MARATTO

article

Beyond Linear Trend Estimation: Uncertainty-weighted Generalised Additive Mixed Modelling of Global Liver Disease Burden

Abstract

Background: Non-linear models, unequal (random) estimation uncertainty, repeated observations, and temporal autocorrelation are not adequately addressed in conventional liver-burden studies, which typically use percentage changes or linear annual trends. The purpose of this study was to classify liver disease trajectories by age and assess their predictive validity from 1980 to 2023. Methods: We computed 672 observations from the Global Burden of Disease (GBD) 2023 for three measures, three age groups and two metrics. Counts and rates were fitted separately using an uncertainty-weighted generalised additive mixed model (GAMM) with a first-order autoregressive AR (1) residual correlation structure. Information criteria, explained deviance, residual diagnostics and prediction errors were used to compare the two specifications: lognormal and Gamma. Results: GAMM–AR (1) models outperformed linear models for counts (AIC: −1,798.963 vs −1,208.529) and rates (−1,881.696 vs −1,259.897), explaining 99.887% and 99.976% of deviance, respectively. The temporal effects were significant for both metrics (p < .001) and had moderate autocorrelation (ρ = 0.534–0.541). Relative to Age Code 15, Age Code 24 had a higher count (rate ratio = 2.381, 95% CI: 2.139–2.651) but a lower rate (0.171, 0.153–0.190). Predictive R2 values ranged from 0.998–0.999. Conclusions: The novel framework combines GBD uncertainty, trajectories and serial dependence. Surveillance systems should monitor counts and rates separately, focusing on recent upward trends; these should be validated using independent datasets and findings from other locations that assess similar risks.

Research topics

  • Liver Disease Diagnosis and Treatment
  • Liver Disease and Transplantation
  • Chronic Disease Management Strategies

Read the original research

This page summarises published work. The authoritative version sits with the publisher.

DOI: 10.9734/irjgh/2026/v9i1161

Is something wrong with this record? Report it or request removal.

Discussion

Discuss this research

Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.

No discussion yet. Open the first thread.