Artificial intelligence–based prognostic models in acute myeloid leukemia: systematic review and meta-analysis

AI AML prognostic models show moderate AUC overall, but performance varies widely and external validation is often limited. Genomic-only models do not consistently outperform clinical models; better reporting and prospective validation are needed. Machine-learning (ML) and deep-learning tools have been proposed to improve survival prediction in acute myeloid leukemia (AML), but comparative benchmarks remain unclear. PRISMA 2020 searches of PubMed, Scopus, and Web of Science (January 2018–March 2025) identified studies developing or externally validating AI-based models for overall or relapse-free survival reporting area under the ROC curve (AUC). Two reviewers extracted design, population, features, algorithms, and training/validation AUCs and assessed risk of bias using PROBAST. Random-effects meta-analysis (DerSimonian–Laird) pooled validation AUCs overall and by horizon (1/2/3/5 years) and feature category (gene-centric vs non-genetic). Optimism bias was the training–validation AUC difference. We included 24 predominantly retrospective studies (137 model cohorts; ∼51,055 patients). Of 120 PROBAST domain ratings, 74% were low risk, 25% unclear, and