Joec-Tech/NDUTH_Model

The Model describes the Cancer cases in Niger Delta University Teaching Hospital, Bayelsa State, Nigeria # NDUTH Cancer Survival Policy Scenario App — Version 2 This package contains a standalone HTML/CSS/JavaScript policy dashboard, the complete R modelling scripts and the bundled synthetic Excel workbook. ## Dynamic scenario interpretation The Scenario Builder includes a plain-language interpretation box that updates whenever a slider or dropdown changes. It converts survival percentages into: - expected patients alive at three years; - expected patients alive at five years; - differences from current care; - projected median-survival change; - five-year follow-up workload; - chemotherapy and radiotherapy service implications; - the operational change needed to achieve the selected early-diagnosis target. The explanation retains the synthetic-data and non-causal-treatment warnings. ## Model-comparison display The Model Evidence page now shows the predicted survival curves from all four candidate AFT distributions: - Exponential - Weibull - Log-normal - Log-logistic The chosen model is displayed as a thicker solid line. The other models use thinner patterned lines. Ranking cards show AIC, delta AIC and BIC, with the selected model clearly labelled. ## Cox–Snell 45-degree diagnostic The Model Evidence page also includes Cox–Snell residual curves for all four candidate AFT models. Each curve is compared with a 45-degree diagonal line. A well-fitting model should follow this reference line reasonably closely. The selected Weibull model is displayed as a thicker solid line. Diagnostic cards report the mean absolute deviation from the diagonal; lower values indicate closer agreement. The tail is limited at the pooled 95th percentile of the residuals because goodness-of-fit curves become unstable where very few patients remain at risk. This is a 45-degree reference line, not a 45% significance threshold. ## Policy controls - **Early-stage diagnosis:** slider - **Treatment pathway:** dropdown - **Treatment timeliness:** dropdown, displayed as exploratory evide …