Meeting notes prepared by Zahraa Mohamed and William Kuipou

At the 2023 AMMnet meeting, members shared ongoing work, celebrated the work of the past year, developed their network, and discussed future activities of AMMnet in the coming year.

Opening remarks

The meeting opened with Jennifer Gardy from the Bill & Melinda Gates Foundation discussing the importance of modelling in the malaria ecosystem. Dr Gardy spoke about the importance in understanding risk and the deployment of old and new interventions. She is proud of this community, which is unique in its inclusivity of placing African leadership at the center of the fight for malaria eradication.

Jennifer Gardy was followed by Jaline Gerardin who spoke about the history of AMMnet. In its 18 months, AMMnet has grown to over 850 members, 8 regional AMMnets, 68 countries, 20 webinars, 16 newsletters, 4 travel awards, 11 board members, 2 language-based groups and has funded 13 local events. The AMMnet story will be highlighted at the Seattle year-end review of the Bill & Melinda Gates Foundation.

This year featured 10 presentations which were both in person and virtual; a poster session; a panel discussion; and small group discussions.

Session 1: Country contexts

Moderated by Akindeh Nji, University of Yaoundé, Cameroon

Jeanne Lemant

Country: Switzerland

Institution: Swiss Tropical and Public Health Institute

Topic : Modeling to support decisions about the geographic and demographic extensions of seasonal malaria chemoprevention in Benin.

Abstract
Seasonal malaria chemoprevention (SMC) has been implemented in Benin since 2019 and targeted more than 400 000 children under 5 in the northern departments of Alibori and Atacora in 2021. The Benin National Malaria Control Program (NMCP) recently considered an extension of SMC, either demographically – children aged 5 to 10 in the same departments would also receive SMC, or geographically to children under 5 in new departments eligible according to WHO criteria. As neither extension had been tested before, the NMCP turned to modeling to estimate their impact. The model OpenMalaria was calibrated to represent the history of malaria interventions and transmission risk in Benin, as well as the age structure of the population. The future interventions which were already planned (mass net distribution campaigns, SMC in children under 5 in Alibori and Atacora, pilot projects of intermittent preventive treatment in infants) were simulated, together with the two extensions of SMC. The model predicted that the demographic extension of SMC could avert on average 4.6 severe cases per 1000 targeted children between 2024 and 2026, while the geographic extension could on average avert between 13 and 14.3 severe cases per 1000 children under 5, depending on the department. To be less cost-effective than the demographic extension, the geographic extension should thus be three times more expensive, when costs from the 2021 campaign indicate it would cost only 40% more. Numbers of severe cases averted per targeted child were similar between operational zones of departments considered for the geographic extension, probably due to similar transmission risks. These findings led to recommend targeting highly populated zones in priority, as SMC in the three most populated zones could avert as many severe cases as in the six other zones. Modeling allowed not only to choose the geographic over the demographic extension, but also to quantify their comparative impact. Modeling can be used to answer questions from decision-makers when they are closely associated to the process, from the refinement of the modeling question to the choice of epidemiological indicators.

Notes:

During her presentation, Jeanne provided a background of SMC in Benin. In 2019, Benin started SMC in two northern zones for 3-59-month olds every month between July and October. After the WHO changed the recommendation to under-10-year-olds, the question arose as to whether to demographically extend the programme to include under-10-year-olds in the 2 zones or to geographically extend SMC to the rest of the country but maintain the 5-year-old cut-off. 

Jeanne’s group used the OpenMalaria model, which includes an agent-based model for humans and a model for mosquitoes with an interaction between the two. Local data was used to tailor the model to the Benin context and two interventions, bednets and SMC, were included. The model predicted at least 4X more cases averted and 3.4 times more cases averted per 1000 USD for the geographic extension compared to the demographic extension. This captures the community impact as it includes cases in all ages.

Limitations included constant SMC coverage at 80%, constant costs and efficacy was based on clinical trials, which may be overestimated (but was the same for both strategies) and constant seasonality. 

During questions, Jeanne spoke about collaborating with the NMCP from whom the initial question came. The group produced a dashboard for the NMCP so they could choose indicators relevant to them.

Moussa Kane

Country: Senegal

Institution: WAMCAD (West Africa Mathematical Modeling Capacity Development) & MARCAD (The Malaria Research Capacity Development Consortium)

Topic : Mathematical modeling of malaria transmission dynamics for low transmission zones: application to Keur Soce, Senegal

Abstract
Senegal is experiencing a considerable drop in the number of malaria cases. The zone of very low transmission is in the north of the country, extending towards the west and center. However, the number of cases can suddenly increase at any given time, and sometimes potentially uncontrollably. This paper investigates how to avoid the resurgence of the malaria epidemic in pre-elimination zones such as Keur Socé by developing a mathematical model of malaria transmission, built by considering a SIRS-SI model where healthy humans are infected with the parasite when bitten by infectious mosquitoes while healthy mosquitoes themselves become infected when bitten by infectious humans, and furthermore the immunity of recovered humans is temporary. We estimated the reproduction number R_0 and discussed the stability of disease-free and endemic equilibria. Numerical simulations were carried out using scilab software to confirm our analytical results. We drew conclusions by analyzing each parameter of the R_0 expression. But first, centered moving averages are used to smooth the chronic malaria morbidity data from Keur Socé, in order to understand the past, analyze and explain the observed values and predict the near future.

Notes:

Moussa outlined his descriptive analysis of incidence including seasonality and explained using behavioural models to make forecasts. A compartmental model with a human and mosquito component was used. The formulation and mathematical analysis of the compartmental model was explained including using Jacobian matrices for the disease-free and endemic equilibria. 

Samuel Oppong

Country: Ghana

Institution: National Malaria Elimination Programme

Topic : Stratification of Malaria Burden and Subnational Tailoring of Interventions for Ghana – comparison of different adjustment methods for malaria incidence.

Abstract
As part of the pillars of HBHI – use of strategic information for decision making, stratification of malaria burden at subnational level is imperative to inform appropriate interventions to be deployed. In 2022, the name of the Ghana National Malaria Program was changed from Control to an Elimination Programme. This called for a new strategic plan to be developed to reflect the scope and strategies to be implemented under the elimination agenda. This abstract outlines the processes and methods used in adjusting crude incidence from health facilities reported in the routine health information system. Crude incidence estimates are adjusted for testing rates (level 1), reporting rates (level 2) and health seeking behaviour (level 3) to ensure representativeness of malaria incidence in the country. A comparative analysis of three (3) different adjustment methods for level 3 incidence was done to determine the one that gave a more conservative estimates for incidence rate. The three methods were the WHO method for adjusting for health seeking behaviour, the MAP adjustment estimates and a country specific adaptation of the WHO adjustment method for health seeking behaviour. Whiles the WHO adjustment method for level 3 overestimated the mean level 2 estimates by 3-folds, MAP estimates showed very little variation between level 2 and 3. The country adopted method for adjusting for level 3 increased the mean level 2 incidence by 2-fold but was considered conservative considering other developments that has impacted the malaria surveillance landscape in the country. The third level adjusted incidence at district level was combined with prevalence estimates from Malaria Atlas Project (MAP) and All-cause mortality estimates from the Institute for Health Metrics and Evaluation to categorize the districts into very low, low, moderate, and high epidemiological strata.

Notes:

Samuel provided background about the malaria landscape in Ghana with recent reductions in malaria incidence and improvements in testing (38% to 98%). SO explained that shapefiles on country-level data have different regions and districts compared to before. He spoke about adjusting for health-seeking behaviour leading to a 3 times increased crude incidence, which might be related to public sector health-seeking. 

Questions included future plans to compare incidence in under-5s compared to over-5s, which will be part of Samuel’s PhD. Another question was about fever due to West Nile virus but Samuel said that it not much of a factor in this work. 

Moureen Ochieng

Country: Uganda

Institution: Infectious Diseases Epidemiological Modelling Unit at Makerere University (IDEMU-Mak)

Topic : Modelling influenza A. 

Notes:

Moureen presented a compartmental SEIR model for influenza A using a set of differential equations. This included a susceptible, exposed, infected and two recovered compartments (one with immunity and one without immunity). She described each of the rates between the compartments and explained the model parameters. 

Session 2: Understanding risk and intervention effectiveness

Moderated by Misonge Kapnang Ivan, University of Dschang, Cameroon

Toussaint Rouamba

Country: Burkina Faso

Institution: Clinical Research Unit Of Nanoro

Topic : Modelling the effects of weather on malaria temporal trend in small area, Burkina Faso

Abstract

For decades, climate change and its effects on diseases have received increasing attention worldwide. This effect can be heterogeneous, especially in regions where off-season crops are predominant, and it is essential to assess the effect of these changes on the intensity of diseases such as malaria. This study explores short-term fluctuations in malaria cases and meteorological variables in the health district of Nanoro, Burkina Faso.
Rainfall, humidity, temperature, and malaria data for the period 2010-2022 were analyzed. The Mann-Kendall test was used to assess the trend of the time series. Change point analysis was performed to detect significant changes in the series mean and variance for 676 weeks. The cross-correlation function (CCF) was used to assess the statistical relationship between weekly weather variables (principal components) and weekly malaria incidence.
Detailed analysis of the 13-year data indicates that the annual maximum and annual minimum temperatures have not shown an increasing trend, while the malaria incidence has shown an increasing trend. Three periods of malaria incidence were found and uniformly distributed annually. Rainfall and temperature were positively and significantly associated with malaria incidence, with a lag time of 9 and 14 weeks, respectively.
A better understanding of the influence of climate on malaria at the local level would allow the development of sensitization campaigns and the readiness of the health system to respond to climate-induced malaria. This implied that programs such as seasonal malaria chemoprophylaxis could be improved by extending to the intermediate period.

Notes:

Unfortunately Toussaint was not able to present his talk due to technical issues.

Justice Aheto

Country: Ghana

Institution: University of Ghana

Topic : Mapping under-five child malaria risk that accounts for environmental and climatic factors to aid malaria preventive and control efforts in Ghana: Bayesian geospatial and interactive web-based mapping methods.

Abstract
Under-five child malaria is one of the leading causes of morbidity and mortality in sub-Saharan Africa. In Ghana, malaria is responsible for about 20000 deaths in children annually of which 25% are those aged <5 years. To provide opportunities for efficient malaria surveillance and targeted control efforts amidst limited public health resources, high resolution interactive web-based spatial maps that characterized geographical differences in malaria risk are required. Methods: This modelling study utilised data from the 2019 Malaria Indicators Survey. A novel Bayesian geospatial modelling and interactive web-based mapping approaches were utilized to examine predictors and geographical differences in under-five malaria. Interactive web-based visualization maps of the predicted malaria risk were developed. In 2019, 718 (25%) of 2867 under-five children surveyed had malaria. Substantial geographical differences in under-five malaria risk were observed. ITN coverage (log-odds 4.5643, 95% credible interval = 2.4086 – 6.8874), travel time (log-odds 0.0057, 95% credible interval = 0.0017 – 0.0099) and aridity (log-odds = 0.0600, credible interval = 0.0079 – 0.1167) were predictive of under-five malaria in the spatial model. The overall predicted national malaria prevalence was 16.3% (standard error (SE) 8.9%) with a range of 0.7 % to 51.4% in the spatial model with covariates and prevalence of 28.0% (SE 13.9%) with a range of 2.4 to 67.2% in the spatial model without covariates. Residing in parts of Central and Bono East regions was associated with the highest risk of under-five malaria after adjusting for the selected covariates.The high-resolution interactive web-based predictive maps can be used as an effective tool in the identification of communities that require urgent and targeted interventions by program managers and implementers. This is key as part of an overall strategy in reducing the under-five malaria burden and its associated morbidity and mortality in a country with limited public health resources where universal intervention is practically impossible.

Notes:

Justice started by introducing the burden of malaria generally. He used the Ghana Malaria Indicator Survey and focused on the outcome of RDT positivity. He used a binomial model with a number of covariates and a spatial component. He used spatial methods to predict malaria prevalence country-wide and presented a web-based interactive tool for the NMCP.

Alvan Coker

Country: Liberia

Institution: National Public Health Institute of Liberia and the School of Public Public Health, University of Liberia

Topic : Modeling the Impact of Gender-Specific Behavior on Enhanced Lassa Fever Transmission Risk among Females in Grand Bassa County, Liberia

Abstract
In Liberia, Lassa fever outbreaks with high case fatality have been widespread despite implementation of a national plan to curb the disease. Grand Bassa County has experienced sustained transmission since 2021. To demonstrate the potential impact of intervening on behavioral transmission routes, we developed a mathematical model for Lassa Fever focused on District 3A&B, Grand Bassa County. A compartmental model for Lassa fever transmission was parameterized using routine surveillance data and detailed outbreak line lists. The model was sex-stratified to capture socio-behavioral practices hypothesized by national stakeholders to enhance transmission risk among females relative to males. A rodent sub-model was fit to account for differential infection among Mastomys rats inside and outside the house and seasonality in vector abundance. The full model was subsequently fit to data on sex-specific case counts and overall death counts. The impact of behavioral interventions on differential case counts between males and females was quantified. Socio-demographic, clinical, epidemiological, and behavioral data on 83 confirmed Lassa fever cases (63% female) reported in the line lists between 2017 and 2022 were available for District 3A&B. The model projected a cumulative incidence of 50.9 care-seeking female cases (IQR: 35.4-76.4) and 22.7 care-seeking male cases (15.4-33.2) for 2023-2027. Environmental clean-up to reduce rodent recruitment into houses led to the largest overall reductions in case counts, while safer handling practices of hunted rodent meat considerably reduced disparities in case counts. Reduced transmission via consumption of uncovered and contaminated food enhanced disparities in case counts. Diverse local data were used to inform Liberia’s first model of Lassa fever. By reflecting aspects of transmission that regularly arise during discussions among national stakeholders, the model is designed to maximize its utilization, particularly by Liberia’s newly developed Lassa fever technical working group.

Notes:

Unfortunately Alvan was not able to present his talk due to technical issues.

Sequoia Leuba

Country: United Kingdom

Institution: Imperial College London

Topic : Quantifying the impact of malaria in pregnancy on maternal anemia and its associated burden across Africa.

Abstract
Plasmodium infection during pregnancy causes maternal anemia but quantitative estimates of the burden of malaria on maternal anemia are lacking, in part because the impact of untreated infections cannot be ethically measured. To address this gap, we used individual-level data on hemoglobin (Hb) concentrations and malaria PCR status at enrolment into 4 recent trials of alternative approaches to malaria prevention in pregnancy, involving 12,608 women in seven countries (Burkina Faso, The Gambia, Ghana, Kenya, Malawi, Mali, and Tanzania). We developed a Bayesian inferential framework to account for the various exclusion criteria of the trials, using data on gestational age at enrolment as a proxy measure for the length of time an infection was left untreated to capture Hb dynamics up to the end of the second trimester (T2). We estimate that among primigravid women, reductions in Hb associated with malaria infection increase throughout gestation, reaching a reduction of 1.24 [95% Credible Interval (CI) 1.13, 1.36] g/dL at the end of T2. Accounting for concomitant declines in Hb throughout gestation in non-infected women, we estimate that, in primigravidae with ongoing infection, the risk of malaria-associated severe anemia (Hb < 7 g/dL) increases from 2.2% [95% CI 1.1-3.5] to 14.3% [95% CI 10.8-17.9] between the ends of the first trimester (T1) and T2. The impact of malaria upon Hb in multigravid women varied by transmission intensity, with the impact similar to primigravidae in areas of lower prevalence but increasingly diminishing in areas of higher transmission, following well-understood patterns in acquisition of pregnancy-specific malaria immunity. Using modelling, we estimate that among women who have experienced infection in one previous pregnancy, the reduction in Hb concentration associated with ongoing infection at end of T2 is 0.38 [95% CI 0.21, 0.55] g/dL, and no reduction in Hb concentration in any subsequent pregnancies. Using this framework, we will extrapolate the associated burden across Africa using estimates of fertility and malaria endemicity collated by the Malaria Atlas Project and the World Health Organization.

Notes:

Sequoia spoke about the number of pregnancies impacted by malaria (13.3 million) and its consequence of anaemia. She used an existing model of malaria in pregnancy and modelled haemoglobin changing as a function of gestational age. She showed how immunity differs based on the individual’s gravidity so they extended it to see how the number of the pregnancy impacts malaria’s impact on anaemia. Overall, the impact of malaria on anaemia worsens through gestation and the impact is greatest amongst women in their first pregnancy. She showed how this impact differed in high vs. low-transmission settings. In the former, the impact differs depending on gravidity (i.e., as more people are infected with malaria in earlier pregnancies) while, in the latter, malaria has an ongoing impact with increasing number of pregnancies. 

During question-answer session, Sequoia said that they will next incorporate IPTp in the model. She also spoke about many of malaria infections being carried over from before pregnancy and the importance of early antenatal care.

Session 3: Modelling antimalarials

Moderated by Antonio Nkondjio Christophe, OCEAC, Yaoundé, Cameroon

Gina Cuomo-Dannenburg

Country: United Kingdom

Institution: Imperial College London

Topic :  Potential suitability of sulfadoxine-pyrimethamine plus amodiaquine for seasonal malaria chemoprevention in areas of high, pre-existing drug resistance.

Abstract
Seasonal malaria chemoprevention (SMC) targets the burden of malaria in children under five in areas of seasonal malaria transmission. Previously, the WHO recommended SMC only in the Sahel region which has low levels of sulfadoxine-pyrimethamine (SP) drug resistance, one of the drugs used for the intervention. However, in 2022 geographic restrictions were removed from WHO guidelines prompting new countries to consider SMC as a possible intervention. There is a need to understand whether SP-AQ would still be effective in areas with high SP resistance to guide the scale-up of SMC to new geographies. Here, we utilize data from the first randomized controlled trials of SMC outside of the Sahel region to estimate the protection provided by SMC which demonstrated a day 28 protective efficacy of 77%, even in areas with established high-level SP resistance. We use Bayesian inference methods to estimate the probability that SMC using SP-AQ would prevent an infection given time since drug administration, seasonality, baseline transmission and existing frequency of SP-resistance conferring mutations. We are currently using these results within an existing, extensively validated individual-based Plasmodium falciparum transmission model to estimate the potential impact of implementing SMC under a variety of scenarios, including exploring the number of cycles, their timing, and the suitable age range. Initial results suggest that in an area with 64.7% of clinical cases in 4 months and established dhfr-dhps quintuple mutation, SMC using four cycles of SP-AQ could prevent 51.1% (95% CI: 37.0 – 65.2%) of annual clinical Plasmodium falciparum cases in children under five years. Despite the high drug resistance already present in east and southern Africa, we predict that SMC has the potential to be a highly effective malaria intervention and could help avert some of the substantial malaria burden in young children in these geographies. It will be important to consider the impact of scaling up SMC on driving resistance to SP and AQ in areas where SP resistance is already relatively high.

Notes:

Gina explained the role of SMC in preventing malaria in children at highest risk. SP + AQ is the most common regimen for SMC but SP was the main first-line treatment in eastern and southern Africa so there is resistance in those regions. A transmission model, Simulation, was used to examine the impact of resistance on the impact of SMC. The model showed that SP resistance does impact SMC efficacy but is still likely to be highly effective in east and southern Africa.

The first question asked about the differences between percentage reduction using Bayesian frameworks compared to classical survival analysis and Gina explained that the former includes varied transmission intensity and drug intensity as a waning effect. The second question was about the likely programmatic impact as this is often lower than trial impact. Gina explained that the results here are conservative and likely more closely approximate programmatic results than trial results.

Constanze Ciavarella

Country: France

Institution: Institut Pasteur

Topic :  Plotting a path through the P. vivax treatment dilemma a modelling study integrating clinical trial data and transmission dynamics.

Abstract
Upon primary infection, some Plasmodium vivax (Pv) parasites develop into hypnozoites that lie dormant in the liver for weeks to months before reactivating to cause relapses. Treatment of Pv thus calls for radical cure, a type of therapy that clears parasites in both blood flow and the liver. Several randomised control trials (RCTs) have compared drug regimens with the liver stage drugs primaquine (PQ) and tafenoquine (TQ). While it is possible to estimate the individual-level efficacy of these drug regimes, their population-level impact still must be evaluated in cluster randomised trials. Such trials are necessary to measure the non-linear effects on transmission due to lower, heterogeneous Pv circulation and decreased population immunity. We developed an infection risk model and fit it to IMPROV trial data (two PQ regimens and a control arm trialled in multiple locations) assuming relapse and biting rates to be location-specific, but drug efficacy to remain constant across locations. Next, we estimated the population-level impact of introducing radical cure in case management using an existing Pv transmission model covering multiple scenarios that vary by transmission intensity, seasonality, Pv relapse rates and care-seeking behaviour. Combining our efficacy estimates with hazard ratios and adherence data from RTCs, we tested several radical cure regimens varying dosage and duration of administration. The efficacies estimated for the 7- and 14-day IMPROV PQ regimens (7 mg/kg total dose) were of 81% (95% CI: 66%-96%) and 86% (95% CI: 72%-99%), respectively. Introducing radical cure may make elimination feasible where transmission is already low (<2% PCR-prevalence). As transmission intensity increases, the efficacy of radical cure is vastly reduced and differences between regimens even out. To date PQ and TQ have been tested under trial conditions, while real-world implementations introduce many constraints that hamper their population-level impact. Rather than focusing on optimal dose and duration of administration, it might thus be more effective to increase adherence and care-seeking rates, and to widen eligibility criteria.

Notes:

Constanze explained the differences between Plasmodium falciparum and vivax. In the latter, hypnozoites remain dormant in the liver for weeks to months and can reactivate, leading to relapse. Two drugs, primaquine and tafenoquine, target the liver-stage parasites but rolling out this treatment is tricky due to severe side effects in certain people. The combination of blood and liver-stage drugs is called radical cure. 

Constanze developed a model including a susceptible compartment with hypnozoites, who could be reinfected or relapse, an infected compartment with blood-stage infection and a susceptible compartment without hypnozoites that can only become reinfected. She presented the potential effect of radical cure on low, medium, and high-transmission populations using an individual-based model of transmission. 

Poster Session

Posters will be posted soon.

Panel Discussion: Experiences establishing and running regional AMMnet chapters

Moderator: Antonio Nkonjio Christophe

Panelists:

• Misonge Kapnang Ivan, AMMnet Cameroon, epidemiologist at the Centre for Infectious Disease, works on modelling projects for the Bill & Melinda Gates Foundation.
• Getachew Teshome, AMMnet Horn of Africa
• Isaiah Agorinya, AMMnet Ghana, teaches epidemiology and biostatistics at the University in Life Sciences and is currently at Northwestern University for a faculty enrichment programme.
• Abdourahamane Diallo, AMMnet Francophone, studies public health and works at Guinea’s national malaria programme.
• Carmene Sandra Ngadjeu, AMMnet Francophone

Note: questions and responses have been paraphrased.

Question 1: Please outline the activities you have undertaken at your AMMnet chapters.

Isaiah Agorinya: the process to start AMMnet Ghana started last October. IA contacted colleagues, formed a WhatsApp group and included anyone interested. The first virtual meeting was held in December 2022, the second 31 March 2023 and the third last week where a colleague presented their work. IA formed a local slack channel alongside WhatsApp for spreading information. AMMnet secretariat awarded a new grant for a workshop that is planned for the second week of December. 

Misonge Kapnang Ivan: contacted Shannon and was very active in AMMnet webinars. He was chosen to be Cameroon representative. There is an anglophone and a francophone part to AMMnet Cameroon. Two grants were awarded; the first to organise a modelling conference and the second for entomologist training. The next plan is to organise a symposium. 

Getachew Teshome: Horn of Africa AMMnet is at its infant stage. They organised a workshop last year, which was useful for collaborative work (rather than in silos). AMMnet Horn of Africa members suffer from malaria themselves so it is especially important to them. There are many technical gaps they hope to address.

Question 2: What are the challenges in the Francophone AMMnet space? Is language a barrier and how could Anglophone AMMnet help to strengthen the Francophone group?

Abdourahamane Diallo: Francophone AMMnet has also created WhatsApp groups and have reached 7 countries. They try to have monthly meetings and invite French researchers to webinars. Training is tricky as Francophone modellers are not many. The plan is to develop training to strengthen capacity. AMMnet encourages modelling in Africa and this should include Francophone Africa. 

Carmene Sandra Ngadjeu: Language is definitely a barrier and more people would be interested in modelling if they spoke English. AMMnet Francophone would like French-led activities and would like to be included in the wider AMMnet community and to be supported. 

Question 3: What about women’s role in modelling? How can we encourage women? 

Carmene Sandra Ngadjeu: Unfortunately, women are less interested in modelling and see it as for men. Only 1 in 10 modellers are women but women can be modellers and should be supported. 

Abdourahamane Diallo: Scholarships to boost women’s engagement could be useful. He agreed that few women in Africa are interested in modelling. Scholarships for training programmes would be useful. 

Question 4: How can AMMnet help support chapters?

Isaiah Agorinya: AMMnet does support with monthly seminars, secretariat assistance, free access to Zoom. Training is essential especially due to diverse backgrounds in the teams. 

Question 5: How do we ensure sustainability?

Misonge Kapnang Ivan: sustainability is linked to innovation, it’s important to learn from other chapters and to have exchanges to build capacity and collaboration. 

Antonio Nkonjio Christophe: Suggested a database of modellers in each country to draw on as potential resources. 

Question 5: What are the challenges in the Horn of Africa chapter in terms of including multiple countries?

Getachew Teshome: Seed money from AMMnet has helped to bring together modellers and policy makers. 

Isaiah Agorinya: how does the Horn of Africa branch handle the need for in-person training when people are in different countries? Is it sustainable? 

Getachew Teshome: there is understanding that AMMnet will not always support every activity. They have discussed scaling up more like an association with membership fees. 

Question 6: How can the global North assist the global South seeing as there is less modelling in the south? How is the relationship between the north and south, how can Africa benefit from this interesting network?

Abdourahamane Diallo: It is important to transfer skills and strengthen capacity in Africa. More projects about malaria modelling should be hosted in Africa. 

Carmene Sandra Ngadjeu: First it’s necessary to create an active network of modellers including the anglophone and francophone modellers with collaboration. Training programmes are important. Northern modellers are far away. Also, it is important to focus within countries and ensuring there is a network. 

Question 7: Take-home messages?

Getachew Teshome: AMMnet is a good opportunity to bring people together.

Abdourahamane Diallo: The ultimate goal is to eliminate malaria and investments should be put into local skills development. Then the human capacity will exist for decision-making. Aid will not always come from the outside and the engagement should be sustained and localised.

Small Group Discussions

See here for slides presented after the small group discussions.

Closing remarks

Jaline Gerardin closed the meeting by thanking all who helped put it together, and looked forward to the next AMMnet meeting in Kigali in 2024.