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 research team, led by Md. Imam Sohel Hossain of the Bangladesh Council of Scientific and Industrial Research, mapped the flood extent using Synthetic Aperture Radar (SAR) images from the Sentinel-1 satellite, processed on the Google Earth Engine cloud computing platform. Unlike optical sensors, radar penetrates monsoon cloud cover and works day and night, making it the tool of choice for flood detection in a region where persistent storm clouds routinely block conventional satellite photographs. The team compared pre-flood images from August 1 to 18 with post-flood images from August 20 to September 1, 2024, and used only the VH polarization channel, which is particularly sensitive to smooth standing water and to double-bounce scattering from flooded vegetation.

The method relied on a bi-temporal change-detection ratio. Radar backscatter values were converted from decibels to linear power, and a pixel-wise ratio image was computed by dividing the post-flood backscatter by the pre-flood value. This ratio approach corrects for scene-wide radiometric differences and seasonal changes in surface conditions. To identify flooded pixels automatically, the researchers applied Otsu’s algorithm, a classic image-segmentation technique that selects the threshold maximizing the variance between flooded and non-flooded classes in a 255-bin histogram. The computed threshold of 1.039 classified pixels where post-flood backscatter had dropped dramatically relative to the pre-flood baseline, since calm water acts as a near-specular reflector that bounces radar energy away from the satellite.

Validation was rigorous. One hundred stratified random validation points, fifty per class and generated with a fixed random seed for full reproducibility, were compared against independent visual interpretation of three-meter PlanetScope imagery acquired during the flood period, supplemented by gauge-measured water levels and field records. The resulting map achieved 94.0 percent overall accuracy, a Kappa coefficient of 0.88, and an F1 score of 0.94 for the flooded class, with errors evenly balanced between false positives and false negatives. Because the user’s accuracies of both classes were symmetric, an area-weighted accuracy estimator returned the same 94.0 percent, confirming that the map faithfully represented the true distribution of inundation rather than merely an artifact of the validation design.

The radar evidence itself was corroborated by striking backscatter statistics. Across the study area, the mean VV backscatter fell from −8.655 decibels to −9.79 decibels after the flood, while VH fell from −16.162 to −17.817 decibels, both signatures of smoother, water-covered surfaces replacing vegetated or rough land. The timing of the radar-detected inundation matched river records closely: the Gumti-Burinadi gauging station at Cumilla recorded its monthly maximum water level of 12.57 meters above mean sea level on August 23, up from a July maximum of 9.21 meters, while the Selonia station on the Little Feni River peaked at 5.92 meters on August 25, far above its July maximum of 3.42 meters.

What caused such an extraordinary event in a region whose last major flood occurred in 1988? The hydro-meteorological evidence points primarily to extreme rainfall over the upstream Tripura catchment in neighboring India, where rainfall exceeded 430 millimeters in a short period during the flood-generating window. Cumulative rainfall during August 19 to 24 reached approximately 4,226 millimeters in the Tripura catchment, dwarfing local totals of roughly 1,355 millimeters in Cumilla, 534 millimeters in Feni, and 1,289 millimeters in Noakhali. Pearson correlation analysis revealed a strong positive association, around 0.82, between cumulative rainfall and district inundation extent, and the peak water level in the Gumti River coincided almost exactly with the period of maximum flooding detected from orbit.

The flood was therefore not primarily a product of local rain. Runoff from the transboundary Gumti and Little Feni catchments surged into low-lying Bangladeshi floodplains already saturated by months of monsoon rainfall, which had reduced infiltration capacity and accelerated surface runoff. Media attention focused heavily on possible releases from the Dumbur Dam in Tripura, but the researchers are careful on this point: without access to verified reservoir operation records or dam release hydrographs, the dam’s contribution remains a plausible but hydrologically unverified factor. They likewise decline to attribute the event directly to climate change, noting that formal attribution would require long-term hydroclimatic analysis, though climate change remains an important context for intensifying extreme monsoon precipitation.

The spatial pattern of damage was far from uniform. Cumilla, sitting squarely in the Gumti River floodplain with extensive low-lying agricultural land, bore the brunt: 55.3 percent of its area, or 1,694.3 square kilometers, was inundated, 62.5 percent of its cropland was damaged, 52.5 percent of its settlements were affected, and 13.4 percent of its population, some 699,839 people, were exposed. Feni saw 31.6 percent of its area flooded, affecting 35.3 percent of its cropland and about 116,946 residents, while Noakhali recorded 24.6 percent inundation with roughly 159,325 people exposed. Cumulatively, 53.05 percent of cropland and 52.75 percent of settlements across the three districts were hit, figures that underline how severely agriculture dominated the loss profile, in contrast to northeastern Bangladesh’s haor floods, where displacement and prolonged wetland inundation dominate.

Beyond the numbers, the study highlights a subtler danger: the risk created by the absence of floods. The authors argue that three decades of quiet conditions eroded community preparedness and risk perception, a phenomenon they link to the concept of hazard memory loss documented in long-interval flood zones elsewhere, such as the Paris region. Worse, recent urbanization in southeastern Bangladesh has ignored the region’s latent susceptibility, with new development spreading across paleo-channels and low-lying basins near silted rivers and inadequate drainage. The prolonged two-week waterlogging that followed the rapid onset reflected the flat floodplain’s limited drainage capacity rather than continuous rainfall, and unplanned growth, river sedimentation, and encroached drainage channels likely amplified both the extent and the persistence of the flooding.

The authors propose a two-track response. In the short term, they emphasize rapid relief and rescue, emergency food and water distribution, restoration of transport and power infrastructure, and deployment of mobile health units to prevent disease outbreaks. Over the long term, they call for regular dredging of the silted Gumti, Feni, and Meghna rivers, modern urban drainage systems, restoration of wetlands as natural flood buffers, community-based early warning systems that blend digital platforms with local knowledge, and flood-resilient construction such as elevated housing and reinforced embankments. At the policy level, they advocate Integrated Water Resource Management and stronger transboundary cooperation with India through the Joint Rivers Commission, including real-time sharing of rainfall, flow, and dam discharge data. Pointing to international models such as the Netherlands’ Room for the River program, China’s Sponge City initiative, and nature-based solutions adopted across Australia, New Zealand, the United Kingdom, and the United States, the team argues that Bangladesh could combine satellite-based monitoring, emerging technologies such as artificial intelligence and the Internet of Things, and ecosystem restoration to confront a risk landscape in which, as this flood proved, low exposure no longer means low risk.

Subject of Research: Rapid satellite-based assessment of the August 2024 flash flood impacts on population and croplands in southeastern Bangladesh

Article Title: Flash floods in traditionally less flood prone areas of southeastern Bangladesh

Article References: Hossain, M. I. S., Mostafa, M. G., Rana, M. S., & Hossain, M. S. (2026). Flash floods in traditionally less flood prone areas of southeastern Bangladesh. Discover Geoscience, 4(1), Article 375. https://doi.org/10.1007/s44288-026-00752-7

Image Credits: AI Generated

DOI: 10.1007/s44288-026-00752-7

Keywords: flash flood, Bangladesh, Sentinel-1 SAR, Google Earth Engine, flood mapping, Cumilla, Feni, Noakhali, Tripura rainfall, cropland damage, disaster risk management, flood exposure