Graph neural network-based risk stratification of prostate cancer using gene expression and SHAP interpretability
Saeed Pirmoradi, Sevil Vaghefi Moghaddam, Mohammadreza Ardalan, Mir Mohsen Sharifi Bonab, Mohammad Teshnehlab, Sepideh Zununi Vahed · 2026
Read Graph neural network-based risk stratification of prostate cancer using gene expression and SHAP interpretability by Saeed Pirmoradi, Sevil Vaghefi Moghaddam, Mohammadreza Ardalan, Mir Mohsen Sharifi Bonab, Mohammad Teshnehlab, Sepideh Zununi Vahed (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/plos-pone-0356727.txt.
Description
Saeed Pirmoradi and coauthors published a 2026 study in PLOS ONE on grading prostate cancer risk from gene activity instead of relying only on biopsy Gleason scores. Biopsy is invasive and can cause pain, infection, or worse. Their framework feeds patients' gene expression profiles and a network of gene interactions into a dual-branch graph neural network, sorting patients into low, medium, and high risk. An explanation method then shows which genes drove each call.
Distinct gene sets marked each risk tier, underlining the disease's different molecular faces. On a held-out test set, the model separated tiers well, most strongly for high risk, with accuracy around four in five patients. Pairing network-aware learning with explanations, they argue, makes predictions sharper and auditable. The main limit is plain: everything rests on one public cancer atlas, so outside cohorts must confirm the signature.
Catalog record
| CREATOR | Saeed Pirmoradi, Sevil Vaghefi Moghaddam, Mohammadreza Ardalan, Mir Mohsen Sharifi Bonab, Mohammad Teshnehlab, Sepideh Zununi Vahed |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 58370 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |