OncoCast-MPM

Overview

OncoCast-MPM is a machine-learning elastic-net penalized Cox regression risk-prediction tool for genomic risk stratification in leiomyosarcoma (soft tissue and uterine). It integrates somatic genomic alterations from targeted panel sequencing to predict disease-specific survival PMID:38488807.

Used by

• PMID:38488807 — OncoCast-MPM was developed using 195 STLMS and 238 ULMS cases from MSKCC profiled by MSK-IMPACT (341–505 genes); defined a 3-tier genomic risk stratification for disease-specific survival; validated using 317 STLMS cases from AACR GENIE; high-risk patients (co-occurrence of RB1 mutation + chr12q deletion or ATRX mutation) had 5-year DSS of 43% vs. 79% for low-risk (no alterations) PMID:38488807.

Notes

• The elastic-net Cox model was preferred over more complex ML approaches to maintain interpretability for clinical application PMID:38488807.
• Corpus-grown slug; not present in canonical ontology.

Sources

• PMID:38488807

This page was processed by crosslinker on 2026-04-11.