ABSTRACT
Background: In Northern Province, Rwanda, stunting is common among children aged under 5 years. However, previous studies on spatial analysis of childhood stunting in Rwanda did not assess its randomness and clustering, and none were conducted in Northern Province. We conducted a spatial-pattern analysis of childhood undernutrition to identify stunting clusters and hotspots for targeted interventions in Northern Province. Methods: Using a household population-based questionnaire survey of the characteristics and causes of undernutrition in households with biological mothers of children aged 1-36 months, we collected anthropometric measurements of the children and their mothers and captured the coordinates of the households. Descriptive statistics were computed for the sociodemographic characteristics and anthropometric measurements. Spatial patterns of childhood stunting were determined using global and local Moran's I and Getis-Ord Gi* statistics, and the corresponding maps were produced. Results: The z-scores of the three anthropometric measurements were normally distributed, but the z-scores of height-for-age were generally lower than those of weight-for-age and weight-for-height, prompting us to focus on height-for-age for the spatial analysis. The estimated incidence of stunting among 601 children aged 1-36 months was 27.1 %. The sample points were interpolated to the administrative level of the sector. The global Moran's I was positive and significant (Moran's I = 0.403, p < 0.001, z-score = 7.813), indicating clustering of childhood stunting across different sectors of Northern Province. The local Moran's I and hotspot analysis based on the Getis-Ord Gi* statistic showed statistically significant hotspots, which were strongest within Musanze district, followed by Gakenke and Gicumbi districts. Conclusion: Childhood stunting in Northern Province showed statistically significant hotspots in Musanze, Gakenke, and Gicumbi districts. Factors associated with such clusters and hotspots should be assessed to identify possible geographically targeted interventions.
ABSTRACT
As found in the health studies literature, the levels of climate association between epidemiological diseases have been found to vary across regions. Therefore, it seems reasonable to allow for the possibility that relationships might vary spatially within regions. We implemented the geographically weighted random forest (GWRF) machine learning method to analyze ecological disease patterns caused by spatially non-stationary processes using a malaria incidence dataset for Rwanda. We first compared the geographically weighted regression (WGR), the global random forest (GRF), and the geographically weighted random forest (GWRF) to examine the spatial non-stationarity in the non-linear relationships between malaria incidence and their risk factors. We used the Gaussian areal kriging model to disaggregate the malaria incidence at the local administrative cell level to understand the relationships at a fine scale since the model goodness of fit was not satisfactory to explain malaria incidence due to the limited number of sample values. Our results show that in terms of the coefficients of determination and prediction accuracy, the geographical random forest model performs better than the GWR and the global random forest model. The coefficients of determination of the geographically weighted regression (R2), the global RF (R2), and the GWRF (R2) were 4.74, 0.76, and 0.79, respectively. The GWRF algorithm achieves the best result and reveals that risk factors (rainfall, land surface temperature, elevation, and air temperature) have a strong non-linear relationship with the spatial distribution of malaria incidence rates, which could have implications for supporting local initiatives for malaria elimination in Rwanda.
