The study, led by Bamidele Nyemike Ogunro of the University of Ibadan together with colleagues from Corona Management Systems in Abuja, the Federal University of Technology Akure, and Nigeria’s National Malaria Elimination Programme, emerged from the Nigeria Malaria Modelling Fellowship, an initiative funded by the Bill & Melinda Gates Foundation. The work addresses a persistent gap in malaria strategy: while bed nets and preventive drugs act on adult mosquitoes or on the parasite once infection has occurred, larva source management attacks the vector at its aquatic stages, killing larvae in ponds, puddles, irrigation channels and other standing water before they can mature into biting adults.

At the heart of the analysis is a deterministic SEIR-based compartmental model, a mathematical framework that divides the human population into susceptible, exposed, infectious and recovered classes, and couples these to the mosquito population across its developmental stages, from eggs and larvae to pupae and adults. This human-vector coupling is what allows the model to capture the feedback loop that drives malaria transmission: more infected people mean more parasites available to mosquitoes, and more adult mosquitoes mean more infectious bites. Seasonality, a defining feature of malaria in the Sahelian zone where Zamfara sits, was incorporated through a Gaussian function that modulates vector dynamics across the year, producing the characteristic seasonal surges that strain health facilities during and after the rainy season.

To ground the model in reality, the researchers calibrated it against historical malaria incidence data from 2020 to 2024, supplied by the National Malaria Elimination Programme. Calibration is the process by which model parameters, such as biting rates, development times and recovery rates, are adjusted until simulated case numbers track observed surveillance data. The team then computed the basic reproduction number, R0, using the next-generation matrix method, a standard technique in mathematical epidemiology that identifies how many secondary infections a single infectious individual generates in a fully susceptible population. For Zamfara under the current intervention mix, R0 came out at 1.18, a value indicating persistent but not explosive transmission, consistent with a setting where control measures are holding the epidemic back but not breaking it.

Because models are only as trustworthy as their sensitivity to assumptions, the team performed a one-at-a-time local sensitivity analysis, varying each parameter individually by plus or minus ten percent while holding all others at baseline values. The results were unambiguous: parameters governing adult mosquito survival and abundance exerted the strongest influence on malaria incidence, which is precisely the life stage that larva source management ultimately reduces by throttling the supply of new adults. This finding aligns with the broader logic of integrated vector management, which holds that reducing vector density complements, rather than replaces, interventions that target the parasite or protect individuals.

A critical test of any predictive model is whether it can forecast data it has never seen. The researchers therefore carried out out-of-sample validation, simulating the calibrated model against independent empirical malaria incidence data from 2025. The model accurately synchronized with the timing of the 2025 seasonal peak, a demanding test given how sharply transmission rises and falls across the year in Zamfara. It did, however, underpredict absolute case volumes during the peak months, a discrepancy the authors attribute to single-year transmission volatility, the inherent randomness of real-world epidemics that a deterministic model smooths over. The validation nonetheless supports the model’s use as a strategic planning tool for comparing intervention scenarios.

The headline result concerns what happens when larva source management is layered onto existing interventions. Simulations showed that adding LSM significantly reduced malaria incidence, with up to 19.3 percent of cases averted when 50 percent of breeding sites were covered. But the relationship between coverage and impact proved strikingly non-linear. Diminishing returns set in beyond 20 percent implementation, meaning each additional percentage point of coverage bought progressively less epidemiological benefit. The sweet spot, according to the analysis, lies at around 30 percent implementation, a threshold the authors identify as balancing epidemiological gain against efficiency loss, with an estimated 132,000 malaria cases averted annually at that level. For resource-constrained health systems, this kind of efficiency analysis is often more actionable than a simple maximum-impact figure, because it identifies where the marginal naira of programme spending stops delivering proportional returns.

Timing emerged as another crucial practical consideration. The impact of larva source management did not appear immediately: measurable reductions in incidence became apparent only after roughly five months of continuous implementation. This lag reflects the biology of the intervention, which works by progressively depleting the adult mosquito population rather than instantly killing infected vectors or curing patients. For programme planners, the implication is that LSM must be initiated well before the transmission season peaks, and that expectations must be managed during the early months of a campaign when case counts may not yet show improvement.

The model also revealed an important distributional dimension of the intervention’s benefits. Larva source management produced higher case reductions among children, the group most vulnerable to severe malaria and malaria-related mortality. Because severe disease in children drives hospital admissions, school absenteeism and long-term developmental harm, an intervention that disproportionately protects this age group delivers benefits beyond raw case counts. This finding strengthens the case for LSM as a child-focused equity measure, complementing seasonal malaria chemoprevention, which already targets children under five with preventive antimalarial drugs during peak transmission months.

The authors conclude that malaria transmission in Zamfara State remains persistent under the current intervention mix, and that incorporating larva source management into the existing strategy would be beneficial for control. The study was conducted under ethical approval from Nigeria’s National Health Research Ethics Committee and in accordance with the Code of Professional Ethics for Simulationists of the Society for Modeling and Simulation International, using only aggregated secondary data for calibration. As Nigerian health authorities weigh how to push toward elimination goals in high-burden northwestern states, this analysis offers a quantified, locally calibrated argument that the fight against malaria should begin where the mosquito does: in the water.

Subject of Research: Mathematical modelling of larva source management as a supplementary malaria control intervention in Zamfara State, Nigeria

Article Title: A compartmental modelling approach for incorporating larva source management into malaria control strategy in Zamfara State, Nigeria

Article References: Ogunro, B. N., Adegbite, G., Obi, C., Eche-George, A., Eshikhena, G., Ojielo, N., Akoma, D., Gayawan, E., Nelson, E., Ogbulafor, N. C., Ntadom, G., & Kaduru, C. (2026). A compartmental modelling approach for incorporating larva source management into malaria control strategy in Zamfara State, Nigeria. BMC Infectious Diseases. https://doi.org/10.1186/s12879-026-14536-4

Image Credits: AI Generated

DOI: 10.1186/s12879-026-14536-4

Keywords: malaria, larva source management, mathematical modelling, SEIR model, vector control, Zamfara State, Nigeria, basic reproduction number, sensitivity analysis, seasonal malaria chemoprevention, integrated vector management, BMC Infectious Diseases

The core of the modeling effort is a multi-component simulation framework that links several distinct biological and economic modules. On the vector side, the model tracks Aedes and Culex mosquitoes using Susceptible Exposed Infectious dynamics, capturing the biology of the two genera that drive transmission in nature. Aedes mosquitoes, whose eggs can survive dry periods and hatch when flooding rains arrive, are classically responsible for initiating outbreaks, while Culex species amplify transmission among animals once an epizootic is underway. On the host side, livestock populations follow Susceptible Exposed Infectious Recovered dynamics, layered on top of a herd demography module that accounts for births, deaths and the gradual turnover of animals that changes the susceptible fraction of the population between outbreaks.

What distinguishes this framework from many earlier RVF models is the explicit coupling of these infection dynamics to end-market processes and demand-side behavior. Livestock producers in pastoral systems such as Ijara County do not operate in a vacuum: when disease strikes, buyers retreat, prices fall and household incomes contract, and these economic signals feed back into decisions about animal holdings and sales. By incorporating demand shocks and behavioral feedback directly into the simulation, the authors allow the economic consequences of an outbreak to unfold endogenously rather than being imposed as fixed external costs. This design reflects the study’s central premise that livestock diseases present systemic threats to animal health, rural livelihoods and national economies, and that nonlinear feedbacks, time lags and cross-sectoral interactions must be captured to evaluate policy honestly.

The simulation runs over a ten-year horizon with a daily time step, a resolution fine enough to represent the rapid course of mosquito-borne transmission and the timing of vaccination campaigns with realistic precision. Parameterization draws on both primary and secondary data, including El Niño associated rainfall patterns that historically precede RVF epizootics in East Africa and the results of producer surveys conducted in the study area. The heavy rains associated with El Niño events create the flooded habitats in which Aedes eggs hatch and Culex populations explode, so anchoring the model’s environmental forcing to these observed patterns helps ensure that simulated outbreaks resemble real ones. The authors report that the model reproduces key outbreak features consistent with historical RVF events, an important credibility check before using it to compare policies that have never been implemented at scale.

With this validated platform in place, the researchers compared alternative vaccination strategies against a business-as-usual baseline and a no-intervention scenario. The business-as-usual strategy corresponds to the reactive posture that has often prevailed in practice: vaccination is delayed until an outbreak is already underway, constrained by the logistical realities of mobilizing vaccines, cold chains and veterinary teams in remote pastoral areas. The preventive alternatives include annual vaccination of susceptible animals, biannual campaigns and triennial schedules, each representing different assumptions about budget, coverage and the frequency with which herds are refreshed with newly born, susceptible stock.

The results are striking in their asymmetry. Under business-as-usual delayed reactionary vaccination, the model shows only marginal improvements in herd recovery and producer income relative to doing nothing at all. By the time a reactive campaign takes effect, the explosive early phase of an RVF epizootic has typically already run its course through the most susceptible animals, so late vaccination buys little additional protection and recovers little of the lost income. In contrast, annual preventive vaccination with sufficient coverage of susceptible animals prevents simulated outbreaks entirely within the model horizon. Biannual and triennial strategies occupy an intermediate position: they reduce outbreak severity but do not fully eliminate risk, because gaps in coverage allow susceptible animals to accumulate and transmission to rekindle when vector populations surge.

Beyond the comparison of fixed schedules, the model illuminates the decisive importance of timing itself. Shortening the delay between the onset of an outbreak and the initiation of vaccination substantially reduces livestock losses and improves the trajectory of income recovery for producers. This finding quantifies a principle that outbreak responders have long understood intuitively: in a fast-moving vector-borne epidemic, every week of delay translates into animals infected that timely vaccination could have protected. The system dynamics formulation makes the cost of delay explicit, translating it into herd-level mortality and household-level economic outcomes rather than leaving it as an abstract warning.

For policymakers in Kenya and other RVF-endemic countries, the practical implications are considerable. Preventive vaccination is expensive and competes with other demands on limited veterinary budgets, and in years without outbreaks it can appear to be money spent for nothing. The model’s ex-ante comparisons offer a way to weigh that recurring cost against the episodic, catastrophic losses of uncontrolled epizootics, and to demonstrate that proactive programs outperform reactive ones in both disease control and economic resilience. The authors position the model as a decision-support tool for livestock health policy, one that can be rerun under different assumptions about rainfall, coverage, vaccine efficacy and market behavior to stress-test strategies before committing public resources to them.

Methodologically, the study illustrates why system dynamics approaches are gaining traction in animal health economics. Because the framework integrates vector ecology, host infection dynamics, herd demography and market processes in one coupled structure, it can reveal behaviors that emerge from interactions among components rather than from any single module. Time lags between rainfall triggers, mosquito population booms, livestock exposure and policy response are represented explicitly, as are the nonlinear thresholds that separate a contained incursion from a runaway epizootic. Conventional static impact assessments, which typically multiply observed losses by fixed prices, cannot capture these dynamics, and the authors argue that this limitation has systematically distorted the apparent value of preventive investment.

The research also speaks to the broader challenge of preparing for climate-sensitive disease threats. RVF outbreaks in East Africa are tightly associated with flooding events whose frequency and intensity may shift with a changing climate, and models of this kind provide a template for evaluating how preparedness investments perform across a range of environmental futures. For the pastoral communities of Ijara County and similar regions, where livestock constitute both income and a form of savings, the difference between a prevented outbreak and a delayed response is measured in household welfare as much as in veterinary outcomes. By demonstrating that annual preventive vaccination with adequate coverage can hold the virus at bay within the simulated decade, while reactive campaigns barely improve on inaction, the study offers a quantified case for shifting RVF policy from reaction to anticipation, and a modeling architecture that other endemic countries can adapt to their own epidemiological and economic circumstances.

Subject of Research: System dynamics modeling of Rift Valley fever vaccination strategies in Kenya

Article Title: Integrating epidemiological and economic dynamics to assess Rift Valley fever vaccination strategies: A system dynamics model from Kenya

Article References: Bahta, S., Wanyoike, F., Bett, B., & Rich, K. M. (2026). Integrating epidemiological and economic dynamics to assess Rift Valley fever vaccination strategies: A system dynamics model from Kenya. PLOS Complex Systems, 3(6), e0000101. https://doi.org/10.1371/journal.pcsy.0000101

Image Credits: AI Generated

DOI: 10.1371/journal.pcsy.0000101

Keywords: Rift Valley fever, system dynamics, vaccination strategy, Kenya, livestock health, epidemiological modeling, economic impact, Aedes mosquitoes, Culex mosquitoes, SEIR model, preventive vaccination, pastoral systems

The core problem the researchers tackle is deceptively simple to state but notoriously hard to solve. Time-series anomaly detection aims to flag observations that deviate sharply from normal behavior, and in surveillance data a surge in symptom-related calls may herald the onset of an epidemic. Yet the field is plagued by flawed evaluation practices. The widely used point-adjust protocol, for example, credits an algorithm with correctly identifying an entire anomalous segment if it detects just a single point within it, a leniency so extreme that even randomly generated anomaly scores can appear highly accurate under it. Point-wise metrics such as precision, recall, and F1 score also treat a late detection as equivalent to a complete miss and struggle to capture event-like anomalies that unfold over days or weeks. As the authors and prior benchmarking studies note, these weaknesses have inflated claims of progress and produced misleading rankings of detection methods.

To build a fairer testbed, the team turned to one of epidemiology’s oldest and most trusted tools: the SEIR model, which divides a population into susceptible, exposed, infectious, and recovered compartments governed by differential equations. Rather than trying to label real historical outbreaks, an exercise fraught with subjectivity and bias, they simulated three synthetic disease outbreaks and injected them into genuine surveillance data. Disease A was designed to resemble measles, with high transmissibility and symptoms including fever, cough, conjunctivitis, and rash. Disease B mimicked COVID-19, with moderate transmissibility and respiratory symptoms, while Disease C modeled mpox, a lower-transmission illness marked by fever, rash, and localized skin effects. Each simulated outbreak was embedded into the call-record time series starting at day 1500 of the dataset and lasting 100 days, with the transmission rate tuned so that roughly 1, 15, or 50 percent of the population recovered by the end of the simulation.

The translation from epidemic model to surveillance signal is where the framework shows its technical sophistication. For each symptom variable, the expected outbreak-related increase in calls was computed by summing, across the three modeled infectious stages, the number of infectious individuals multiplied by the probability that a person in that stage exhibits the symptom, the probability that a symptomatic individual calls the hotline, and a demographic weighting factor of 0.8 for adult variables and 0.2 for child variables. Because the injected signals are deterministic given the chosen parameters, the resulting benchmark is fully reproducible, allowing different detection algorithms to be compared on identical, objectively labeled outbreak events, something real-world surveillance data can almost never provide.

Against this benchmark, the researchers evaluated five reconstruction-based anomaly detectors: principal component analysis, a vanilla autoencoder, a variational autoencoder, an LSTM autoencoder, and a Transformer autoencoder. All of these learn a representation of normal temporal behavior and flag anomalies when reconstruction error, the mismatch between input data and the model’s attempt to reproduce it, spikes above a threshold set at the 99th percentile of training scores. The models were tested in two modes. In the offline setting, the entire time series is analyzed retrospectively. In the online setting, which mirrors real deployment, the model starts with one year of training data and then processes incoming batches sequentially, incorporating each evaluated batch into its training set and recalculating its anomaly threshold, an incremental expanding-window strategy chosen because ground-truth labels are unavailable during real surveillance.

The results reveal a clear hierarchy among the architectures and an honest picture of the difficulty of the task. In offline testing, every model detected all three injected outbreaks at the event level, but the Transformer autoencoder led consistently, achieving the highest point recall for all three diseases and the fastest time to detection, flagging Disease A after 61 days, compared with 74 to 78 days for the other models. In the harder online setting, the Transformer autoencoder again performed best on Disease A, with a point recall of 0.395 and detection after 60 days. For the weaker signals of Diseases B and C, however, its advantage narrowed or vanished: the LSTM autoencoder matched it on Disease B, and only those two models managed to detect Disease C at all. The authors attribute the Transformer’s edge to its ability to capture complex temporal and cross-variable dependencies, while noting that the variational autoencoder’s probabilistic constraints can smooth out subtle anomalies and that PCA, limited to linear relationships, was least sensitive.

Sensitivity analyses added crucial nuance. Lower transmission rates degraded detection across the board, and at the lowest rate only the Transformer autoencoder still registered the outbreak. More strikingly, the timing of the outbreak within the surveillance record mattered enormously. When a measles-like outbreak was injected during the early pandemic phase of the data, the online Transformer detector achieved a point recall of 0.722 at baseline reporting probability, and up to 0.950 when the reporting probability was raised. The same outbreak injected in the post-pandemic period yielded a recall of only about 0.4 and a detection delay of 60 days, and at a reduced susceptible fraction of 50 percent it went entirely undetected. Outbreak detectability, in other words, depends not just on the pathogen but on the surrounding statistical background, the signal-to-noise ratio of the moment.

To test the framework against reality, the team also ran the detectors on the unmodified dataset, which contains four expert-identified COVID-19 waves and recurrent seasonal outbreaks of influenza and norovirus between 2019 and mid-2023. Using DBSCAN clustering to group point-wise anomaly detections into temporally coherent events, the two best models, the Transformer and LSTM autoencoders, overlapped nearly all known outbreak periods, including all four norovirus seasons and three influenza seasons, though the Transformer missed the second COVID-19 wave entirely and both models detected only fragments of the fourth. Detections that matched no known outbreak were reported as unmatched clusters rather than dismissed as false positives, a deliberate design choice reflecting the fact that real surveillance data may contain undocumented epidemiological events, seasonal behavioral shifts, or reporting artifacts that no complete ground truth can capture.

The authors are candid about the framework’s limitations. The SEIR-based injection assumes a fully mixed population and simplified symptom-reporting behavior, so it cannot reproduce the full complexity of real epidemics, and the ground-truth onset is defined as the true start of transmission in the simulation rather than the point where changes become visible in the data, which deliberately penalizes the detectors and keeps performance metrics conservative. Keeping potentially anomalous observations in the online training stream risks contamination, whereby a prolonged outbreak gradually becomes absorbed into the learned definition of normal. The comparison was also restricted to reconstruction-based methods, so no claim of superiority over classical statistical surveillance tools such as CUSUM and EWMA, or over forecasting-based approaches, is being made. Future work, the team writes, should compare these families of methods directly, explore spatiotemporal neural networks to identify at-risk municipalities, and add explainability techniques to make the detections interpretable for public health decision-makers.

What the study ultimately delivers is less a single winning algorithm than a discipline. By pairing epidemiologically grounded synthetic outbreaks with event-oriented evaluation metrics, point recall, event recall, time to detection, and an overlap duration coefficient analogous to the Intersection over Union measure, the framework offers the outbreak-detection field something it has badly needed: a benchmark in which performance claims can be trusted, timeliness is measured honestly, and the metric matches the public health reality that interventions target outbreak episodes, not isolated data points. As syndromic surveillance streams grow larger and faster, frameworks of this kind may determine whether the next emerging pathogen is noticed in days rather than weeks.

Subject of Research: Real-time detection of infectious disease outbreaks in symptom surveillance data using epidemiology-guided machine learning anomaly detection

Article Title: An epidemiology-guided machine learning framework for real-time anomaly detection in disease outbreak monitoring using symptom surveillance data

Article References: Hashemi, A. S., Dietler, D., Ohlsson, M., & Björk, J. (2026). An epidemiology-guided machine learning framework for real-time anomaly detection in disease outbreak monitoring using symptom surveillance data. Machine Learning with Applications, 26, Article 101015. https://doi.org/10.1016/j.mlwa.2026.101015

Image Credits: AI Generated

DOI: 10.1016/j.mlwa.2026.101015

Keywords: anomaly detection, disease surveillance, SEIR model, syndromic surveillance, machine learning, Transformer autoencoder, LSTM autoencoder, outbreak detection, time series analysis, evaluation metrics, DBSCAN clustering, public health