Validating
Unsupervised genomic classification of acute myeloid leukemia
Leukemia classification and genomics
Within the HARMONY Alliance, the group is developing and internationally validating an unsupervised genomic classification of AML. This work has been presented in the EHA 2025 plenary session and as an ASH 2025 oral abstract. It is not yet represented here as a peer-reviewed article.
What problem are we trying to solve?
Can a fully unsupervised model of mutations and cytogenetics recover reproducible biological classes of AML that refine existing risk groupings?
Why it matters
AML is biologically heterogeneous. Current risk scores remain incomplete. A generalizable genomic classification could support more precise prognosis and trial design if it holds outside the derivation cohort.
Approach
Hierarchical Dirichlet mixture modeling on large HARMONY genomic cohorts, with independent validation in additional international datasets. Public description is limited to what has been presented at EHA 2025 (plenary) and ASH 2025 (oral abstract).
Who is working on it
Papers and abstracts
- Development and Independent Validation of Unsupervised Genomic Classification of Patients With Acute Myeloid Leukemia Using Hierarchical Dirichlet Mixture Models: A HARMONY Alliance Study EHA 2025 Congress, Plenary Abstract Session 2025 · Conference abstract
- Independent, international validation of the unsupervised harmony classification of Acute Myeloid Leukemia Blood (ASH Annual Meeting Abstracts) 2025 · Conference abstract
- Covering Hierarchical Dirichlet Mixture Models on binary data to enhance genomic stratifications in onco-hematology PLoS Comput Biol 2024
- Unravelling co-mutational patterns with prognostic implications in NPM1 mutated adult acute myeloid leukemia - a HARMONY study Leukemia 2026
Talks
- HARMONY results presented at EHA Congress 2025