The research, led by Md. Mamun Miah and colleagues including Kabir Hossain, Hafiz T. A. Khan, and H. M. Shahadat Ali, focused on two of Bangladesh’s most flood-prone districts, Noakhali and Feni. Both lie in the country’s coastal delta region, where monsoon rains, tidal surges, and river overflow regularly inundate villages and overwhelm sanitation infrastructure. The team conducted a cross-sectional survey, going door to door to interview 501 respondents selected through simple random sampling. The face-to-face interviews captured a detailed picture of each household: who lived there, what they earned, where their drinking water came from, how they treated it, how often and how severely floods reached them, and whether medical help and community preparedness programs were within reach.

The headline finding was stark. In these flood-affected areas, 63.7 percent of surveyed households—319 of the 501—reported at least one member suffering from a waterborne disease. That figure underscores why the researchers argue that conventional, reactive approaches to post-flood health crises are insufficient. If nearly two out of three households are affected, interventions based on guesswork or broad geographic targeting risk wasting scarce resources on families at comparatively lower risk while missing those most vulnerable. A predictive model, the authors reasoned, could sharpen the focus.

Before any algorithm could be trained, the raw survey data required careful preparation. The researchers ran a sequence of feature selection techniques to distill dozens of potential variables into a compact, informative set. They began with correlation analysis to detect redundant variables, then assessed multicollinearity using the variance inflation factor, a standard diagnostic that flags predictors which overlap too heavily with one another. Mutual information analysis followed, measuring how much each variable contributed non-linear information about disease outcomes. Finally, the team applied Recursive Feature Elimination with Cross-Validation, or RFECV, an iterative method that repeatedly trains a model, drops the least useful feature, and tests whether predictive performance holds up across data splits. The process retained ten predictors: age, gender, monthly household income, household size, flood frequency, flood severity, source of drinking water, availability of medical assistance, water purification method, and community preparedness for preventing waterborne diseases.

With the feature set fixed, the researchers benchmarked five widely used machine learning algorithms: Logistic Regression, Random Forest, Support Vector Machine, K-Nearest Neighbors, and Extreme Gradient Boosting, better known as XGBoost. Each model was optimized through randomized hyperparameter tuning, a search strategy that samples combinations of algorithm settings rather than exhaustively testing every possibility, paired with five-fold cross-validation to ensure the tuning generalized beyond a single data split. The models were then evaluated on an independent test set they had never seen during training, using a battery of metrics including accuracy, precision, recall, specificity, F1-score, and the area under the receiver operating characteristic curve, or ROC-AUC, which captures how well a model separates affected from unaffected households across all decision thresholds.

The Random Forest classifier emerged as the strongest performer, delivering the most balanced results on the test set. It achieved an accuracy of 61.4 percent, precision of 71.9 percent, recall of 64.1 percent, specificity of 56.8 percent, an F1-score of 67.8 percent, and a ROC-AUC of 0.643. Those numbers tell a nuanced story. The precision figure is particularly notable: when the model flags a household as high risk, it is right roughly seven times out of ten, which matters greatly in resource-constrained settings where misdirected aid carries real costs. The ROC-AUC of 0.643 indicates performance modestly better than random guessing—useful, but not infallible. The researchers were transparent about this, and they also tested more elaborate approaches, including stacking and weighted soft voting ensembles that combine multiple models. Neither outperformed the single Random Forest, a reminder that in applied machine learning, complexity does not automatically buy accuracy, especially with moderate-sized survey datasets.

What elevates the study beyond a standard prediction exercise is its commitment to explainability. Black-box models are notoriously difficult for health officials to act on, so the team turned to SHAP—SHapley Additive exPlanations—a technique rooted in cooperative game theory that assigns each feature a contribution value for every individual prediction. The SHAP analysis identified flood severity, community preparedness, water purification practices, and socio-economic factors as the most influential drivers of predicted risk. In practical terms, households facing the deepest or most prolonged flooding, lacking effective water treatment, and situated in communities without organized prevention efforts were consistently ranked as most vulnerable, with income and household size shaping the risk profile as well.

These findings carry immediate implications for how flood response is organized in Bangladesh and beyond. Rather than distributing water purification tablets or deploying medical teams uniformly across an affected district, health authorities could use a trained model to triage, prioritizing households whose characteristics match the high-risk profile identified by SHAP. The same logic supports early warning strategies: because flood frequency and severity are among the top predictors, seasonal forecasts and river gauge data could feed into risk models before floodwaters arrive, allowing pre-positioning of supplies in the most exposed communities. The emphasis on community preparedness as a key predictor also suggests that investments in local education, sanitation planning, and coordinated response networks yield measurable reductions in disease risk that algorithms can detect.

The authors are careful to frame the framework as promising but not yet deployment-ready. External validation—testing the model on data from other flood-affected regions—is required before it can be trusted at scale, and the moderate ROC-AUC signals that unmeasured variables, from pathogen concentrations in specific water sources to individual hygiene behaviors, likely influence outcomes. Still, the study demonstrates a replicable pipeline: rigorous survey design, statistically grounded feature selection, hyperparameter-optimized modeling, and transparent interpretation. All analyses were conducted in Python, and the methods are documented in enough detail to be adapted by other research teams working in low- and middle-income countries where waterborne disease remains a leading burden.

As climate change intensifies monsoon variability and sea-level rise pushes saline floodwater deeper into the Bengal delta, the stakes of this kind of predictive public health work will only grow. Bangladesh has long been on the front line of climate adaptation, and studies like this one point toward a future where the response to a flood begins not when the first patients arrive at a clinic, but when a model, trained on the experiences of hundreds of households, flags which doors relief workers should knock on first. The 319 affected families in Noakhali and Feni are, in that sense, more than statistics; they are the training signal for a smarter, faster defense against the diseases that follow the water.

Subject of Research: Machine learning prediction of household-level waterborne disease risk in flood-affected districts of Bangladesh

Article Title: Predicting households risks of waterborne diseases in flood-affected areas using machine learning: A study of Noakhali and Feni, Bangladesh

Article References: Miah, M. M., Hossain, K., Khan, H. T. A., & Ali, H. M. S. (2026). Predicting households risks of waterborne diseases in flood-affected areas using machine learning: A study of Noakhali and Feni, Bangladesh. PLOS Water, 5(8), e0000542. https://doi.org/10.1371/journal.pwat.0000542

Image Credits: AI Generated

DOI: 10.1371/journal.pwat.0000542

Keywords: waterborne diseases, machine learning, Bangladesh, flooding, Random Forest, SHAP, public health, feature selection, XGBoost, early warning, Noakhali, Feni

The Brazilian Semiarid, or semiarido, is one of the most densely populated semi-arid regions on Earth, home to millions of people whose taps depend on erratic rains rather than reliable piped networks. Climate change is tightening that squeeze: rainfall arrives in fewer, more violent bursts, evaporation strips shallow reservoirs, and drought years arrive in punishing clusters. Wet seasons deliver most of the year’s water in a handful of months; the rest of the year becomes a slow negotiation with whatever remains. For households outside the public water supply network—the norm in the rural communities examined here—drinking and washing water is assembled from a shifting mosaic of sources. Rainfall cisterns store the wet season’s bounty beside the house; hand-dug wells and surface reservoirs supply bathing and cleaning; tanker trucks and the Integrated Rural Sanitation System, known locally as SISAR, fill the gaps. Each source carries its own chemical signature and its own microbial community, and each swells and fades with the seasons. That variability, the new study argues, is not merely an inconvenience of rural life. It is a public health mechanism—one that quietly converts water scarcity into sickness.

To capture that mechanism in action, the researchers conducted descriptive field research at three moments in 2025—February, April, and September—bracketing the transition from the rainy season into the depths of the dry. Working in communities made up entirely of households without access to the public water supply network, they administered structured questionnaires on water use, storage, and handling, and collected water samples from every domestic source then in service. In the laboratory, each sample was characterized physicochemically—temperature, oxidation-reduction potential, alkalinity, electrical conductivity, and related parameters—and screened microbiologically for total coliforms and E. coli, the classic indicator of fecal contamination. Hydroclimatic analysis framed the picture, characterizing the regional climate scenario against which household behavior unfolded. The team then pressed the data through a battery of statistical tests: the Kruskal-Wallis test with Dunn’s post-hoc comparisons for physicochemical measurements, Fisher’s exact test for microbiological findings, and Cochran’s Q and chi-square tests to evaluate the incidence of diarrhea across the three communities.

The hydroclimatic backdrop was stark. Cumulative precipitation in 2025 fell by 35.2 percent in Quixeramobim and 27.9 percent in Quixadá relative to 2024, a shortfall that rippled immediately through household decisions. Cisterns that had filled during the rainy months drained quickly, and the depletion was accelerated by a deeply rooted survival strategy: sharing. Families whose stores ran dry drew on the cisterns of neighbors and relatives, spreading the available water thinner but shortening its life for everyone. Rainfall cistern storage fluctuated sharply across the seasons, and the shared tanks emptied faster than any single household’s would have. As storage fell, the hierarchy of water use shifted in a predictable but consequential direction. Families continued to reserve cistern water for drinking whenever possible, but hygiene—bathing, washing dishes, laundering clothes—was pushed outward onto surface reservoirs and wells, sources that are typically more exposed to animals, sediment, and runoff.

Those behavioral adaptations matter because the physicochemical data reveal how different the fallback options really are. Across seasons and communities, the most striking discriminator was oxidation-reduction potential, or ORP—a measure of water’s tendency to accept or donate electrons, expressed in millivolts. In practical terms, ORP reflects the oxidative capacity of water: higher values generally indicate conditions hostile to microbial survival, often associated with freshly treated or well-aerated water, while lower values signal reducing environments where organic matter and microorganisms can persist. Because ORP responds quickly to disinfectants such as chlorine, it is often used as a rapid proxy for microbial safety in drinking-water systems. The analyses uncovered significant seasonal and inter-community variation in water quality, particularly in ORP, and significant differences across all physicochemical parameters between source types. Water delivered by tank trucks and through the SISAR system registered higher temperatures and higher ORP, while wells stood out for their higher alkalinity and electrical conductivity—a chemical fingerprint of long groundwater residence times, dissolved mineral loads, and, in many semiarid aquifers, progressive salinization under drought and intensive pumping.

Electrical conductivity and alkalinity deserve a closer look, because together they tell a story of geology meeting water stress. Conductivity climbs as dissolved ions—calcium, magnesium, sodium, bicarbonate, chloride—accumulate in water; the longer water lingers in contact with rock, the more dissolved material it picks up. In the Semiarid, where aquifers are often brackish and recharge is thin, wells frequently yield water that is chemically unpalatable long before it becomes biologically dangerous. High alkalinity buffers pH but also signals substantial dissolved mineral content. Neither property makes water unsafe on its own, yet both shape taste, discourage consumption, and interact with disinfection: chlorination, the backbone of treatment for trucked and SISAR supplies, behaves differently in high-alkalinity, high-temperature water, where the free chlorine that kills pathogens is consumed faster and works less predictably. The elevated temperatures and ORP recorded in trucked and SISAR water suggest supplies that have been recently treated but are also transported and stored in conditions that can erode that protection quickly.

The microbiological results cut through every comforting assumption. Total coliform contamination—a broad family of bacteria used as sentinels for environmental and fecal pollution—was widespread, and the persistent presence of E. coli, a definitive marker of fecal contamination, was observed across every source type. No category of water was consistently clean. Yet there was a clear gradient: rainfall cisterns demonstrated the lowest levels of E. coli contamination of any source, confirming their status as the best option available to these families. The finding carries a warning as well as a reassurance. Rainwater harvested from roofs is vulnerable at every step of its journey—dust, leaves, and animal droppings on catchment surfaces, the first flush of each storm washing months of accumulated debris into storage, and above all the hands, cups, ladles, and containers that move water from cistern to mouth. Cistern water arrives relatively clean; whether it stays that way depends entirely on storage conditions and domestic water handling, which the study identifies as likely contributors to the contamination it measured.

The health consequences were continuous rather than episodic. Diarrhea cases were reported in all three communities throughout the study period, and a quarter of all cases affected children—underscoring the heightened vulnerability of a group whose smaller bodies dehydrate faster and whose immune defenses are still under construction. Statistical evaluation of the illness data, using Cochran’s Q and chi-square tests, reinforced a pattern that clinicians in the region know well: diarrheal disease in the Semiarid is not a rare catastrophe triggered by a single spectacular contamination event. It is a chronic burden, ebbing and flowing with the seasons, sustained by the constant rotation of households through water sources of uneven quality. Every trade-down from a cistern to a well or a reservoir is a fresh roll of the dice, and over months the dice accumulate. A quarter of cases striking children is a signal that this burden falls hardest on those least equipped to bear it.

The study’s central conclusion is quietly subversive for water policy. Brazil’s semiarid region has spent decades building rainwater cisterns as the cornerstone of rural water security, and the new data validate that investment: cisterns are relatively safer sources. But the results also show that safety on paper evaporates in practice. During the dry season, cisterns are frequently replaced by less reliable alternatives—wells, surface reservoirs, tanker trucks—precisely because scarcity forces the swap, and the act of switching, sharing, and storing water under stress appears to raise microbiological contamination further. Storage conditions, domestic water handling, and adaptive strategies born of scarcity likely contribute to increased contamination, the authors conclude. In other words, the concrete infrastructure is only half of the water-safety system; the other half is the invisible household choreography of fetching, sharing, decanting, and rationing that determines what actually arrives in the cooking pot.

The implications reach well beyond the communities studied. The research, funded by Brazil’s National Council for Scientific and Technological Development and the Dutch Research Council, suggests that protecting rural health in a drying climate demands more than building cisterns: it requires safeguarding the water inside them through maintained gutters and first-flush systems, covered and regularly cleaned storage, point-of-use treatment such as chlorination or filtration during the high-risk months, and hygiene practices that travel with the water. It means treating tanker deliveries and small rural sanitation networks as water-quality infrastructure, not merely water-quantity infrastructure, and monitoring them with equal rigor. And it means planning explicitly for the dry season, because the study shows that the moment families are forced to trade down to riskier sources is the moment disease follows. In Brazil’s Semiarid, water scarcity and diarrheal disease are not separate problems to be solved by separate agencies. They are a single problem, moving together from source to sickness with every shift in the rain.