Dynamic platelet trajectory patterns and their nonlinear association with mortality in acute myocardial infarction

While platelet counts are known prognostic markers in acute myocardial infarction (AMI), the predictive value of dynamic platelet trajectories and their nonlinear association with nadir platelet counts remains unexplored. We aimed to identify distinct platelet trajectory patterns and evaluate their independent association with 28-day mortality in AMI patients. Using the MIMIC-IV database, we analyzed longitudinal platelet counts during the first 4 days of ICU admission in 3,657 AMI patients, Platelet trajectory patterns were identified using an unsupervised clustering approach combining factor analysis and k-means clustering based on 24 predefined summary measures describing temporal platelet dynamics. Patients were classified into ascending (AS), stable (ST), and descending (DS) trajectory groups. Associations with 28-day mortality were assessed using multivariate logistic regression, inverse probability weighting (IPW), and doubly robust estimation, with the DS group as the reference. Restricted cubic spline analysis was used to evaluate nonlinear relationships between nadir platelet counts and mortality. The cohort (median age 65 years; 58% male) revealed three distinct platelet trajectories: ascending (AS, n = 860), stable (ST, n = 2,270), and descending (DS, n = 527). Compared to the high-risk DS group, both AS (OR = 0.42, 95% CI 0.33-0.54) and ST (OR = 0.48, 95% CI 0.39-0.59) groups showed significantly lower 28-day mortality (all p < 0.001), with consistent results in IPW (AS: OR = 0.45; ST: OR = 0.51) and doubly robust models (AS: OR = 0.43; ST: OR = 0.49). Nadir platelet counts demonstrated a significant J-shaped association with mortality (P-nonlinear < 0.001), with the lowest estimated mortality risk observed at approximately 168 × 10⁹/L. Ascending/stable platelet trajectories are associated with reduced mortality risk compared to descending patterns in AMI, while extreme nadir values independently predict poor outcomes. Dynamic monitoring of platelet trajectories and nadir platelet levels may provide additional prognostic information for early risk stratification in AMI patients.