What question did this study set out to answer?

To evaluate the performance of AI-based prognostic models for survival in acute myeloid leukemia (AML) through systematic review and meta-analysis.

March 13, 2026Blood NeoplasiaOpen Access

Artificial Intelligence-Based Prognostic Models in Acute Myeloid Leukemia: Systematic Review and Meta-Analysis

View Full Paper

Ask AI

Bookmark

Authors

Systematic review finds AI models show moderate performance in predicting survival in AML, highlighting the need for better validation.

To evaluate the performance of AI-based prognostic models for survival in acute myeloid leukemia (AML) through systematic review and meta-analysis.
Conducted PRISMA 2020 searches of multiple databases (PubMed, Scopus, Web of Science) from 2018 to 2025.
Extracted data on model design, population, features, algorithms, and AUCs from studies.
Performed random-effects meta-analysis to pool validation AUCs across studies and analyze subgroups by horizon and feature category.
Assessed risk of bias using the PROBAST tool.
Pooled validation AUC across 73 cohorts was 0.769, indicating moderate prognostic discrimination.
Validation AUCs improved over time, with 5-year models achieving 0.833.
Non-genetic models had higher validation AUCs compared to genomic models (0.776 vs 0.741).
High variability was found between studies, with substantial optimism bias noted.

Zhang et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark