What question did this study set out to answer?

July 26, 2026Open Access

Global burden of traumatic brain injury from 1990 to 2021 and projections to 2050: A GBD 2021–based study using interpretable machine learning

SDShiying DongTianjin Medical University General Hospital MQMing QiuTianjin Medical University General Hospital MLMintao LiTianjin Medical University General Hospital

Key Points

This study aims to assess the global burden of traumatic brain injury from 1990 to 2021 and project future trends to 2050.
Utilized Global Burden of Disease Study 2021 estimates for TBI incidence, prevalence, and YLDs across 204 countries.
Employed age-period-cohort modeling, frontier analysis, and Bayesian models for projections and interpretation.
Applied machine learning techniques for prediction, with model performance validated using cross-validation.
Incident cases of TBI increased by 22.35%, prevalence by 53.27%, and YLDs by 52.65% from 1990 to 2021.
Age-standardized TBI rates declined by 20.16% for incidence, 16.52% for prevalence, and 16.19% for YLDs.
Projected demographic shifts indicate declining age-standardized rates through 2050, with significant disparities based on age, sex, and socioeconomic factors.

Abstract

Traumatic brain injury (TBI) is a major cause of disability and socioeconomic burden worldwide. Using updated Global Burden of Disease Study 2021 estimates, this study assessed global TBI burden from 1990 to 2021 and projected future patterns to 2050. Global Burden of Disease Study 2021 estimates were used to evaluate TBI incidence, prevalence, and years lived with disability (YLDs) in 204 countries and territories. Age-standardized rates were stratified by age, sex, region, and sociodemographic index (SDI). Age-period-cohort (APC) modeling, frontier analysis, and inequality metrics were applied to characterize temporal patterns and disparities. Bayesian APC models projected population-level burden to 2050, and eXtreme Gradient Boosting-SHapley Additive exPlanation models were used as supplementary tools for prediction and interpretation. Model performance was assessed using cross-validation. From 1990 to 2021, incident, prevalent, and YLD cases increased by 22.35%, 53.27%, and 52.65%, respectively, whereas the corresponding age-standardized rates declined by 20.16%, 16.52%, and 16.19%. Men consistently had a higher burden than women. The APC model suggested a transition around 52.5 years, indicating a shift toward older populations. In low-SDI regions, the age-standardized YLD rate changed by 1.14% (95% uncertainty interval: −2.32 to 5.03), indicating no statistically significant change. SDI showed an inverted U-shaped association with YLD rates. Age-standardized incidence rate increased in the Caribbean and Oceania. Cross-validation showed close agreement between observed and predicted values (cross-validated R 2 = 0.997–0.998; mean absolute error = 2.27–16.64). Projections suggested declining age-standardized rates through 2050. Global TBI burden increased in absolute numbers but declined in age-standardized rates. Persistent age, sex, socioeconomic, and regional disparities support targeted prevention, trauma-care strengthening, and rehabilitation planning.

AI에게 질문

Bookmark

View Full Paper