Development and validation of a long-term survival prediction model for older adults with asthma

BACKGROUND: Asthma prevalence is increasing among older adults globally, yet mortality risk factors remain poorly characterized. This study aimed to identify key mortality risk factors in older adults with asthma and develop a validated survival prediction model to guide clinical decision-making. METHODS: Using two longitudinal cohorts (SHARE and CHARLS), we included 1,584 older adults with asthma. Risk factors were identified through comprehensive analysis including permutation-based feature importance, concordance index variation curves, and survival model comparison. Five machine learning algorithms were developed and compared using concordance index (C-index). The optimal model was evaluated using integrated brier score, time-dependent area under the curve (td-AUC), calibration curves, and decision curve analysis. SHapley Additive exPlanations (SHAP) analysis quantified individual risk factor contributions, and subgroup survival analyses validated risk factor associations. A nomogram and web-based clinical tool were developed. RESULTS: During follow-up, 311 deaths occurred in SHARE (10-year mortality: 27.99%) and 183 in CHARLS (38.69%). Six key mortality risk factors were identified: advanced age, higher frailty index, male, reduced PEF%pred, lower BMI, and cardiovascular disease. The Cox proportional hazards model achieved optimal performance with C-index of 0.774 (training), 0.771 (testing), and 0.743 (external validation). SHAP analysis revealed advanced age, frailty index, and male as the strongest risk predictors. Subgroup analyses in the SHARE cohort demonstrated significant survival differences for all six risk factors (P < 0.0001), while the CHARLS cohort showed consistent trends but with some variation in statistical significance across subgroups. CONCLUSIONS: Six clinical variables effectively predict mortality risk in older adults with asthma. This survival prediction model enables risk stratification and identification of high-risk patients who may benefit from intensified monitoring and targeted interventions. The findings provide new insights into mortality risk factors and offer a practical tool for personalized management of asthma in older adults.