<p>We present a framework for analyzing AIDS mortality rates across countries by incorporating temporal changes in antiretroviral therapy (ART) accessibility into both the predictive model and visualization. A gradient-boosted tree-ensemble model is used to predict mortality rates and enable more accurate cross-country comparisons. The framework combines a weighted loss function to address data sparsity across regions and obtains feature importance scores via post-processing. The model standardizes predictions into templated outputs for consistent comparison and creates multiple visualizations, including heatmaps, to show ART-driven decreases in mortality. The framework is also tested on external data to ensure it generalizes across various epidemiological settings. A key feature of the framework is end-to-end adjustment of ART confounders and interactive visualizations that bridge the gap between predictive models and leveraged public health information. This approach is effective in increasing the reliability of mortality estimates but also offers a scalable method for detecting disparities in access to treatment worldwide. The framework can inform policymaking by providing a quantitative measure of the estimated associations between ART and reductions in AIDS mortality, thereby supporting efforts to provide equitable healthcare services.</p> Graphical abstract <p></p>