Fig 1.
WHO IMAI Algorithm for Dehydration Assessment in Patients with Acute Diarrhea [10,13].
Fig 2.
Patient Enrollment.
Table 1.
Population Characteristics.
Table 2.
Clinical Symptoms and Signs on Arrival by Age.
Table 3.
Clinical Symptoms and Signs on Arrival by Dehydration Category.
Table 4.
Age-Specific NIRUDAK Model.
Fig 3.
Discrimination and calibration for NIRUDAK models based on training data.
The top row shows Receiver Operating Characteristic Curves. Shaded area gives 95% confidence regions. The middle row plots observed versus predicted by deciles of predicted probabilities. The vertical lines give the 95% confidence intervals for the observed probability in each decile. The bottom row gives histograms of the predicted probabilities.
Table 5.
Full and Simplified NIRUDAK Models.
Fig 4.
Comparison of m-indices computed from NIRUDAK full and simplified models (dark green) and WHO IMAI algorithm (light green). The m-index for each model is computed both from those observations from the original dataset that were included in each bootstrap sample (training m-index) and from those observations excluded from each set (testing m-index). Each histogram shows the distribution of m-indices derived from the bootstrap samples. Histograms for the WHO IMAI algorithm are the same in the left and right columns of the figures. The bar at the top of each histogram gives the mean m-index and a 95 percent confidence interval derived from the bootstraps.