Impact of climate and acquired immunity on malaria burden in Senegal and The Gambia: an agent-based modeling study

Malaria is a disease that does not treat every person the same way. When a mosquito carrying the parasite bites a human, the outcome depends heavily on how many times that person has been bitten before. Over a lifetime, repeated exposure teaches the body to fight the infection, but this protection builds slowly and is never perfect. A young child, with little history of exposure, will likely fall ill from a bite that an adult, who has survived many seasons of the disease, might shrug off. This protection is not a fixed trait; it is a record of personal history shaped by the local environment. Because the mosquitoes that carry malaria need water to breed, the amount of rain in a region determines how many mosquitoes there are, which in turn dictates how often people get bitten and how quickly they build up their defenses.

Scientists who study these diseases often use computer models to predict how many people will get sick and to plan how to stop the spread. For a long time, these models have relied on a simplifying assumption: they treat the entire population as if everyone has the exact same average level of protection. In this view, a child and an adult in the same village are given the same immune defense, calculated as a middle ground for the whole community. While this makes the math easier, it ignores the reality that immunity is a personal journey. If the models are wrong about who is protected and who is not, the plans based on them might miss the people who need help the most.

A team of researchers set out to test whether this common simplification leads to significant errors in understanding the disease burden in Senegal and The Gambia. To do this, they built a new kind of computer simulation called VECTRI-ABM. Unlike older models that treat the population as a single block, this new framework is a grid-based system that couples a climate-driven vector component with an agent-based representation of the human population, resolving immunity at the individual level. The model tracks mosquitoes using climate data and was evaluated against reported entomological inoculation rates, Malaria Atlas Project estimates of parasite rates, and confirmed cases from Senegal's National Malaria Control Program. The researchers ran the simulation across a wide area that stretches from the dry Sahel to the wetter tropical zones, capturing the full range of climates found in Senegal and The Gambia.

The team first checked their new model against real-world data collected from two villages, Dielmo and Ndiop, where scientists had carefully recorded mosquito bites and malaria cases for over a decade. Once the simulation matched these real-life observations, the researchers ran it twice across the entire region. In the first run, every virtual person developed immunity based on their own personal history of bites. In the second run, they forced every person to share the same average level of immunity that the whole community possessed, mimicking the standard approach used in most current models. By comparing the two runs, they could see exactly what was lost when the individual histories were erased.

The results showed that the standard approach of averaging immunity creates a distorted picture of the disease. When the researchers used the simplified model, it consistently underestimated how many children were getting sick, particularly in the wettest areas where the mosquitoes are most abundant. At the same time, the simplified model overestimated the number of adults who were carrying the parasite without showing symptoms. This matters because public health strategies often target specific groups; if a model suggests children are less sick than they really are, resources might be directed away from the people who need them most. The study found that the error introduced by averaging immunity is not the same everywhere; it depends on the local climate. In the regions with the highest rainfall and the most intense transmission, the mistake in predicting the burden on children was the largest.

However, the study also identified where the simpler model still works well. For people older than fifteen years, the two versions of the simulation largely agreed on the number of clinical cases. By this age, most people in these regions have built up enough immunity that their individual histories matter less for predicting whether they will get sick. The main difference remained in the younger population, where the accumulation of protection is still in progress. The researchers concluded that while the simple, averaged approach might be sufficient for understanding the disease in older adults, it fails to capture the true burden on children, especially in the wettest parts of the region. This suggests that to accurately plan interventions and understand the spread of malaria, models must account for the fact that immunity is a personal story, not a community average.