population coverage. Overall success with new parasite clearance strategies depends on continued coverage of vector control interventions to ensure sustained gains in reduction of disease burden.

Modeling the cost effectiveness of malaria control interventions in the highlands of western Kenya.

INTRODUCTION: Tools that allow for in silico optimization of available malaria control strategies can assist the decision-making process for prioritizing interventions. The OpenMalaria stochastic simulation modeling platform can be applied to simulate the impact of interventions singly and in combination as implemented in Rachuonyo South District, western Kenya, to support this goal. METHODS: Combinations of malaria interventions were simulated using a previously-published, validated model of malaria epidemiology and control in the study area. An economic model of the costs of case management and malaria control interventions in Kenya was applied to simulation results and cost-effectiveness of each intervention combination compared to the corresponding simulated outputs of a scenario without interventions. Uncertainty was evaluated by varying health system and intervention delivery parameters. RESULTS: The intervention strategy with the greatest simulated health impact employed long lasting insecticide treated net (LLIN) use by 80% of the population, 90% of households covered by indoor residual spraying (IRS) with deployment starting in April, and intermittent screen and treat (IST) of school children using Artemether lumefantrine (AL) with 80% coverage twice per term. However, the current malaria control strategy in the study area including LLIN use of 56% and IRS coverage of 70% was the most cost effective at reducing disability-adjusted life years (DALYs) over a five year period. CONCLUSIONS: All the simulated intervention combinations can be considered cost effective in the context of available resources for health in Kenya. Increasing coverage of vector control interventions has a larger simulated impact compared to adding IST to the current implementation strategy, suggesting that transmission in the study area is not at a level to warrant replacing vector control to a school-based screen and treat program. These results have the potential to assist malaria control program managers in the study area in adding new or changing implementation of current interventions.

Seasonally dependent relationships between indicators of malaria transmission and disease provided by mathematical model simulations.

Evaluating the effectiveness of malaria control interventions on the basis of their impact on transmission as well as impact on morbidity and mortality is becoming increasingly important as countries consider pre-elimination and elimination as well as disease control. Data on prevalence and transmission are traditionally obtained through resource-intensive epidemiological and entomological surveys that become difficult as transmission decreases. This work employs mathematical modeling to examine the relationships between malaria indicators allowing more easily measured data, such as routine health systems data on case incidence, to be translated into measures of transmission and other malaria indicators. Simulations of scenarios with different levels of malaria transmission, patterns of seasonality and access to treatment were run with an ensemble of models of malaria epidemiology and within-host dynamics, as part of the OpenMalaria modeling platform. For a given seasonality profile, regression analysis mapped simulation results of malaria indicators, such as annual average entomological inoculation rate, prevalence, incidence of uncomplicated and severe episodes, and mortality, to an expected range of values of any of the other indicators. Results were validated by comparing simulated relationships between indicators with previously published data on these same indicators as observed in malaria endemic areas. These results allow for direct comparisons of malaria transmission intensity estimates made using data collected with different methods on different indicators. They also address key concerns with traditional methods of quantifying transmission in areas of differing transmission intensity and sparse data. Although seasonality of transmission is often ignored in data compilations, the models suggest it can be critically important in determining the relationship between transmission and disease. Application of these models could help public health officials detect changes of disease dynamics in a population and plan and assess the impact of malaria control interventions.

Estimating malaria transmission through mathematical models.

Evaluating the effectiveness of malaria control interventions on the basis of their impact on transmission is increasingly important as countries move from malaria control to pre-elimination programs. Mathematical modeling can examine relationships between malaria indicators, allowing translation of easily measured data into measures of transmission, and addressing key concerns with traditional methods for quantifying transmission. Simulations show these indicators are statistically correlated, allowing direct comparisons of malaria transmission using data collected using different methods across a range of transmission intensities and seasonal patterns. Results from such models can provide public health officials with accurate estimates of transmission, by seasonal pattern, that are necessary for assessing and tailoring malaria control and elimination programs to specific settings.

Simulation of malaria epidemiology and control in the highlands of Western Kenya.

BACKGROUND: Models of Plasmodium falciparum malaria epidemiology that provide realistic quantitative predictions of likely epidemiological outcomes of existing vector control strategies have the potential to assist in planning for the control and elimination of malaria. This work investigates the applicability of mathematical modelling of malaria transmission dynamics in Rachuonyo South, a district with low, unstable transmission in the highlands of western Kenya. METHODS: Individual-based stochastic simulation models of malaria in humans and a deterministic model of malaria in mosquitoes as part of the OpenMalaria platform were parameterized to create a scenario for the study area based on data from ongoing field studies and available literature. The scenario was simulated for a period of two years with a population of 10,000 individuals and validated against malaria survey data from Rachuonyo South. Simulations were repeated with multiple random seeds and an ensemble of 14 model variants to address stochasticity and model uncertainty. A one-dimensional sensitivity analysis was conducted to address parameter uncertainty. RESULTS: The scenario was able to reproduce the seasonal pattern of the entomological inoculation rate (EIR) and patent infections observed in an all-age cohort of individuals sampled monthly for one year. Using an EIR estimated from serology to parameterize the scenario resulted in a closer fit to parasite prevalence than an EIR estimated using entomological methods. The scenario parameterization was most sensitive to changes in the timing and effectiveness of indoor residual spraying (IRS) and the method used to detect P. falciparum in humans. It was less sensitive than expected to changes in vector biting behaviour and climatic patterns. CONCLUSIONS: The OpenMalaria model of P. falciparum transmission can be used to simulate the impact of different combinations of current and potential control interventions to help plan malaria control in this low transmission setting. In this setting and for these scenarios, results were highly sensitive to transmission, vector exophagy, exophily and susceptibility to IRS, and the detection method used for surveillance. The level of accuracy of the results will thus depend upon the precision of estimates for each. New methods for analysing and evaluating uncertainty in simulation results will enhance the usefulness of simulations for malaria control decision-making. Improved measurement tools and increased primary data collection will enhance model parameterization and epidemiological monitoring. Further research is needed on the relationship between malaria indices to identify the best way to quantify transmission in low transmission settings. Measuring EIR through mosquito collection may not be the optimal way to estimate transmission intensity in areas with low, unstable transmission.