2024 Annual Meeting

Opening Remarks

Emilie Pothin, Co-founder of AMMnet and Group Leader of Analytics and Intervention Modelling at Swiss TPH, kicked off the 2024 AMMnet Meeting by extending a warm welcome to all attendees.

She expressed special gratitude towards Shannon, Gianpaolo, and other members in charge of organizing this two- day  event,  as well as the presenters, panelists and trainers and other contributors for their involvement in the meeting. Lastly, on behalf of AMMnet, Emilie extended a special thanks to Bill & Melinda Gates Foundation for supporting the 2024 AMMnet Meeting, as well as all events organized by AMMnet. Proceeding, Emile highlighted that the purpose of AMMnet is to connect people, knowledge-sharing and networking for advancing malaria research.

Jaline Gerardin, Executive Director and Board Chair of AMMnet, followed Emilie and  began by congratulating AMMnet for its second anniversary. She emphasised the remarkable growth of AMMnet, which has expanded to include over a thousand members since its inception. She then proceeded to introduce the AMMnet’s Advisory Board, along with the 4 committees and subgroups.

Regarding the AMMnet funding opportunities, Jaline said that AMMnet has supported 18 AMMnet Local events so far, with 32 still remaining. She also announced a new funding opportunity of about $40,000, for  training projects or for in-person events in a malaria endemic country. She also recalled the approval of the new policy of local chapters in 2024.

‍

SESSION 1: Overview of Type of Models

Moderated by Hannah Slater
In this section Punan Amratia, Caitllin Bever and Shazia Ruybal-Pesantez gave us a flavour about different types of model.

Dynamic Modelling (Caitlin Bever)

Caitlin began her presentation by asking some questions about modelling including: what is a model? Why do we use models? What is a dynamic model? And what is a mechanistic model?

According to Caitlin, models allow us to test ideas that would be challenging to explore in the real world, due to complexity, ethics, impracticalities of time and measurement that are hard to make. In her presentation, Caitlin recalled an example application of dynamic modelling to make real decisions, using the example of modelling as one component of subnational tailoring in Nigeria. She ended by saying that dynamic models can be powerful tools for exploring diverse questions about how to drive the fight against malaria. However, no model that can truly reflect the future.

Geospatial Modelling (Punam Amratia)

The presenter commenced  by quoting Elliot&waternburg 2004 who defined spatial epidemiology as the description of geographical variation in disease with respect to demographic, environmental, behavioural, socioeconomic, genetic and infectious risk factors. This modelling approach essentially links data with specific geographic locations.  Responding to the question why are geospatial outputs useful, Punan emphasised that geospatial outputs provide an easy-to-understand representation, capturing the pattern with space and  trends overtime. Such outputs are used to support evidence -based decision-making.

‍

• Genomic modelling (Shazia Ruybal-Pesantez)

The presenter started her presentation by highlighting that generating genomic data is now cheaper and more accessible due to high-throughput sequencing. In addition, genetic data complements malaria surveillance by providing insights into epidemiology and transmission patterns.

The presenter also provided programmatic applications of genetic epidemiology, such as detecting drug resistance, assess transmission intensity by estimating R0, identify imported cases, characterize local transmission chains and validate genetic metrics for malaria control strategies.  

Group Discussion
Questions addressed:

• What types of questions can be addressed with this type of modelling?
• What are the challenges (data or analytical/technical) in developing this type of model?
• What are the strengths of this type of model?  Provide a real world example of this model being applied to decision making.
• What are the challenges in communicating the results of this type of model?

Dynamical Model
Types of questions that can be addressed using modelling:
Assessing the potential impact of various interventions;

• Predicting the future trends of malaria burden;
• Drug Development: Genomic information can significantly contribute to the development of new drugs.
• Determine the most effective strategies or combinations of strategies for malaria elimination control and elimination, including resource allocation;
• Evaluate strategies to mitigate the impact of drug and insecticide resistance.

Challenges (data/technical) in developing this type of model:

•  Data accessibility;
•  High quality data;

 Strengths of this type of models:

• Prediction of future trends;
• Provides a platform for assessing the potential impact of different interventions strategies

 Real word examples of this model being applied to decision making:

• To support countries’ Malaria Strategic Plan (Nigeria, Tanzania and Mozambique);
• Shaping long-term strategies such as GTS 2016-2030;

 Challenges in communicating the results of this type of model:

• Use the right language and visualizations to effectively communicate the results to different audiences;
• Presenting the results in a format that is relevant actionable for different decision-makers;

Report on Breakout Session (Genomic, Led by Monica Golumbeanu and Shazia Ruybal-Pesantez)

Discussion Points:

Expected Outcomes from Genomic Research:

• Understanding Disease Origins: Participants emphasized the importance of genomic research in identifying the origins of diseases, especially in countries with porous borders.
• Understanding the genes of dormant vectors and parasites can aid in surveillance efforts despite the challenges.
• Drug Development: Genomic information can significantly contribute to the development of new drugs.
• Pharmacogenetics: This field is crucial for understanding how different age groups respond to drugs, which is essential for personalized medicine.
• Genomic Bioinformatics: This area can provide insights into the genetic basis of disease and help in the development of targeted therapies.
• Insecticide and Drug Resistance: Understanding the genetic basis of resistance in vectors and drugs is crucial for the development of effective control strategies.

Sources of Genomic Data:

• PCR and QPCR: These techniques are commonly used for genetic analysis.
• DNA and RNA Data: Both types of data are valuable for genomic research.
• Molecular Sequencing: This method is essential for obtaining detailed genetic information.
• Other Comments: Participants discussed the importance of data availability and access, the relevance of collected data over time, and the necessity of longitudinal studies to understand patterns.

Data Collection and Relevance:

• Data Collection Duration: Participants agreed that data should be collected over a sufficient duration to ensure its relevance for research.
• Simultaneous Collection of Genomic and Epidemiological Data: There was a consensus on the importance of collecting both types of data simultaneously to provide a comprehensive understanding of disease patterns.

Potential Questions for Modelers:

• Compartmental Models: These models can help understand the genomic information.
• Data Collection Locations: Using models to determine where specific genomic data can be collected is a potential area of research.
• Geospatial Outputs: These can provide insights into where certain information is relevant, aiding in targeted research efforts.

Conclusion:The breakout session on genomic research highlighted the potential of genomic data in understanding disease origins, drug development, and resistance mechanisms. Participants emphasised the importance of data availability, the relevance of longitudinal studies, and the need for simultaneous collection of genomic and epidemiological data. The session also identified potential research questions for modelers, focusing on the use of models to guide data collection and analysis.

‍

Scientific Session 1

In a session moderated by Benedicta Mensah, the scientific section featured presentations from Charlène Mfangnia, Lydia Braunak-Mayer, Olasapu Isaac, Kaba Lancei, Eric Ali, and Hillary Topazin, who shared their research findings.

Charlène Naomie Tedto Mfangnia
Country: Kenya
Institution: International Centre of Insect Physiology and Ecology (Icipe)
Topic: A systems dynamics tool to assess the effectiveness of releasing MB-infected mosquitoes in the fight against malaria

The speaker briefly explained the transmission mechanism of Plasmodium and the strategy of controlling malaria by modifying the vector population. The method they used was system thinking and dynamic systems, through data calibration and performed analysis. They observed a decrease in incidence as well as the number of deaths. He concluded by indicating that there is a need to indicate an optimal strategy with a cost-benefit ratio. The limitations of the study were the availability of a large number of data.

Olasupu Isaac
Country: Nigeria
Institution: Federal University, Oye-Ekiti
Topic: Modelling the effect of mosquito heterogeneous biting pattern on vector control intervention for Plasmodium falciparum Malaria in Nigeria

The study aimed to evaluate the impact of mosquito biting patterns on the effectiveness of LLINs in Nigeria. The research findings indicate that while scaling up LLINs coverage could reduce malaria incidence, combining this with efforts to reduce outdoor biting rates of mosquitoes could be more effective, especially in areas with significant outdoor biting mosquitoes, such as environmental management, to enhance the success of LLINs in malaria control efforts.  

Lanceï Kaba
Country: Mali
Institution: Institut Supérieur des sciences et de Médecine Vétérinaire de Dalaba
Topic: Modelling the impact of expansion of seasonal malaria chemoprevention in Guinea: demographic and geographic expansions

After the presentation of the context of malaria in Guinea, he sought to answer the question asked by the  NMCP regarding the choice between the demographic extension and the geographical extension of the SMC in Guinea. Using the EMOD model for data calibration and simulation, the results showed that geographic extension would be better. However, the presenter mentioned that he would complete the study with a cost-effectiveness evaluation for a final decision, which will be the next stage of his study.

Eric Ibrahim
Country: Kenya
Institution: University of Kwazulu Natal and ICIPE
Topic: Spatial-temporal dynamics of malaria vector niche overlaps in Africa

Using cellular automata models the study aim is on examining the expanding niche overlaps between primary and secondary malaria vectors in Africa. The study revealed that co-existence of these vectors significantly impacts malaria transmission, emphasising the importance of indoor control strategies and bionomics. The presenter has highlighted that expansion of niche overlaps places numerous areas at risk of sustained and prolonged malaria transmission and raise concerns about the potential spread of malaria if reintroduced into currently malaria-free zones

Hillary Topazin
Country: United Kingdom
Institution: Imperial College London
Topic: The use of surveillance test-and-treat posts to reduce Malaria in border regions in sub-Saharan Africa

Used an individual based, mathematical metapopulation model of P. falciparum to estimate the effectiveness of border posts on total cases in malaria endemic sub-Saharan. The study focused on malaria elimination strategies at border posts, showing that interventions at these posts averted cases and reduced the PfPr2-10, especially in low-high transmission settings. The presenter emphasised border posts will not allow a country to reach elimination in isolation and the need for coordinated effort, as only a small proportion of individuals are mixing at each time unit.

Lydia Braunak-Mayer
Country: Switzerland
Institution: Swiss TPH
Topic: Defining target product profile criteria for next-generation malaria vaccines to accelerate elimination efforts: a modeling study

Using a modeling framework of malaria epidemiology and control (OpenMalaria), the study focuses on accelerating the development of next-generation malaria vaccines by exploring the required product properties for different deployment scenarios with the ultimate goal of malaria elimination. The presenter highlighted the importance of high intervention coverage and stringent vaccine properties in settings with PfPr2-10>10% to achieve substantial impact and accelerate elimination. The presenter emphasized the need for multi-target approaches and the benefits of deploying specific vaccines based on initial prevalence level. The presenter also underscores the significance of immunogenic and protective vaccines for all age groups, especially in settings with low initial prevalence, to interrupt transmission effectively.

Day 2

On day 2  of the AMMnet Kigali meeting Benedicta Mensah, gave us a brief recap on what transpired on day 1 ,she was then followed by Isambi Mbalawata  the MC of the day who welcomed participants.
Isambi’s welcome address was followed by the launching of the AMMnet Mentoring Program, by Luc Djogbenou and Vusi Magagula from the AMMnet Career Development Committee.

The AMMnet Career Development Committee’s presentation was followed by the training session ,
Participants later broke in groups after the Panel of discussion where Guinea, Ethiopia and Benin gave an overview of the work each country did as far as Malaria eradiction is concerned.
Model Calibration and Validation Using Bayesian Statistical Inference Application to Malaria Models (Led by Rabiu Musa)

The session began with an introduction to model calibration, validation, and illustration. Rabiu Musa provided a comprehensive overview of parameter estimation, highlighting two primary approaches:

• Least Squares Methods: This approach lacks provable statistical properties, making it a less reliable method for parameter estimation.
• Maximum Likelihood Estimation: Unlike least squares methods, maximum likelihood estimation has provable statistical properties, making it a more reliable method for parameter estimation.

Rabiu further delved into two types of maximum likelihood estimation:

• Bootstrapping: A resampling technique that involves generating a large number of resamples of the original data set and using these resamples to estimate the distribution of the estimator.
• Bayesian Inference: A method that combines prior knowledge with observed data to update the probability of hypotheses. It is noted for being computationally intensive.

Following the theoretical introduction, participants gained practical experience with Bayesian estimation. Rabiu led a hands-on session using the “STAN” package in R language to estimate the influenza outbreak in Britain in 1978. This practical application demonstrated the application of Bayesian statistical inference in real-world scenarios, providing participants with valuable insights into the process of model calibration and validation.

‍

Session 2:

The day 2 scientific session  was based on Strengthening Malaria Modeling Capacity in Sub-Saharan Africa with presentations from Trainees form the BMGF Grands Awardees Consortia

Spatiotemporal patterns of malaria transmission among children aged 6-59months in Ghana; application of Bayesian Hierachical models

Wisdom Kwami Takramah,WAMCAD

In his presentation, Wisdom first described the problem of malaria in children. About a quarter (25%) of the 20,000 annual deaths of malaria in Ghana is from children under the age of 5. For the research, he extracted malaria data from the Ghana demographic and Health Survey as well as from the Ghana Malaria Indicator Survey. He used climatic variables, population variables as well as ITN coverage to determine the number of children under 5 years of age with positive rapid diagnostic test for malaria. Wisdom then estimated a geospatial map of the clustering of relative risk of malaria among the children. The clusters were determined with high-high, low-low, high-low and low-high values of relative risk of malaria. He then developed a Bayesian hierarchical spatio-temporal models for predicting malaria risk. This enabled him to identify the high and low smooth relative risk of malaria among children under 5 years in the different regions of Ghana. From the hotspots, public interventions can be deployed, and from the coldspots, lessons can be learnt.

Modelling the prevalence and dynamics of Malaria burden in Bauchi State, Nigeria: using SIRh_SIm – Compartmental Model

Dr Amusaidu  Saleh,Nigeria MMF

Dr Saleh, in his presentation started with a map showing the risk distribution of Malaria burden across the states and zones of Nigeria. He then talked about the Bauchi state which has a high malaria prevalence. In his research he used an Susceptible-Infectious-Recovered Susceptible-infectious compartmental model to help predict prevalence and understand malaria transmission dynamics in Bauchi state for better decision making. He used 2 years of malaria incidence data from the district health information system-2 to build the model to simulate transmission in Bauchi state. He reported that the model estimated Bauchi state malaria prevalence at 3.19%, which suggests that existing interventions are effective.  Other findings were that there is a high transmission rate from susceptible to infected humans; low rate of recovery from malaria; moderate transmission from infected human to susceptible mosquito and high transmission rate from infectious mosquitoes to susceptible humans.

Geospatial Modelling for improved Anopheles Gambiae s.i field surveillance through adatative spatial sampling design

Gabriel Michel  Monteiro, MaModAfrica

Gabriel, in his presentation, started by pointing out the importance of field surveillance of malaria vectors. He said that it monitors malaria vector populations and dynamics, biting rates, insecticide resistance and resting behaviours. These help in targeting interventions as well as monitoring strategies for malaria control. He then defined the adaptive sampling design and explained that it is preferred because it is guided by a predictive target, and provided evidence of adaptive design for disease control or surveillance in the form of publications. He then talked about the study objective of evaluating the efficacy of spatial adaptive sampling designs for vector surveillance in terms of the reduction of model uncertainty and the increase in the accuracy of describing mosquito spatial patterns. Entomological collection was done in 2018 (phase 1) and 2021 (phase 2) in Benin. Gabriel then detailed the steps in his adaptive design before describing the catches of malaria vectors by sampling phase (2018 and ). He presented that the overall framework allowed for more catches of malaria vectors, model performance improved but the estimation error did not improve due to adaptive target. He found that there was an increased risk of Anopheles Gambiae in most regions. In conclusion, he talked about the advantages of using adaptive spatial design.

Investigating a coordinated regional approach to Malaria elimination using Mathematical Modelling

Hilja Eelu, MMALA

In her presentation, Hilja started off by showing a map of malaria transmission intensity in southern Africa. She then talked about multi-country malaria initiatives in the area. She then described and compared populations at risk, cases, deaths, ITN and IRS coverage, and malaria funding in the status in Angola, Namibia and Zambia. She then went on to describe the cross coupled metapopulation model for the three patches of Angola, Namibia and Zambia and illustrated the model for each patch. This model will be used to determine the influence of infectiousness of neighbouring patches. She then described the training she received from Modelling and simulation Hub Africa (MASHA), and organizations like Elimination 8 (E8) and the Namibia National vector-borne Disease control center  that were instrumental in her study.

Spread of insecticide resistance into a Malaria vector population: a Mathematical Modelling approach

Sylvere Kezeta, ACoMVec

Sylvere started by describing the burden of malaria in the world.  He then talked about the vector control measures of ITNs and IRS, and explained that the scale-up of ITNs contributed to the reduction in burden of malaria in 2005-2015. He further said that the slowdown in the reduction of malaria was due to insecticide reduction. His main objective was to use mathematical modelling to understand how insecticide resistance emerges. His specific objectives  are to build and analyze a dynamical system, identify the main factors and then propose strategies. He then explained an illustration of a flow diagram of his model and the assumptions. Sylvere believes that his findings showed that the model is well posed by proving the existence of a unique, positive and boundedness solution. The next step in the study is to do mathematical analysis of the  model , simulate and calibrate the model and then generate different scenarios.

‍

AMMnet Hackathon

Justin Miller, Christian Selinger

The duo started their presentation by explaining the problem of attending conferences focused on improving coding skills but fail to focus on maintaining and growing skills once we return home. The aim of the hackathon is to provide support for learning to code, and how to improve coding skills remotely specifically in data handling, presentation and analysis; experimental design, quantitative understanding, coding and problem solving, and transmission modelling. The idea is for skilled individuals to sign up and provide support and participants to submit coding related problems for consideration. Two or three problems are selected and discussed by a small team of experts that also prepares solutions. The solution is then presented in a 2 hour session with codes generated and stored. They ended their presentation by inviting people to sign up as experts or submit problems.

‍

AMMnet as part of BMGF Strategy

Jennifer Gardy
In her remarks, she said that in the past modelers were not representative of the communities that were being helped, but things are changing. She congratulated the growth of AMMnet and it including friends of modelers in their fold. She said that modelling fits with global health strategy at the BMGF. The strategy guides pandemic preparedness, malaria and TB/HIV and looks into aims and objectives over a period of a few years, and the ways to achieve these. The aim of this strategy is to eliminate malaria in our lifetime but there are some challenges. The first is that between 2005 and 2015 malaria incidence dropped but has since levelled off. The second is that donor funding has declined and the world needs three times more funding than we currently have. Lastly, the world needs to stay ahead of the evolution of the vector to prevent resistance. This can be achieved with new drugs and vaccinations. She further said that modelling can be useful in defining the public health strategy by helping in subnational tailoring of malaria interventions since blanket coverage is not helping. She concluded by saying that modelling is a key tool in the work that is going on at the Gates foundation.

Closing Remarks
The closing remarks were given by Jaline Gerardin closed the meeting by thanking the  whole AMMnet Management Team and members, but also the funders, such as Bill and Melinda Gates Foundation, for their financial support that enabled the organization of the event  and looked forward to the next AMMnet meeting  in 2025.